diff --git a/README.md b/README.md index e1a7b92..dded186 100644 --- a/README.md +++ b/README.md @@ -24,7 +24,7 @@ Both purposes share one schema, one prompt set, one methodology. ## October 2026: image, video and music generation -The [Predator media campaign](https://benchmarks.weeyuga.com/benchmarks/predator-media-20261021.html) adds 123 retained generated files and 130 attempt records, including poor outputs and no-output failures. Read the [findings article](https://benchmarks.weeyuga.com/findings/images-video-music-on-6gb.html). The full media evidence lives in `runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/`; its per-group records also follow the existing `runs//` layout. These records use a media-generation schema rather than invented token-throughput metrics. Measurements and prose follow the archive data license; generated media and upstream model material are excluded from a blanket license grant and retain the applicable terms linked in the campaign README and dataset. +The [Predator media campaign](https://benchmarks.weeyuga.com/benchmarks/predator-media-20261021.html) now retains 133 generated files and 140 attempt records, including poor outputs and no-output failures. Read the [findings article](https://benchmarks.weeyuga.com/findings/images-video-music-on-6gb.html). The full media evidence lives in `runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/`; its per-group records also follow the existing `runs//` layout. These records use a media-generation schema rather than invented token-throughput metrics. Measurements and prose follow the archive data license; generated media and upstream model material are excluded from a blanket license grant and retain the applicable terms linked in the campaign README and dataset. ## Quick start — for friends contributing a benchmark diff --git a/catalogue.json b/catalogue.json index 4b650f0..397a7a1 100644 --- a/catalogue.json +++ b/catalogue.json @@ -1,7 +1,7 @@ { "schema_version": "1.0", - "generated_at": "2026-10-21T05:35:13Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "generated_at": "2026-10-21T09:28:54Z", + "git_sha": "d304145ed0046408a75e3e3c957e93bcb6c0737f", "site_base_url": "https://benchmarks.weeyuga.com", "data_base_url": "https://benchmarks.weeyuga.com/data", "gitea_archive_url": "https://github.com/slobodanmargetic988/weeyuga-benchmarks-public", @@ -56,15 +56,18 @@ "scatter" ] }, - "site_grade_count": 23, + "site_grade_count": 24, "site_grade_breakdown": { "flagship": 4, - "standard": 19, + "standard": 20, "archive-only": 16, "legacy": 103 }, - "total_runs_in_archive": 142, + "total_runs_in_archive": 143, "by_model": { + "unsloth-qwen-image-2-1-turbo-gguf-q4-k-m": [ + "bd801c85-11e3-5033-b076-acd84846b820" + ], "zai-org-cogvideox-2b-q8-0": [ "a657b671-dbc0-5434-bce1-fa2aa99bac1b", "69e67f23-518d-5b32-a277-0720a153274c", @@ -370,6 +373,7 @@ }, "by_hardware": { "predator": [ + "bd801c85-11e3-5033-b076-acd84846b820", "a657b671-dbc0-5434-bce1-fa2aa99bac1b", "69e67f23-518d-5b32-a277-0720a153274c", "bfcf95b8-1682-5963-b5dc-a9b7ea7e535c", @@ -524,6 +528,7 @@ }, "by_family": { "generation": [ + "bd801c85-11e3-5033-b076-acd84846b820", "a657b671-dbc0-5434-bce1-fa2aa99bac1b", "69e67f23-518d-5b32-a277-0720a153274c", "bfcf95b8-1682-5963-b5dc-a9b7ea7e535c", @@ -702,6 +707,7 @@ }, "by_task": { "media-generation": [ + "bd801c85-11e3-5033-b076-acd84846b820", "a657b671-dbc0-5434-bce1-fa2aa99bac1b", "69e67f23-518d-5b32-a277-0720a153274c", "bfcf95b8-1682-5963-b5dc-a9b7ea7e535c", @@ -854,6 +860,7 @@ }, "by_engine": { "comfyui": [ + "bd801c85-11e3-5033-b076-acd84846b820", "a657b671-dbc0-5434-bce1-fa2aa99bac1b", "69e67f23-518d-5b32-a277-0720a153274c", "bfcf95b8-1682-5963-b5dc-a9b7ea7e535c", @@ -1112,6 +1119,7 @@ }, "by_grade": { "standard": [ + "bd801c85-11e3-5033-b076-acd84846b820", "a657b671-dbc0-5434-bce1-fa2aa99bac1b", "69e67f23-518d-5b32-a277-0720a153274c", "2ef120f1-a624-57c8-85fe-ed58e29b74b3", @@ -1374,6 +1382,62 @@ ] }, "benchmarks": [ + { + "id": "bd801c85-11e3-5033-b076-acd84846b820", + "kind": "media-generation", + "title": "Predator media: Qwen Image 2.1 Turbo Q4 \u00b7 ten-image seed/cache trial", + "date": "2026-10-21", + "started_at": "2026-10-21T08:41:18.067000+00:00", + "git_sha": "63be2f9b82ad0dac88a7ceaa28a9cf3154125a18", + "hardware": "predator", + "hardware_label": "Predator \u00b7 GTX 1060 6 GB \u00b7 28 GB RAM \u00b7 Windows", + "engine": "comfyui", + "harness": "packet-foundry-media-v1", + "model_family": "generation", + "model_sizes": [ + "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M" + ], + "task_kind": "media-generation", + "tags": [ + "image", + "video", + "music", + "retained-failures" + ], + "headline": "10 retained outputs \u00b7 10 accepted results \u00b7 10 recorded attempts \u00b7 median Factory trace 202.0s", + "janie_blurb_md": "[Browse the complete campaign and media](/benchmarks/predator-media-20261021.html). [Read the findings](/findings/images-video-music-on-6gb.html).", + "janie_blurb_status": "published", + "caveat": "Technical delivery and media decoding do not establish visual or musical quality. Timings include adapter overhead; cache behavior differs between runs. Failed and cancelled attempts remain in the evidence.", + "caveat_severity": "warning", + "methodology_ref": "packet-foundry-media-v1", + "methodology_deviations_md": "Source commit identifies ComfyUI 0.42.1. Private bridge revisions and workflow hashes, where captured, are recorded separately in the campaign evidence. No clean matched-quality comparison or isolated quantization speedup is claimed.", + "results_table": [], + "cells": [], + "n_calls_total": 10, + "n_errors_total": 0, + "chart_spec": null, + "site_grade": "standard", + "site_grade_reason": "Retained media with explicit verification and quality limitations", + "raw_data_urls": { + "jsonl": "/data/predator-media-20261021/runs/bd801c85-11e3-5033-b076-acd84846b820/run.jsonl", + "md": "/data/predator-media-20261021/runs/bd801c85-11e3-5033-b076-acd84846b820/run.md", + "log": "/data/predator-media-20261021/failures.json", + "gitea_dir": "/data/predator-media-20261021/runs/bd801c85-11e3-5033-b076-acd84846b820/metadata.json" + }, + "synthesis_doc": "/findings/images-video-music-on-6gb.html", + "synthesis_docs_all": [ + "/findings/images-video-music-on-6gb.html" + ], + "related_ids": [ + "051eb12c-1d98-59a1-9225-4a1addb622de", + "071f90f8-7fb6-5652-9dcd-8999e6f2360b", + "2ef120f1-a624-57c8-85fe-ed58e29b74b3", + "3e0cbed8-136a-5de3-9323-b0c087d342a2" + ], + "status": "complete", + "visibility": "public", + "media_group": "qwen21-turbo-q4km-ten" + }, { "id": "a657b671-dbc0-5434-bce1-fa2aa99bac1b", "kind": "media-generation", @@ -2331,7 +2395,7 @@ "title": "Pascal Vulkan-vs-CUDA Cross-Machine Benchmark \u2014 2026-05-06", "date": "2026-05-06", "started_at": "2026-05-06T08:30:00Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "pavilion+predator (cross-machine)", "hardware_label": "pavilion+predator (cross-machine)", "engine": "llamacpp", @@ -3730,7 +3794,7 @@ "title": "vps50-cpu-matrix-1 \u2014 gemma/phi/qwen2.5/qwen3 on vps50", "date": "2026-05-05", "started_at": "2026-05-05T10:23:13Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "vps50", "hardware_label": "VPS50 \u00b7 cloud \u00b7 16 vCPU AMD EPYC \u00b7 62 GB RAM \u00b7 no GPU", "engine": "llamacpp", @@ -3993,7 +4057,7 @@ "title": "predator-a3b-ngl6-ctx-1 \u2014 qwen3 on predator", "date": "2026-05-05", "started_at": "2026-05-05T09:44:34Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "predator", "hardware_label": "Predator \u00b7 gaming laptop \u00b7 GTX 1060 6 GB \u00b7 28 GB RAM", "engine": "llamacpp", @@ -4672,7 +4736,7 @@ "title": "predator-a3b-ngl-ctx-2d-1 \u2014 qwen3 on predator", "date": "2026-05-05", "started_at": "2026-05-05T07:43:00Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "predator", "hardware_label": "Predator \u00b7 gaming laptop \u00b7 GTX 1060 6 GB \u00b7 28 GB RAM", "engine": "llamacpp", @@ -5963,7 +6027,7 @@ "title": "vps50-cpu-matrix-1 \u2014 gemma/phi/qwen2.5/qwen3 on vps50", "date": "2026-05-05", "started_at": "2026-05-05T07:00:10Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "vps50", "hardware_label": "VPS50 \u00b7 cloud \u00b7 16 vCPU AMD EPYC \u00b7 62 GB RAM \u00b7 no GPU", "engine": "llamacpp", @@ -6282,7 +6346,7 @@ "title": "vps50-cpu-matrix-1 \u2014 gemma/phi/qwen2.5/qwen3 on vps50", "date": "2026-05-05", "started_at": "2026-05-05T03:07:15Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "vps50", "hardware_label": "VPS50 \u00b7 cloud \u00b7 16 vCPU AMD EPYC \u00b7 62 GB RAM \u00b7 no GPU", "engine": "llamacpp", @@ -6484,7 +6548,7 @@ "title": "vps50-gemma-e4b-1 \u2014 gemma on vps50", "date": "2026-05-05", "started_at": "2026-05-05T03:07:13Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "vps50", "hardware_label": "VPS50 \u00b7 cloud \u00b7 16 vCPU AMD EPYC \u00b7 62 GB RAM \u00b7 no GPU", "engine": "llamacpp", @@ -6579,7 +6643,7 @@ "title": "vps50-gemma-e4b-1 \u2014 gemma on vps50", "date": "2026-05-04", "started_at": "2026-05-04T23:53:09Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "vps50", "hardware_label": "VPS50 \u00b7 cloud \u00b7 16 vCPU AMD EPYC \u00b7 62 GB RAM \u00b7 no GPU", "engine": "llamacpp", @@ -6674,7 +6738,7 @@ "title": "predator-a3b-ctx-sweep-1 \u2014 qwen3 on predator", "date": "2026-05-04", "started_at": "2026-05-04T23:48:01Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "predator", "hardware_label": "Predator \u00b7 gaming laptop \u00b7 GTX 1060 6 GB \u00b7 28 GB RAM", "engine": "llamacpp", @@ -7353,7 +7417,7 @@ "title": "vps50-cpu-matrix-1 \u2014 gemma/phi/qwen2.5/qwen3 on vps50", "date": "2026-05-04", "started_at": "2026-05-04T23:39:53Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "vps50", "hardware_label": "VPS50 \u00b7 cloud \u00b7 16 vCPU AMD EPYC \u00b7 62 GB RAM \u00b7 no GPU", "engine": "llamacpp", @@ -7556,7 +7620,7 @@ "title": "predator-a3b-ngl-matrix-1 \u2014 qwen3 on predator", "date": "2026-05-04", "started_at": "2026-05-04T22:50:47Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "predator", "hardware_label": "Predator \u00b7 gaming laptop \u00b7 GTX 1060 6 GB \u00b7 28 GB RAM", "engine": "llamacpp", @@ -8212,7 +8276,7 @@ "title": "A3B cross-machine \u2014 Pavilion side", "date": "2026-05-04", "started_at": "2026-05-04T22:11:43Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "pavilion", "hardware_label": "Pavilion \u00b7 HP laptop \u00b7 GTX 1050 4 GB \u00b7 16 GB RAM \u00b7 i7-9750H", "engine": "llamacpp", @@ -8468,15 +8532,15 @@ ], "status": "complete", "visibility": "draft", - "title_override": "A3B cross-machine \u2014 Pavilion side", - "headline_override": "Pavilion (GTX 1050) running Qwen3-30B-A3B IQ2-XXS at ~6 tok/s mean, p50 58s. The smaller half of the cross-machine pair." + "headline_override": "Pavilion (GTX 1050) running Qwen3-30B-A3B IQ2-XXS at ~6 tok/s mean, p50 58s. The smaller half of the cross-machine pair.", + "title_override": "A3B cross-machine \u2014 Pavilion side" }, { "id": "d1cff064-9141-48e0-b1d2-729b134e0543", "title": "Predator Qwen rerun", "date": "2026-05-04", "started_at": "2026-05-04T22:02:28Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "pavilion", "hardware_label": "Pavilion \u00b7 HP laptop \u00b7 GTX 1050 4 GB \u00b7 16 GB RAM \u00b7 i7-9750H", "engine": "llamacpp", @@ -8573,7 +8637,7 @@ "title": "A3B cross-machine \u2014 Predator side", "date": "2026-05-04", "started_at": "2026-05-04T22:02:17Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "predator", "hardware_label": "Predator \u00b7 gaming laptop \u00b7 GTX 1060 6 GB \u00b7 28 GB RAM", "engine": "llamacpp", @@ -8830,15 +8894,15 @@ ], "status": "complete", "visibility": "draft", - "title_override": "A3B cross-machine \u2014 Predator side", - "headline_override": "Predator (GTX 1070) running the same A3B IQ2-XXS at ~3.9 tok/s mean, p50 102s \u2014 the larger half of the pair." + "headline_override": "Predator (GTX 1070) running the same A3B IQ2-XXS at ~3.9 tok/s mean, p50 102s \u2014 the larger half of the pair.", + "title_override": "A3B cross-machine \u2014 Predator side" }, { "id": "8ec3c8cb-58a7-4f33-974c-9c1bfb7824ef", "title": "predator-a3b-1 \u2014 qwen3 on predator", "date": "2026-05-04", "started_at": "2026-05-04T21:24:11Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "predator", "hardware_label": "Predator \u00b7 gaming laptop \u00b7 GTX 1060 6 GB \u00b7 28 GB RAM", "engine": "llamacpp", @@ -8935,7 +8999,7 @@ "title": "Predator Qwen rerun", "date": "2026-05-04", "started_at": "2026-05-04T21:24:10Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "pavilion", "hardware_label": "Pavilion \u00b7 HP laptop \u00b7 GTX 1050 4 GB \u00b7 16 GB RAM \u00b7 i7-9750H", "engine": "llamacpp", @@ -9032,7 +9096,7 @@ "title": "pavilion-a3b-1 \u2014 qwen3 on pavilion", "date": "2026-05-04", "started_at": "2026-05-04T20:38:52Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "pavilion", "hardware_label": "Pavilion \u00b7 HP laptop \u00b7 GTX 1050 4 GB \u00b7 16 GB RAM \u00b7 i7-9750H", "engine": "llamacpp", @@ -9129,7 +9193,7 @@ "title": "predator-a3b-1 \u2014 qwen3 on predator", "date": "2026-05-04", "started_at": "2026-05-04T18:50:19Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "predator", "hardware_label": "Predator \u00b7 gaming laptop \u00b7 GTX 1060 6 GB \u00b7 28 GB RAM", "engine": "llamacpp", @@ -9251,7 +9315,7 @@ "title": "Predator Qwen rerun matrix", "date": "2026-05-04", "started_at": "2026-05-04T17:49:07Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "predator", "hardware_label": "Predator \u00b7 gaming laptop \u00b7 GTX 1060 6 GB \u00b7 28 GB RAM", "engine": "llamacpp", @@ -9844,15 +9908,15 @@ ], "status": "complete", "visibility": "draft", - "title_override": "Predator Qwen rerun matrix", - "headline_override": "Three Qwen variants on Predator: 36 calls, ~10.9 tok/s mean, p50 24s. The single-machine flagship for the rerun campaign." + "headline_override": "Three Qwen variants on Predator: 36 calls, ~10.9 tok/s mean, p50 24s. The single-machine flagship for the rerun campaign.", + "title_override": "Predator Qwen rerun matrix" }, { "id": "09d8fbde-0008-49bb-99da-03eeaca72be1", "title": "Predator trio bench", "date": "2026-05-04", "started_at": "2026-05-04T16:01:52Z", - "git_sha": "570121a4fb9e4b1a7b4058fae5dd9e027a7e5618", + "git_sha": null, "hardware": "predator", "hardware_label": "Predator \u00b7 gaming laptop \u00b7 GTX 1060 6 GB \u00b7 28 GB RAM", "engine": "llamacpp", diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/README.md b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/README.md index 178c8a5..b270ed4 100644 --- a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/README.md +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/README.md @@ -4,7 +4,7 @@ Real local-hardware observations from October 20–21, 2026, using a Windows com ## Coverage -The archive retains **123 output files**: **110 images, 11 videos and two music samples**. Of those, 122 were accepted and acknowledged by the Warehouse; one additional Qwen image finished locally but delivery was cancelled after network and lease failures. The image total includes the exact 100-image prompt-complexity suite and ten earlier image outputs. Two pairs of early accepted images have identical original file hashes, so the archive contains **121 unique original media byte sequences**, while preserving every delivery individually. +The archive retains **133 output files**: **120 images, 11 videos and two music samples**. Of those, 132 were accepted and acknowledged by the Warehouse; one additional Qwen image finished locally but delivery was cancelled after network and lease failures. The image total includes the exact 100-image prompt-complexity suite, ten earlier image outputs and a ten-image Qwen Image 2.1 Turbo follow-up. Two pairs of early accepted images have identical original file hashes, so the archive contains **131 unique original media byte sequences**, while preserving every delivery individually. Eleven video contact sheets and the image used as LTX video input are derived reference material, not additional model outputs. Two synthetic codec-preflight MP4 fixtures are excluded from the model gallery and generation counts. @@ -36,6 +36,15 @@ Delivery, checksums, full decoding and actual LangSmith media attachment retriev ## Public-copy verification and attribution -Original media remain retained privately. PNG copies were losslessly re-encoded with ancillary metadata removed, with decoded pixel bytes compared against each source. Video and FLAC containers were remuxed without metadata and without re-encoding; encoded stream hashes were compared against each original. Source and public-copy SHA-256 hashes are both recorded. Every public output was fully decoded; `publication-audit.json` records the 123-file checks. Video contact sheets are derived JPEGs, not replacement original clips. +Original media remain retained privately. PNG copies were losslessly re-encoded with ancillary metadata removed, with decoded pixel bytes compared against each source. Video and FLAC containers were remuxed without metadata and without re-encoding; encoded stream hashes were compared against each original. Source and public-copy SHA-256 hashes are both recorded. Every public output was fully decoded; `publication-audit.json` retains the original 123-file checks. `append-audit-qwen-turbo-ten.json` records the ten additional PNG checks and proves original media/records were preserved. Video contact sheets are derived JPEGs, not replacement original clips. Prompts, measurements and observations are supplied as benchmark evidence. Model and companion terms remain separate from Packet Foundry's Apache-2.0 code license. In particular, FLUX.1 dev uses its non-commercial model license; YuE2 has CC-BY-NC-4.0 model terms; Qwen Image 2.1 uses its research terms; FLUX.3 components link their upstream FLUX Kommunity and companion terms. The model-source and license links in the dataset identify the relevant upstream material. This archive does not grant a new blanket license to model weights or override upstream conditions on their outputs. + + +## Qwen Turbo follow-up and choosing seed variants + +Ten further local images use the distinct Unsloth Qwen Image 2.1 Turbo Q4_K_M checkpoint, the official eight-step ManualSigmas schedule, CFG1 and Euler. The first two share a brief with seeds 42/43; eight subsequent requests use new briefs. All ten outputs are retained, including visible material, geometry or constraint errors. Independent original-image visual review is separate from successful relay, decoding and actual LangSmith attachment/experiment checks. + +The production cache finding is supported by all 50 prior FLUX pairs: every second request reused both conditioning outputs while sampling again. Excluding initial loading, native medians were 70.05 s for the new brief and 38.49 s for its next seed. The Qwen Turbo pair took 320.246 s then 95.234 s; its first included model loading, so that delta combines startup/model and encoding reuse. `seed-variant-cache-findings.json` separates these scopes, includes the actual measured gaps and maps all 50 FLUX pairs to unchanged public records. No five-variant batch was tested or extrapolated as measured. + +Turbo model terms: Qwen research license; not Apache-2.0. https://huggingface.co/Qwen/Qwen-Image-2.1-Turbo/blob/fc13ba1986fcb905854dc8de2f52eee395f49b5b/LICENSE Workflow attribution and MIT notice: `qwen-turbo-workflow-NOTICE.md`. Existing model/media license limitations remain in force; no weights are redistributed. diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/append-audit-qwen-turbo-ten.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/append-audit-qwen-turbo-ten.json new file mode 100644 index 0000000..505f262 --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/append-audit-qwen-turbo-ten.json @@ -0,0 +1,638 @@ +{ + "completed_at": "2026-10-21T09:23:32.364175+00:00", + "scope": "Append ten Qwen Turbo originals; preserve the existing 123 generated files and 130 attempt records.", + "new_original_media": 10, + "new_original_media_full_decode_checks": 10, + "new_public_copy_checksums_verified": 10, + "new_original_content_preserved_checks": 10, + "expected_campaign_retained_outputs": 133, + "expected_campaign_recorded_attempts": 140, + "expected_no_output_failure_rows": 7, + "old_baseline_dataset_sha256": "e67fd94071a63873c156e97e4acd3230b38e4928dfc7c01f01bf79272988c451", + "old_record_objects_preserved": 130, + "old_media_sha256_preserved": { + "media/qwen-01.png": "f3a0f6a8fc2db57a92bb4828202de902fe2b32eb4657c11d80592fb6ef4269e9", + "media/flux3-01.png": "8a6f6acba48baa9a8932a5d480dcf8c7cdbd99550d61c8e2d94adaebae86a29d", + "media/flux3-02.png": "386ee3050f72fcd8c8d6cac4b04925c76eb6ed1af89d8f291bcd20aab84814d0", + "media/qwen-02.png": "e6e63fd8fcdcff56e913be15b74ee17c648ee8f1bfca91cc93d2768625fd49eb", + "media/qwen-03.png": "e6e63fd8fcdcff56e913be15b74ee17c648ee8f1bfca91cc93d2768625fd49eb", + "media/qwen-04.png": "88922cfd817857f78ac9c8188781bf26fe3186cefa27229693fee0f4ff427d11", + "media/flux3-03.png": "8a6f6acba48baa9a8932a5d480dcf8c7cdbd99550d61c8e2d94adaebae86a29d", + "media/flux3-04.png": "33a814e3934c8ad934cd5dfecfa687761f9639590f59ed753364c8653db4a750", + "media/flux1-01.png": "2ef4e49593fd964cd76091599c833e2bc12673cfa049c0550f0971a2c6173650", + "media/flux1-02.png": "922a91c7b98d9a7195a76bba70980994366ca2eb8a15500ff8be3d4fc234fd84", + "media/yue2-01.flac": "305088e03699e617f3402bbf110bc009c414c09e89ded9b93c17ff05ac6cd7c4", + "media/yue2-02.flac": "2749f7d5dd4d0d22efd4fc53c375b7494feb2e978986fdbef3f4d95129d7a430", + "media/wan-01.mp4": "adcc3baae187708e268c74781ee8cf7f491934ededf3b15979979bb915bbb363", + "media/wan-02.mp4": "eb76558eda9d4f34b79be8d3885b5fe35cff8ede12cb33766f4c7aecf302571e", + "media/wan-03.mp4": "7735cf2a321e43ef8248a36eb2f363c1d37e938ce9a2f9b140efeca506b4334c", + "media/ltx-01.mp4": "56b30d098647ecfa4cd2d5d6d9788bb09d1348c734fba053fecf86f17387c4ce", + "media/ltx-02.mp4": "b20b579efa2d6c5dada6f7da27af0ec75246a28cb90d2b8fac65084ad86e4fe1", + "media/ltx-03.mp4": "aef9f338efcd5b731d3b7568d8cabc9fb3085622e5b0e758aac9462420ac39e1", + "media/ltx-04.mp4": "7903f25d4ce4a1996d75bf0569599f0b8bc42d9bda94b925eeb71a4c41261e53", + "media/cog-01.mp4": "4dd3b927cac4950eccd9fb7fbe1df583f7ec34d099c812da84add0cb6bc2012c", + "media/cog-02.mp4": "32b9da79aadb9a01b0fa08091ec65fd2d369a5b488524780ab7829c0bc50e7df", + "media/pf-img100-01-t1-s42.png": "4d011862443f564d1ce85ff0b28114c308eec185babae44a0b2228ad675ef68c", + "media/pf-img100-01-t1-s43.png": "89472b519b9b60a6a7d257d0c1dd7bffb3c3185fb9ddc5f7f64aa1a4c15bebc7", + "media/pf-img100-02-t1-s42.png": "455450e669d4eff2d8ffae9765d6a4e7ed2d680bbc7c9a0644245909b53d5aa4", + "media/pf-img100-02-t1-s43.png": "2db96f2e0ecf12c3e20349db4331512179c2db6ae8345da5f4be808c7d71a843", + "media/pf-img100-03-t1-s42.png": "f81df06cabf652c8d82a2df2777f05e57240c80b8db7bcc30d60ab507beaf820", + "media/pf-img100-03-t1-s43.png": "7ead5afbd3c3676bca51ba7c961acbba62c0c359cafd181df0bfd2014b7f34ad", + "media/pf-img100-04-t1-s42.png": "d7c0e80f1a9cc3870c7b2e36cbdba7efc9eb59dd40a275bc36798d52be472829", + "media/pf-img100-04-t1-s43.png": "55374949261af7e4219c8dc8fef459e824a5d37760e9aac6445f2c4a02281ea4", + "media/pf-img100-05-t1-s42.png": "c0a35138f89e00e10717d035f3583b4b3b5a673877abecf72d3fd9ff8a76a47e", + "media/pf-img100-05-t1-s43.png": "227ea3b005c226ed05ea506c5b834fb6493c012ed4dcc8b4e889aec15a62e4cd", + "media/pf-img100-06-t1-s42.png": "67391e3223ae7961380c81c8b10bcf257282f7d0f83b6d710d4a893fc5dccd4d", + "media/pf-img100-06-t1-s43.png": "2539e8cb8d8ee547611c749759dc13bd444f864c460ae7d9b9743c1b43d8a044", + "media/pf-img100-08-t1-s42.png": "399d70f4e4b769414c1a1051fd604d0072e04e08ce72daa441c6ca0bb7d0a495", + "media/pf-img100-08-t1-s43.png": "8fbabd3ab0261cdae18930060a40482b32e717fa169e505f62207eac61986c01", + "media/pf-img100-09-t1-s42.png": "3913267fa1e0aa7b1bb5497b2d4a8e09da1da284fc150d269334d47c875191a1", + "media/pf-img100-09-t1-s43.png": "b6d5c6c9396bd7f411756e000764f7f7a6ac46ff0213003fcfca13b4fd4bb835", + "media/pf-img100-16-t1-s42.png": "5c8e09d40f555f99120ea41ad28e966b312d4f540628f091bfdce8dbfb514f28", + "media/pf-img100-16-t1-s43.png": "2bb44ea2b05ec722baf3ca67e9fb56bc1507a15399671fcb9d22745a4636227c", + "media/pf-img100-20-t1-s42.png": "542eb45fec9df79ae9339a1dda2f3cc0211b75492ce02cf68b0d4e6a0a60f345", + "media/pf-img100-20-t1-s43.png": "9e36188d971e6084f329718f68d4f0fa37c6179ff9a6ba55ac9f880a44ac8c21", + "media/pf-img100-01-t2-s42.png": "d47a03923b13f722ac90c4a36983ce0feb9924759a5e26f384a6a5933cbf27e1", + "media/pf-img100-01-t2-s43.png": "67d4129bd8b19f8abdecae021eec2d0a6c10e54da6457959037bb3e66a64f012", + "media/pf-img100-02-t2-s42.png": "96506641ec6e742a0973ee932337212a7c93132eb4093477f1897e29c2807283", + "media/pf-img100-02-t2-s43.png": "4d384901e455073e60ed5d023b8a3f58c6fd3b3ac7e0190c2a27e4591348ec9b", + "media/pf-img100-03-t2-s42.png": "f0691b4c987040cda7b2af982c20183f40dfce29bacc2896f8f6da7fea3706fa", + "media/pf-img100-03-t2-s43.png": "88b8b21bac327a1ad1e88d98a9f98426b14ac1d088f867022c122be84d589435", + "media/pf-img100-04-t2-s42.png": "5a744f7d40382e341d51c76abee39272eb64d353d247160e7db3add08a0cee82", + "media/pf-img100-04-t2-s43.png": "03ff21d1588a073bdb6f5311ba276d477eefc2138283c116ab21234890996a30", + "media/pf-img100-05-t2-s42.png": "75581bb9146a02807e7e7894f88aad9c7ccec320300c20397bffe754a9cfa5a4", + "media/pf-img100-05-t2-s43.png": "b88067882a5645f862ccf7f0c9f61fe03d967faab0b9b20d9a236ee7951b30dd", + "media/pf-img100-06-t2-s42.png": "3a33f8768d0d5bf54cd9e92a97c7461fcef83f46da17abcbd01af55392355ea9", + "media/pf-img100-06-t2-s43.png": "20d3003590d0b8e3bec8bce99a444f27ba50372bb60411af8ed600dfd01a27ae", + "media/pf-img100-08-t2-s42.png": "6aa65295f92b91492eaaf1415f6cb09b3338003df3502c86f47ae71bf44d70bf", + "media/pf-img100-08-t2-s43.png": "6c8f7d5ea5e413ed3895de953d792119669fc8f4aee4f1b506d396ef335c8aa6", + "media/pf-img100-09-t2-s42.png": "1811f70d4d86df600b2d174ad58e2529a80e0bdeae177c41a9c8308a8dea3f51", + "media/pf-img100-09-t2-s43.png": "f313a8b6da015d12eb32923265aef93a4695a1e7e8c0e330a405b13a73325dae", + "media/pf-img100-16-t2-s42.png": "149548593b20ba8324a37b9dabd9f2726e3a7877f56c57522ced9c6b8feae2f5", + "media/pf-img100-16-t2-s43.png": "d632e079fc728c6da8fd6f1f060579651c60c7fe2cab7bf5d013266eb3b6cb8d", + "media/pf-img100-20-t2-s42.png": "55ae2c4edeff03d2c1f34100e8b86eedb563813ee12262e590893f0d3c283f01", + "media/pf-img100-20-t2-s43.png": "98f6dc93e64a129ffb8f32a94c04f09cb16621f1dfd595e87025693c41d00c0d", + "media/pf-img100-01-t3-s42.png": "e3de1fe5e729106e7a00349a2f6e55c46f9f1f8c4f8e781b829854844d537cc1", + "media/pf-img100-01-t3-s43.png": "e8f52bcf88af1957b44aa998ba3114720f768516fa040633cf3c65eee3130ef2", + "media/pf-img100-02-t3-s42.png": "5a74641bb956f93f2bddcb4f7cd32456d7638aa209ef4d38e6d46c9ad06649eb", + "media/pf-img100-02-t3-s43.png": "0916ff1c51f837b29ccaa20fa4fd93d98d05cb263cf446c0127ee394ab2eb3e1", + "media/pf-img100-03-t3-s42.png": "60ca4597f393d6cb2f31086fa631a1f377772c7effcc049d6ac8707b668a1b12", + "media/pf-img100-03-t3-s43.png": "9e9c2105cc8301d50aaecde59d13e3cb57e7e99dcc404dc9b904244f8e9b6f6a", + "media/pf-img100-04-t3-s42.png": "934d7c3d4b03113f263b3ba66404c3d7c8d60011e41254090a92de19d8c95596", + "media/pf-img100-04-t3-s43.png": "43b2bc08989f06e8fa32ba1171d8920013372a0cf1bf870f3aff90a07bcbae08", + "media/pf-img100-05-t3-s42.png": "0c09d09036675811d612bbf1358dfe84e9379d06a77d0e40d7e7c7c364e619ce", + "media/pf-img100-05-t3-s43.png": "c2a68b852c2c76ce6697d8435381189906291ba22bbb0ec034bca21b075cef43", + "media/pf-img100-06-t3-s42.png": "2b1d624dd277033b32daea289006227c1286ccef6471d11fbcf72eb52546894a", + "media/pf-img100-06-t3-s43.png": "d5752edb0af59d76724c850392a0f92c7fe3930340ba216feb1ea91edbbffc59", + "media/pf-img100-08-t3-s42.png": "414d1114494372a676e0630cabe00216594afe347b268c5d91aa7a56c66b5ca6", + "media/pf-img100-08-t3-s43.png": "67c1bd9b83181f4e6122f60d603b07728b3eb6878f9490b936fe9f149a88abba", + "media/pf-img100-09-t3-s42.png": "79f783753dcbb750f80854888b6fcac21b08708565ba4c497b6db063d29b8af5", + "media/pf-img100-09-t3-s43.png": "76aa5c31fe67d4d656b02691bc868533736c1cb7a54eae293187c6027f429e48", + "media/pf-img100-16-t3-s42.png": "6e4635cac388b40ad5f8d5aae53eb4727ed12587063facc3100f4bdbe3ba179c", + "media/pf-img100-16-t3-s43.png": "796998cdd7b86fbcb05cf516315efd87cc9eff9635ca89a4ec22b9ea07296239", + "media/pf-img100-20-t3-s42.png": "428876e01d8b449b03f6e2dd5549a35a7cce383bd928c890f0454ee7c4071ea1", + "media/pf-img100-20-t3-s43.png": "3fbbd890f7cbafc2729d98412133fbad3a47f7475596874c92598ca34a82dc09", + "media/pf-img100-01-t4-s42.png": "f78b82656ab97b41e423321512a53fc7b666c7bc14344a8ee7edcd0f2557a087", + "media/pf-img100-01-t4-s43.png": "2fb4ab887090b381ac0dbe8c2d7b53fcfb1f3efe702ecc7074459a4e6ea844e0", + "media/pf-img100-02-t4-s42.png": "b8e28c5f3128ff1cab97600a781817ef9a8f2fa368ac6c3f61cbf18f810ea26b", + "media/pf-img100-02-t4-s43.png": "d4b819508f19e8732a05f726a6cbfa69abc43a4449776b5ac04fb64d5df24ae8", + "media/pf-img100-03-t4-s42.png": "a3adb9f82f156192b2fa89c29feb4e6519380041a6482df5046b7f042819e6a9", + "media/pf-img100-03-t4-s43.png": "9063a3f0b63e83fc846cef57efc5dbd9649f0963ec57f672e9f5cb2b8b91cad9", + "media/pf-img100-04-t4-s42.png": "5fca74f8e128d5e4e16154e14ef583d609b38c8415a693e73758c9d5e7b66920", + "media/pf-img100-04-t4-s43.png": "0fa71a1950c82dc3020307f087d270f21caaa137ee281e6ed7c15f77ff1b0479", + "media/pf-img100-05-t4-s42.png": "53ab61359e2c27e1ac2d0b74d72bb8eab474b7c8c4617a0d03294488a9f6ca06", + "media/pf-img100-05-t4-s43.png": "4748cbf7e6b1f1c3e7efa9c312ce0104b5fb818e2c62a5ccb4dd1c46ee47bce1", + "media/pf-img100-06-t4-s42.png": "78909142e418be12dc5792e2e8fd1ee40564f46830c95f5b5247daaf919eb71a", + "media/pf-img100-06-t4-s43.png": "a564761843a0f3f0631c5cd59e816662323a9dc415e12171aeded43c9572f62b", + "media/pf-img100-08-t4-s42.png": "65eb7c0a22907028e602f7236994b0fe38758c89f47f560138366017d23a7a59", + "media/pf-img100-08-t4-s43.png": "89472c8fa56c9059cccd9b014ba95a983c8481d7f0a0c6ed059510a1efd8cd69", + "media/pf-img100-09-t4-s42.png": "46027e9f44014747c45ececf58e270747e4d2d41639e0396f3682e9df0d76a4f", + "media/pf-img100-09-t4-s43.png": "04a3e58482bcc56912f9e0291484b18c10286e970e23e34a661f3da0d2d34ecc", + "media/pf-img100-16-t4-s42.png": "0331f4135cc5b0b1308818885ef1d4654b652d28f188e9fee3e694e73315ff00", + "media/pf-img100-16-t4-s43.png": "b2d37e14d218cdff81fb08f631c20f60bc706b40da4534b2605742bb6ef0c373", + "media/pf-img100-20-t4-s42.png": "9cedc1bf978bf81d9b904f71735f865ae21847f556245ce745edaeea5e6ae0a1", + "media/pf-img100-20-t4-s43.png": "ecb08b25069c6b9d4422321da866391132e5af79d4556ec95bf03d488f93f2f5", + "media/pf-img100-01-t5-s42.png": "a391cabbff6837a961a6c903a3d3fb3551525e0e7d27c513cc038a61323022ee", + "media/pf-img100-01-t5-s43.png": "f97225ce6251bab1704e03666e324f5054161eecbba39dd82485fc1adf35648d", + "media/pf-img100-02-t5-s42.png": "2612918a0c200599e36f03a6e68673f84866d16008b12ab1a4584395cef037db", + "media/pf-img100-02-t5-s43.png": "071d4e45bfdeb6d582c9ad9e7069bab68623c7df6f0467659838b3720552cb57", + "media/pf-img100-03-t5-s42.png": "4b77aad9976034ae24657004520358a7baa84fd84b009cee26499a106d6966b3", + "media/pf-img100-03-t5-s43.png": "328116ac07109fad53e34bd0e2c72f9375955217da84c10d7b2b47c041120bfa", + "media/pf-img100-04-t5-s42.png": "7041179c88974444906d56f8762363d2468901acc1cb7a81bcba68921175035b", + "media/pf-img100-04-t5-s43.png": "2ea4b06e4bb8127a61d49b3401f0824ab45162dad7681755aa59bee62c05b7df", + "media/pf-img100-05-t5-s42.png": "f1c1aa1fc11d2f94930d9515074f5e556c6db9e7801eb8b5d02000706337410e", + "media/pf-img100-05-t5-s43.png": "eced8b3843f88a1d9d02b5402625a31d65b840f3562807fefda9b708abb78abe", + "media/pf-img100-06-t5-s42.png": "08865bf86f1856bfec5b77bf43702c1e3fe255254a8af2f005054e426065871b", + "media/pf-img100-06-t5-s43.png": "28a10a6d476ef7e2ba3edbe0394d5372c6f2a5592f97378f5fd32cf74f8e8bc0", + "media/pf-img100-08-t5-s42.png": "58daf884844773d6907505cafa2302394ec587c766d2ed89fe6b1f0c0893efb6", + "media/pf-img100-08-t5-s43.png": "be26c9bdf28b88322a03a97646c46744a9cd018c13af73ed947e7925e86fa24f", + "media/pf-img100-09-t5-s42.png": "0db31157b184e937321c08c63a5a3a0dccf1e2fb964b7b516a0253c10898caaa", + "media/pf-img100-09-t5-s43.png": "ac473dd27fd0752c2383399fdf1a642035463dd4b95fc38fe12fefc8b24abfd4", + "media/pf-img100-16-t5-s42.png": "83932709fc4ff79ca74152153c596ae15b833c04b9ddaa21e4ea9ab1217b1a1c", + "media/pf-img100-16-t5-s43.png": "7cd179dbce208d84d0f81bf0b4cc84626aa8a84c208a2900a5b4cf2e5614ad91", + "media/pf-img100-20-t5-s42.png": "44332c7ea8a35e709d14b55b28384167f740449911589da2bfc81af8fe76100e", + "media/pf-img100-20-t5-s43.png": "c44512027c26fe213a4cfb979380c25ea045f7872233f29575e19a2af76dd7af", + "media/cog-03.mp4": "4ab00c3b745f3893bee84552cd7dd261ec113747e2544ecac917be75bd920a4c", + "media/cog-04.mp4": "190985d4238c7e0e5c51490fe700b18c67e7cfca80bce4cbd7e50ddac7dd7648" + }, + "all_existing_media_files_preserved": { + "media/pf-img100-16-t2-s43.png": "d632e079fc728c6da8fd6f1f060579651c60c7fe2cab7bf5d013266eb3b6cb8d", + "media/pf-img100-05-t1-s43.png": "227ea3b005c226ed05ea506c5b834fb6493c012ed4dcc8b4e889aec15a62e4cd", + "media/pf-img100-20-t4-s42.png": "9cedc1bf978bf81d9b904f71735f865ae21847f556245ce745edaeea5e6ae0a1", + "media/pf-img100-20-t4-s43.png": "ecb08b25069c6b9d4422321da866391132e5af79d4556ec95bf03d488f93f2f5", + "media/pf-img100-05-t1-s42.png": "c0a35138f89e00e10717d035f3583b4b3b5a673877abecf72d3fd9ff8a76a47e", + "media/pf-img100-16-t2-s42.png": "149548593b20ba8324a37b9dabd9f2726e3a7877f56c57522ced9c6b8feae2f5", + "media/pf-img100-03-t5-s43.png": "328116ac07109fad53e34bd0e2c72f9375955217da84c10d7b2b47c041120bfa", + "media/pf-img100-08-t1-s43.png": "8fbabd3ab0261cdae18930060a40482b32e717fa169e505f62207eac61986c01", + "media/cog-04.mp4": "190985d4238c7e0e5c51490fe700b18c67e7cfca80bce4cbd7e50ddac7dd7648", + "media/pf-img100-02-t3-s42.png": "5a74641bb956f93f2bddcb4f7cd32456d7638aa209ef4d38e6d46c9ad06649eb", + "media/pf-img100-06-t4-s43.png": "a564761843a0f3f0631c5cd59e816662323a9dc415e12171aeded43c9572f62b", + "media/pf-img100-06-t4-s42.png": "78909142e418be12dc5792e2e8fd1ee40564f46830c95f5b5247daaf919eb71a", + "media/pf-img100-02-t3-s43.png": "0916ff1c51f837b29ccaa20fa4fd93d98d05cb263cf446c0127ee394ab2eb3e1", + "media/pf-img100-08-t1-s42.png": "399d70f4e4b769414c1a1051fd604d0072e04e08ce72daa441c6ca0bb7d0a495", + "media/pf-img100-03-t5-s42.png": "4b77aad9976034ae24657004520358a7baa84fd84b009cee26499a106d6966b3", + "media/cog-02-contact.jpg": "15806f437b3d98f08d06f1e97553c47b6879edb71b236fa9299a2d27124238f8", + "media/pf-img100-03-t4-s43.png": "9063a3f0b63e83fc846cef57efc5dbd9649f0963ec57f672e9f5cb2b8b91cad9", + "media/pf-img100-06-t5-s43.png": "28a10a6d476ef7e2ba3edbe0394d5372c6f2a5592f97378f5fd32cf74f8e8bc0", + "media/cog-01.mp4": "4dd3b927cac4950eccd9fb7fbe1df583f7ec34d099c812da84add0cb6bc2012c", + "media/pf-img100-02-t2-s42.png": "96506641ec6e742a0973ee932337212a7c93132eb4093477f1897e29c2807283", + "media/pf-img100-02-t2-s43.png": "4d384901e455073e60ed5d023b8a3f58c6fd3b3ac7e0190c2a27e4591348ec9b", + "media/pf-img100-06-t5-s42.png": "08865bf86f1856bfec5b77bf43702c1e3fe255254a8af2f005054e426065871b", + "media/wan-01-contact.jpg": "531f01cb909a32e67e7e3e5050bdbc673146aed25b0a93ac0e3f4d75c9704439", + "media/pf-img100-03-t4-s42.png": "a3adb9f82f156192b2fa89c29feb4e6519380041a6482df5046b7f042819e6a9", + "media/pf-img100-16-t3-s43.png": "796998cdd7b86fbcb05cf516315efd87cc9eff9635ca89a4ec22b9ea07296239", + "media/pf-img100-20-t5-s42.png": "44332c7ea8a35e709d14b55b28384167f740449911589da2bfc81af8fe76100e", + "media/cog-02.mp4": "32b9da79aadb9a01b0fa08091ec65fd2d369a5b488524780ab7829c0bc50e7df", + "media/cog-03.mp4": "4ab00c3b745f3893bee84552cd7dd261ec113747e2544ecac917be75bd920a4c", + "media/pf-img100-20-t5-s43.png": "c44512027c26fe213a4cfb979380c25ea045f7872233f29575e19a2af76dd7af", + "media/pf-img100-16-t3-s42.png": "6e4635cac388b40ad5f8d5aae53eb4727ed12587063facc3100f4bdbe3ba179c", + "media/pf-img100-08-t2-s43.png": "6c8f7d5ea5e413ed3895de953d792119669fc8f4aee4f1b506d396ef335c8aa6", + "media/pf-img100-09-t4-s42.png": "46027e9f44014747c45ececf58e270747e4d2d41639e0396f3682e9df0d76a4f", + "media/ltx-03-contact.jpg": "9af621b0aa91cc3e494df53f60cdcec5416ef68472e062dfecda76f3f7793ca5", + "media/yue2-01.flac": "305088e03699e617f3402bbf110bc009c414c09e89ded9b93c17ff05ac6cd7c4", + "media/qwen-04.png": "88922cfd817857f78ac9c8188781bf26fe3186cefa27229693fee0f4ff427d11", + "media/pf-img100-09-t4-s43.png": "04a3e58482bcc56912f9e0291484b18c10286e970e23e34a661f3da0d2d34ecc", + "media/pf-img100-08-t2-s42.png": "6aa65295f92b91492eaaf1415f6cb09b3338003df3502c86f47ae71bf44d70bf", + "media/pf-img100-04-t4-s42.png": "5fca74f8e128d5e4e16154e14ef583d609b38c8415a693e73758c9d5e7b66920", + "media/wan-02-contact.jpg": "2391cbc5c9b6bedaec4c192572e96592d4aa6b3d0c55ca592e2f45e0ba1ae718", + "media/cog-04-contact.jpg": "b05b605af378c04efb82f0744b05939135f940e1a450f46c6462fef7e7f87636", + "media/pf-img100-16-t1-s43.png": "2bb44ea2b05ec722baf3ca67e9fb56bc1507a15399671fcb9d22745a4636227c", + "media/pf-img100-05-t2-s43.png": "b88067882a5645f862ccf7f0c9f61fe03d967faab0b9b20d9a236ee7951b30dd", + "media/pf-img100-01-t5-s42.png": "a391cabbff6837a961a6c903a3d3fb3551525e0e7d27c513cc038a61323022ee", + "media/pf-img100-01-t5-s43.png": "f97225ce6251bab1704e03666e324f5054161eecbba39dd82485fc1adf35648d", + "media/pf-img100-05-t2-s42.png": "75581bb9146a02807e7e7894f88aad9c7ccec320300c20397bffe754a9cfa5a4", + "media/pf-img100-16-t1-s42.png": "5c8e09d40f555f99120ea41ad28e966b312d4f540628f091bfdce8dbfb514f28", + "media/pf-img100-04-t4-s43.png": "0fa71a1950c82dc3020307f087d270f21caaa137ee281e6ed7c15f77ff1b0479", + "media/cog-01-contact.jpg": "0ed1a9f64cca26220bd7df16dfae30f7ea01d34b27e34283c3cdf5b9b15bb23f", + "media/pf-img100-04-t5-s42.png": "7041179c88974444906d56f8762363d2468901acc1cb7a81bcba68921175035b", + "media/wan-02.mp4": "eb76558eda9d4f34b79be8d3885b5fe35cff8ede12cb33766f4c7aecf302571e", + "media/pf-img100-01-t4-s42.png": "f78b82656ab97b41e423321512a53fc7b666c7bc14344a8ee7edcd0f2557a087", + "media/qwen-03.png": "e6e63fd8fcdcff56e913be15b74ee17c648ee8f1bfca91cc93d2768625fd49eb", + "media/pf-img100-05-t3-s43.png": "c2a68b852c2c76ce6697d8435381189906291ba22bbb0ec034bca21b075cef43", + "media/pf-img100-05-t3-s42.png": "0c09d09036675811d612bbf1358dfe84e9379d06a77d0e40d7e7c7c364e619ce", + "media/qwen-02.png": "e6e63fd8fcdcff56e913be15b74ee17c648ee8f1bfca91cc93d2768625fd49eb", + "media/pf-img100-01-t4-s43.png": "2fb4ab887090b381ac0dbe8c2d7b53fcfb1f3efe702ecc7074459a4e6ea844e0", + "media/wan-03.mp4": "7735cf2a321e43ef8248a36eb2f363c1d37e938ce9a2f9b140efeca506b4334c", + "media/pf-img100-04-t5-s43.png": "2ea4b06e4bb8127a61d49b3401f0824ab45162dad7681755aa59bee62c05b7df", + "media/pf-img100-02-t1-s42.png": "455450e669d4eff2d8ffae9765d6a4e7ed2d680bbc7c9a0644245909b53d5aa4", + "media/pf-img100-08-t3-s43.png": "67c1bd9b83181f4e6122f60d603b07728b3eb6878f9490b936fe9f149a88abba", + "media/wan-01.mp4": "adcc3baae187708e268c74781ee8cf7f491934ededf3b15979979bb915bbb363", + "media/pf-img100-09-t5-s42.png": "0db31157b184e937321c08c63a5a3a0dccf1e2fb964b7b516a0253c10898caaa", + "media/qwen-01.png": "f3a0f6a8fc2db57a92bb4828202de902fe2b32eb4657c11d80592fb6ef4269e9", + "media/pf-img100-09-t5-s43.png": "ac473dd27fd0752c2383399fdf1a642035463dd4b95fc38fe12fefc8b24abfd4", + "media/pf-img100-08-t3-s42.png": "414d1114494372a676e0630cabe00216594afe347b268c5d91aa7a56c66b5ca6", + "media/pf-img100-02-t1-s43.png": "2db96f2e0ecf12c3e20349db4331512179c2db6ae8345da5f4be808c7d71a843", + "media/flux1-01.png": "2ef4e49593fd964cd76091599c833e2bc12673cfa049c0550f0971a2c6173650", + "media/pf-img100-03-t2-s42.png": "f0691b4c987040cda7b2af982c20183f40dfce29bacc2896f8f6da7fea3706fa", + "media/pf-img100-06-t3-s42.png": "2b1d624dd277033b32daea289006227c1286ccef6471d11fbcf72eb52546894a", + "media/flux3-04.png": "33a814e3934c8ad934cd5dfecfa687761f9639590f59ed753364c8653db4a750", + "media/pf-img100-02-t4-s43.png": "d4b819508f19e8732a05f726a6cbfa69abc43a4449776b5ac04fb64d5df24ae8", + "media/pf-img100-02-t4-s42.png": "b8e28c5f3128ff1cab97600a781817ef9a8f2fa368ac6c3f61cbf18f810ea26b", + "media/pf-img100-06-t3-s43.png": "d5752edb0af59d76724c850392a0f92c7fe3930340ba216feb1ea91edbbffc59", + "media/pf-img100-03-t2-s43.png": "88b8b21bac327a1ad1e88d98a9f98426b14ac1d088f867022c122be84d589435", + "media/ltx-01-contact.jpg": "2d0c8d1d11498438d7e30c0fa836c19a352063978f57bb6ac20f26af88486eac", + "media/flux1-02.png": "922a91c7b98d9a7195a76bba70980994366ca2eb8a15500ff8be3d4fc234fd84", + "media/ltx-04.mp4": "7903f25d4ce4a1996d75bf0569599f0b8bc42d9bda94b925eeb71a4c41261e53", + "media/pf-img100-16-t5-s42.png": "83932709fc4ff79ca74152153c596ae15b833c04b9ddaa21e4ea9ab1217b1a1c", + "media/pf-img100-01-t1-s43.png": "89472b519b9b60a6a7d257d0c1dd7bffb3c3185fb9ddc5f7f64aa1a4c15bebc7", + "media/pf-img100-20-t3-s43.png": "3fbbd890f7cbafc2729d98412133fbad3a47f7475596874c92598ca34a82dc09", + "media/pf-img100-20-t3-s42.png": "428876e01d8b449b03f6e2dd5549a35a7cce383bd928c890f0454ee7c4071ea1", + "media/pf-img100-01-t1-s42.png": "4d011862443f564d1ce85ff0b28114c308eec185babae44a0b2228ad675ef68c", + "media/pf-img100-16-t5-s43.png": "7cd179dbce208d84d0f81bf0b4cc84626aa8a84c208a2900a5b4cf2e5614ad91", + "media/ltx-01.mp4": "56b30d098647ecfa4cd2d5d6d9788bb09d1348c734fba053fecf86f17387c4ce", + "media/pf-img100-16-t4-s42.png": "0331f4135cc5b0b1308818885ef1d4654b652d28f188e9fee3e694e73315ff00", + "media/cog-03-contact.jpg": "e91ba5d3ae684f78c23e2a61ffd3a4d0c4a9556d4390b65406585c24de04b47c", + "media/pf-img100-20-t2-s43.png": "98f6dc93e64a129ffb8f32a94c04f09cb16621f1dfd595e87025693c41d00c0d", + "media/pf-img100-04-t1-s43.png": "55374949261af7e4219c8dc8fef459e824a5d37760e9aac6445f2c4a02281ea4", + "media/flux3-02.png": "386ee3050f72fcd8c8d6cac4b04925c76eb6ed1af89d8f291bcd20aab84814d0", + "media/flux3-03.png": "8a6f6acba48baa9a8932a5d480dcf8c7cdbd99550d61c8e2d94adaebae86a29d", + "media/pf-img100-04-t1-s42.png": "d7c0e80f1a9cc3870c7b2e36cbdba7efc9eb59dd40a275bc36798d52be472829", + "media/pf-img100-20-t2-s42.png": "55ae2c4edeff03d2c1f34100e8b86eedb563813ee12262e590893f0d3c283f01", + "media/pf-img100-16-t4-s43.png": "b2d37e14d218cdff81fb08f631c20f60bc706b40da4534b2605742bb6ef0c373", + "media/pf-img100-09-t1-s43.png": "b6d5c6c9396bd7f411756e000764f7f7a6ac46ff0213003fcfca13b4fd4bb835", + "media/ltx-02.mp4": "b20b579efa2d6c5dada6f7da27af0ec75246a28cb90d2b8fac65084ad86e4fe1", + "media/pf-img100-03-t3-s42.png": "60ca4597f393d6cb2f31086fa631a1f377772c7effcc049d6ac8707b668a1b12", + "media/pf-img100-02-t5-s43.png": "071d4e45bfdeb6d582c9ad9e7069bab68623c7df6f0467659838b3720552cb57", + "media/pf-img100-06-t2-s42.png": "3a33f8768d0d5bf54cd9e92a97c7461fcef83f46da17abcbd01af55392355ea9", + "media/flux3-01.png": "8a6f6acba48baa9a8932a5d480dcf8c7cdbd99550d61c8e2d94adaebae86a29d", + "media/pf-img100-06-t2-s43.png": "20d3003590d0b8e3bec8bce99a444f27ba50372bb60411af8ed600dfd01a27ae", + "media/yue2-02.flac": "2749f7d5dd4d0d22efd4fc53c375b7494feb2e978986fdbef3f4d95129d7a430", + "media/ltx-04-contact.jpg": "c7f5d5c989cf6f43cd4d673bbb86fcd2a54d0a29ca8c85ef196121f24d3fb2a0", + "media/pf-img100-02-t5-s42.png": "2612918a0c200599e36f03a6e68673f84866d16008b12ab1a4584395cef037db", + "media/pf-img100-03-t3-s43.png": "9e9c2105cc8301d50aaecde59d13e3cb57e7e99dcc404dc9b904244f8e9b6f6a", + "media/ltx-03.mp4": "aef9f338efcd5b731d3b7568d8cabc9fb3085622e5b0e758aac9462420ac39e1", + "media/pf-img100-09-t1-s42.png": "3913267fa1e0aa7b1bb5497b2d4a8e09da1da284fc150d269334d47c875191a1", + "media/pf-img100-04-t3-s43.png": "43b2bc08989f06e8fa32ba1171d8920013372a0cf1bf870f3aff90a07bcbae08", + "media/pf-img100-01-t2-s43.png": "67d4129bd8b19f8abdecae021eec2d0a6c10e54da6457959037bb3e66a64f012", + "media/pf-img100-05-t5-s42.png": "f1c1aa1fc11d2f94930d9515074f5e556c6db9e7801eb8b5d02000706337410e", + "media/pf-img100-05-t5-s43.png": "eced8b3843f88a1d9d02b5402625a31d65b840f3562807fefda9b708abb78abe", + "media/pf-img100-01-t2-s42.png": "d47a03923b13f722ac90c4a36983ce0feb9924759a5e26f384a6a5933cbf27e1", + "media/pf-img100-04-t3-s42.png": "934d7c3d4b03113f263b3ba66404c3d7c8d60011e41254090a92de19d8c95596", + "media/pf-img100-08-t5-s42.png": "58daf884844773d6907505cafa2302394ec587c766d2ed89fe6b1f0c0893efb6", + "media/pf-img100-03-t1-s42.png": "f81df06cabf652c8d82a2df2777f05e57240c80b8db7bcc30d60ab507beaf820", + "media/pf-img100-09-t3-s43.png": "76aa5c31fe67d4d656b02691bc868533736c1cb7a54eae293187c6027f429e48", + "media/pf-img100-09-t3-s42.png": "79f783753dcbb750f80854888b6fcac21b08708565ba4c497b6db063d29b8af5", + "media/pf-img100-03-t1-s43.png": "7ead5afbd3c3676bca51ba7c961acbba62c0c359cafd181df0bfd2014b7f34ad", + "media/pf-img100-08-t5-s43.png": "be26c9bdf28b88322a03a97646c46744a9cd018c13af73ed947e7925e86fa24f", + "media/pf-img100-06-t1-s42.png": "67391e3223ae7961380c81c8b10bcf257282f7d0f83b6d710d4a893fc5dccd4d", + "media/pf-img100-08-t4-s42.png": "65eb7c0a22907028e602f7236994b0fe38758c89f47f560138366017d23a7a59", + "media/pf-img100-09-t2-s43.png": "f313a8b6da015d12eb32923265aef93a4695a1e7e8c0e330a405b13a73325dae", + "media/pf-img100-09-t2-s42.png": "1811f70d4d86df600b2d174ad58e2529a80e0bdeae177c41a9c8308a8dea3f51", + "media/wan-03-contact.jpg": "60b9741370ec284cbfb778c09ec3b64ba8e21219c5732e7a139e940cd370d6b5", + "media/pf-img100-08-t4-s43.png": "89472c8fa56c9059cccd9b014ba95a983c8481d7f0a0c6ed059510a1efd8cd69", + "media/pf-img100-06-t1-s43.png": "2539e8cb8d8ee547611c749759dc13bd444f864c460ae7d9b9743c1b43d8a044", + "media/ltx-reference-fox.png": "4d50bd2de7bd42a4deca86ddb8cd35dd2b47ef1f240146f8a501630e5b97ac87", + "media/pf-img100-20-t1-s43.png": "9e36188d971e6084f329718f68d4f0fa37c6179ff9a6ba55ac9f880a44ac8c21", + "media/pf-img100-04-t2-s43.png": "03ff21d1588a073bdb6f5311ba276d477eefc2138283c116ab21234890996a30", + "media/ltx-02-contact.jpg": "e85eeb20f92d06c59c8ae61e92a64dc1ef703b9d3fffa8c2e4045e84653c3300", + "media/pf-img100-05-t4-s42.png": "53ab61359e2c27e1ac2d0b74d72bb8eab474b7c8c4617a0d03294488a9f6ca06", + "media/pf-img100-01-t3-s43.png": "e8f52bcf88af1957b44aa998ba3114720f768516fa040633cf3c65eee3130ef2", + "media/pf-img100-01-t3-s42.png": "e3de1fe5e729106e7a00349a2f6e55c46f9f1f8c4f8e781b829854844d537cc1", + "media/pf-img100-05-t4-s43.png": "4748cbf7e6b1f1c3e7efa9c312ce0104b5fb818e2c62a5ccb4dd1c46ee47bce1", + "media/pf-img100-04-t2-s42.png": "5a744f7d40382e341d51c76abee39272eb64d353d247160e7db3add08a0cee82", + "media/pf-img100-20-t1-s42.png": "542eb45fec9df79ae9339a1dda2f3cc0211b75492ce02cf68b0d4e6a0a60f345" + }, + "all_existing_workflow_files_preserved": { + "workflows/pf-img100-20-t1-s42.json": "326c3236f63462fca8794b3e389d30959afc94b82f980d5192260fcc44576b96", + "workflows/pf-img100-05-t4-s42.json": "f1fc51a0d327e47675f1b07cc7fb56ac9dc8f044aeae849861b6154793a89a02", + "workflows/pf-img100-05-t5-s42.json": "3d189c20e031c238e215d91539cfbb22dadfd1772c096690e106305a663591a8", + "workflows/h3-dynamic-failure.json": "5c7fd74009dedf7cb5dc21d0a58b9f7ccf81c8bb08d27134601299659c3034a3", + "workflows/pf-img100-16-t2-s43.json": "ba885ee5be5af2afd9c06bff1a9ccfce937cc88480cd1485b3a9b717c1471385", + "workflows/pf-img100-16-t3-s43.json": "4b74efa1c1b1a44a14ef4f51268851dfee841a5b325a3e5af47ae9e458c75d1f", + "workflows/qwen-01.json": "db6d6b7acf7b9476202a1e844dc0a88576ff5581b0614eeb08bfaeb971806719", + "workflows/cog-03.json": "fcb4bc417fd364d877497a3677db5ac0dc2cdadfa687465cad91cd5552af07c2", + "workflows/pf-img100-03-t3-s43.json": "bbe37203767caf3f0b2cd2c470253f9c288b48912c26ec0604b24237be85f2c2", + "workflows/pf-img100-03-t2-s43.json": "ab380d8ff081007929da88adfc210fcfb7756d4c075a0521c290a37b46dfc733", + "workflows/pf-img100-09-t5-s43.json": "e8436d7ff85453036e97b7cf659428ca522a891cb75ace4826cf67c79dfc0158", + "workflows/pf-img100-09-t4-s43.json": "c9f8428593f26af6f2b8ae21d3834278da45a40020e36fde8557eb257fe62ccd", + "workflows/ltx-02.json": "b7f0a15f413e553e2d7f40d959eb93eed21efbb6d18b6f7969ee8fb99d0b62ff", + "workflows/pf-img100-06-t3-s42.json": "d42869d820ea818fba44010d67648b014fb6ca4bb8c8e53c15935e1cb5128454", + "workflows/pf-img100-06-t2-s42.json": "ce4d2060691b78b43cbf8165dd0f1fcf796e79bea00cebb62866c1035921b95d", + "workflows/flux1-02.json": "34bef4e31688f6abea0ec4b2ffff41af1c3dd8442e773e8b13d006eae78d2b80", + "workflows/pf-img100-02-t1-s43.json": "89b907bf0b2b7eb309bde3ca2a70fe1fd9708c9c55b2241abb6661ade27cb402", + "workflows/pf-img100-02-t1-s42.json": "05e3a0f415497356e28bc3efa49f19a49177857746f75db456878342cfdf5142", + "workflows/pf-img100-06-t3-s43.json": "c509aa3f276b3dc85acd68e7203675e22b809fe528465e6f3362b016ef4a535d", + "workflows/pf-img100-06-t2-s43.json": "186b743a0d970cf37a5c9f67f10feb2a87e67281f2c88bdd74fd60f86bc74e39", + "workflows/ltx-03.json": "ec933114dfabd1fd07c72a96a7a95623bab5e61d86b41548e093978a855f766a", + "workflows/pf-img100-09-t5-s42.json": "13f6b7e98f9188c2c3c3f8d2585286fc42797fba84abd130be64e5b18692898c", + "workflows/pf-img100-09-t4-s42.json": "78c3d0c8e77579e81299bea04288f734be685165903106be9ede77243b62346e", + "workflows/flux3-04.json": "2506ab9f2aeb0642165d35391978c1f48681fe63cfcaa68afd8204b98d6e0e75", + "workflows/pf-img100-03-t3-s42.json": "cd6581098bab9947643f97dad349079ed95336ad846d6e236bdbebb909f774c6", + "workflows/pf-img100-03-t2-s42.json": "deb5c23f0b1358bc714e3a68e6d1dea4ca9882db9bf2fbef0960b898e1818d5e", + "workflows/cog-02.json": "a297179138978dd412d9e251729ebe3607ea9357aab40827e73df0fb0b8ddab0", + "workflows/pf-img100-16-t2-s42.json": "4957c0ed0f00a087ed02b6bf725dc03096ffebe37f60d221b1c454efda005da4", + "workflows/pf-img100-16-t3-s42.json": "b3f07a4a42d8ff964b7092f1df1f654c1b434d7d545248354e5d7c4746bce2fb", + "workflows/pf-img100-05-t4-s43.json": "068b40a65074c864cb27a2cd14e2c37a835b63f6ba6142f28f09bf4797f30eaa", + "workflows/pf-img100-05-t5-s43.json": "8cb34f8ce1add49089da2c9ce4e5498f97f729e6de336deb559e08ce497c8b00", + "workflows/pf-img100-20-t1-s43.json": "f78796fa5fb1430672a22a34e5590daa25dee69ebcf38731fee0f30d9163ef4f", + "workflows/pf-img100-06-t4-s43.json": "155543012cd686147d0ee34e98daa9bc928bb2a90a757ba620c64f66d5aec6c7", + "workflows/pf-img100-06-t5-s43.json": "3d63c510aa36e5367222a0ba9b94e3c16b3739db81f0b260234fa4673c0b54a2", + "workflows/yue2-01.json": "62d224bd6fe710dff747657dfed600a61eb2e0abd77bc7eca1f1f0f0dc92c611", + "workflows/wan-01.json": "e4b6b473ba78c4ce93407ebfc26d2c0f2bb565d846ddf2d26419716f5a0a5849", + "workflows/pf-img100-09-t2-s42.json": "6e555f23f42c87af772573401a361ea6246c64b15f872a73b670b67cff81c4f9", + "workflows/pf-img100-09-t3-s42.json": "2951e486a53b310fb61467ba1479dccdc727c1231d2e56395df20e9bfc247b55", + "workflows/flux3-03.json": "03192b7693b36b64d331ee5e541d89af116b3564102dd714d64a9a11458cb4b4", + "workflows/pf-img100-03-t4-s42.json": "8365bb5f4116c64274843ed67a160677595395e0c95836f6fc66469226e2962a", + "workflows/pf-img100-03-t5-s42.json": "a4361af7011ecc70c1ecdaafa01f368a623042c4b8a791b9d509c6e426c03a18", + "workflows/ltx-04.json": "c586a4532e154ac333b8fa2c1475434c361930fd24e5ae0989da06bb4d63a479", + "workflows/pf-img100-08-t1-s42.json": "27b013d340702706b88ea5697bfd9eafe8e38c1e48e96f11a6b8f71e519a13ba", + "workflows/pf-img100-04-t1-s43.json": "e7d064d4418bc17fcb7f4bf0867563bde1d978d2760428dbac38f61485ac0946", + "workflows/pf-img100-16-t5-s42.json": "c40482cb2155a412f0b7886f6342b82e488cd5aee23d8560bbeb952b3659a052", + "workflows/pf-img100-16-t4-s42.json": "cef17d3605831e83faf8e7ca7c20f0056357e03e4296200c13c9a098aa653eac", + "workflows/pf-img100-01-t1-s42.json": "69192e8663227d662cccd273fa854b5f250b1fcf9b11d641c64c6e6a22d8f109", + "workflows/pf-img100-05-t3-s43.json": "24992f7e8510ad5fb15ecc7b047e273599cb417badc455312541ab7ef15a065f", + "workflows/pf-img100-05-t2-s43.json": "5a887229d45496dd6c18c08c21f57b03b93a4186098c932f2ffabc8eec3baa3d", + "workflows/pf-img100-05-t3-s42.json": "e7f2cd4fbfa237c2ea6261a3ef5b1a1fb35444693bf68290d497ad9525b3b878", + "workflows/pf-img100-05-t2-s42.json": "6e0dfd9b337b0a25b294581a982355e2f4c70bc85d3452153b3b40f343aa8130", + "workflows/pf-img100-01-t1-s43.json": "92bc29db6f3eb1d7dc37a27aed9f9e98028e03dadb9998eed47f3e0832f58a1c", + "workflows/pf-img100-16-t5-s43.json": "1bf19470e992ef39e0ff1f0859d94c5429ee14d047d73ac81f85e11bec9fa6ee", + "workflows/pf-img100-16-t4-s43.json": "a02ae35031f7471e2e27c754543bfad8adc14fe2e8428af3695eb5d2676dc83b", + "workflows/pf-img100-04-t1-s42.json": "7ba965be7c794c9049d14880aba8c541866d390b5f616fa9c9c88316047d5462", + "workflows/pf-img100-08-t1-s43.json": "b14b1b3dcbd7e4d53368c9925630d10decee5d759edcec6a4e9b89db0719042f", + "workflows/pf-img100-03-t4-s43.json": "861c8303d7b5ac952c036fa9b70997b366f25044dbb74ef6e24d5178c97529a9", + "workflows/pf-img100-03-t5-s43.json": "0fc13e34379d4a8c61268c4e38512c801eda926b740c695ed4a69703a8543f48", + "workflows/flux3-02.json": "91a012e46fb9cc14667024409a8c9309b1678c23d085cc7db3c286c04c8df831", + "workflows/pf-img100-09-t2-s43.json": "f3fb20060f47b094249e50c4edbf6fc8f9972f85f3446947fc9a11bf3b2315c1", + "workflows/pf-img100-09-t3-s43.json": "04155a0d60354766c99a64602a3cf9e3a16f3ecfbf7bc517963f4d2bd76d72a0", + "workflows/pf-img100-06-t4-s42.json": "ba99d1322d3ec1a42a3d0faa7a273f7f256af1815cbf6e53a4e1f931f897291d", + "workflows/pf-img100-06-t5-s42.json": "f605d02cabb52b01adfce3b16e5d1a1a67be429037ed263fb52e5a6f3bf57c7e", + "workflows/cog-04.json": "4dc8d577d94aafb5f5ea7f0aff9d9f256281509f31dec8bc8c3167742bd34042", + "workflows/pf-img100-16-t1-s42.json": "c290d994a04bfbc8b7847e48aa6fa3193d1b0205d611026ef6f0371a0e0486dd", + "workflows/pf-img100-04-t5-s43.json": "703b3002b92bd728c2ecc1d64609d76408fbe6f8ea726bc8caa7e25c7e2ed716", + "workflows/pf-img100-04-t4-s43.json": "ffd5498e567d50e056f18c4632cee52c766bac33d64e458adebde9b1a9f22dbf", + "workflows/flux3-01.json": "03192b7693b36b64d331ee5e541d89af116b3564102dd714d64a9a11458cb4b4", + "workflows/yue2-cpu-cancelled.json": "d55b45fe9f441376981c34ff1890e3956cb945cd72c19a6a430df379cb4235fc", + "workflows/wan-03.json": "fe66a989e0b9bf1f1a0911db2da109b76ff81b62e9952d1541bee0fd69cb34db", + "workflows/pf-img100-20-t3-s43.json": "2ed6bfde21ddfab9e5b41590d7c96c90c8311999619e24b3b429156ae31f6a0b", + "workflows/pf-img100-20-t2-s43.json": "741e19041bf8af0c6fd618c8e8d7b25834ac256cb7e9b3d97ec5eb256254ef33", + "workflows/pf-img100-01-t5-s42.json": "1df2ea3eff67a6aa66efb15db8f2e2355765c5841643eeb368456153772da23d", + "workflows/pf-img100-01-t4-s42.json": "98e5b57af07357d32eac81f011056ab09c760ee54e1ca7c8b731df9dbb42d8a3", + "workflows/pf-img100-02-t2-s42.json": "9b7a3a418ccb7117027be4ab81c2c95f3ac8f6a3d35fd1521343532ed3719428", + "workflows/pf-img100-02-t3-s42.json": "dce5f39712fd09783d22b12643ed0847d99edc91dbebe72e3086fc9967e7f02a", + "workflows/pf-img100-06-t1-s43.json": "ed0eb2c0c565ad3429ace95fcbd96d5cb638a4c961ad5a99de81e6d680b2a962", + "workflows/pf-img100-03-t1-s42.json": "df842b7f64af75ed37c864c6789b63cc2f7f2cf5cccfc6f7380f6fa311bf5933", + "workflows/pf-img100-08-t4-s42.json": "6002b00485ad3eec2ab0d1ef13e12be8dd18eb832c566931fdd0ebbd2d4875d9", + "workflows/pf-img100-08-t5-s42.json": "03a0e960f4bef600295ea2f593b5eb2385ab6d5199a89665f65ce6ffd818a027", + "workflows/pf-img100-08-t4-s43.json": "bfffa67908a5343013b930518bd37898c424d1d50ee8133135b19e4543540107", + "workflows/pf-img100-08-t5-s43.json": "9f3e1743ca3c3a9887b4c996a96354b6628e28bacb5a7349fcafc552a800c392", + "workflows/pf-img100-03-t1-s43.json": "6b777772e17143c904a868d4d3e13ea4edbc83a9f98625844f6beac1277e2129", + "workflows/qwen-04.json": "c4ebf613b7fe95ed256c2a25def1bd7826fa73ed493e6097e2ef79444ef74bac", + "workflows/pf-img100-06-t1-s42.json": "75065b3087aad24e846d0bb973b5444d3c5e0e65e0ff5a85248c97299fb59e2e", + "workflows/pf-img100-02-t2-s43.json": "21ae7b3ab8c5033ba124bd08f4ffd3b50f072f5e9576e941ade085a877475836", + "workflows/pf-img100-02-t3-s43.json": "f5fbe8031404dde65ba8f4987797007209d472b0588e4d48bdc946aa43a94a6c", + "workflows/pf-img100-01-t5-s43.json": "f1546c8c65b8bd5e6b6ead0203f90d9b9d58fb21624a51f7a7e416ffb16af296", + "workflows/pf-img100-01-t4-s43.json": "40c851bffaab0347ec646d06b7f7ac58cdaac3b45d1f3bdcb844ce776573289b", + "workflows/yue2-02.json": "0b0decfdfee681258e987290b58c31461d4929468628aaca161b76225bc4c09e", + "workflows/wan-02.json": "d84e1cad9b2e385a8f6560f7bb2037267d11cc383785728ad7136e8efac34d65", + "workflows/pf-img100-20-t3-s42.json": "cd848c79a8302d5f2539d7b2b28e2412d0da4c6bbea307191ae85185ccca55be", + "workflows/pf-img100-20-t2-s42.json": "877cce3e64e8edfc1fa11e3a8f3e6238334a137ea6710121845d85fc13781f13", + "workflows/pf-img100-04-t5-s42.json": "e1029a71844b6c827ce716e4814fbebcf80520a99c70e5ff63d4230e3f7e136b", + "workflows/pf-img100-04-t4-s42.json": "9cbc7ca1ddfc202ddbacd217c4ae4ed7fc9ff0c23da6efeda784673f78d848df", + "workflows/pf-img100-16-t1-s43.json": "9f54fb65b552983dcf4897145f07f8e89a2fa5d122b1841211bd27bdbef99023", + "workflows/pf-img100-08-t3-s43.json": "945f0d17fb530c5c7344254954b7d0672529bf2eda98771b8d66e6448907fec1", + "workflows/pf-img100-08-t2-s43.json": "10f0848b7744ed6c09d9abfaf1c75e02e7b0e83130115caa1e0bd0f5b3932f2c", + "workflows/qwen-03.json": "923c283c6c2f9230299d3c633deebb620aefc38f74e9fd1cc485e8e3511c209a", + "workflows/pf-img100-09-t1-s43.json": "cb742c519147b037809ade5fc93e63ef04a9ff690ee4be9b55f9db2d4c232af9", + "workflows/pf-img100-02-t5-s43.json": "38403ace19dad13b80cdedf5b80e7df1287ded371ec09ddf141f7b9b5b605169", + "workflows/pf-img100-02-t4-s43.json": "79b56a8d0e04dfc1a8778221c8d804390e4d2cab67eed7524cc9bf8b92f0d324", + "workflows/pf-img100-05-t1-s42.json": "e7930d33f62d5e2041a39c055a2cde3c59b5368a3ddc7b96bcb4c0ae38a7dff1", + "workflows/pf-img100-01-t2-s43.json": "41623117f79244e095c7d78e50ae648fcf3f1a4822829ab453029beea344837f", + "workflows/pf-img100-01-t3-s43.json": "2509d69feb665f1e6da93a54d4f703aae7c277686757436bd11a320bf6b98f0b", + "workflows/pf-img100-20-t4-s42.json": "46e72eb5e2fb104e825ecca535d4b9822965ea696c98de80d75f6d175ab89213", + "workflows/pf-img100-20-t5-s42.json": "380a47774a298a2ce3b1bb12bd13f81ad28b9d2d1751d1b9affdff9820c9654f", + "workflows/pf-img100-04-t2-s42.json": "3f4c18ceaf1327de2bd4ede333af57c1e7a1f5cd25dcafefd91b739e0b786bdd", + "workflows/pf-img100-04-t3-s42.json": "600f1995d2e55415fd86d5d16753c45a01e97426542171aae2d5834babd4cebc", + "workflows/cog-01.json": "3c5947dea7d5687b93b3ef13c7f6c746e69337dbb618fdab6f77274402f2ef56", + "workflows/h3-legacy-failure.json": "5c7fd74009dedf7cb5dc21d0a58b9f7ccf81c8bb08d27134601299659c3034a3", + "workflows/pf-img100-04-t2-s43.json": "e00f923da14d3b2f5e1fca78ab1b6f2fad700f24da4aa3d84765d37bb292e58c", + "workflows/pf-img100-04-t3-s43.json": "1014331b2e1fcaf370004e7ad66e91ea2350a9580cd519eccc7c18766cc112eb", + "workflows/ltx-01.json": "c6abcc8def208e4fe7733297e3353e32b50570a417e3a8dbf6c814194e5fede3", + "workflows/pf-img100-20-t4-s43.json": "90306f0a755419a6f9db9c3f8635946edb256dfe569fa4871bee14d7c9787785", + "workflows/pf-img100-20-t5-s43.json": "1ac73ff33e46c92e9e480622cd213249c197f92baf3911047d9c9056a405e1c4", + "workflows/pf-img100-01-t2-s42.json": "d45a430ce08c44a993b7bfe209210544ff4db7eb72c019a3d1179cfd5bd605c5", + "workflows/pf-img100-01-t3-s42.json": "fb0880905acbf14a2e59c48aef3b433e515da82195ee9095b8c4ef070d1b796f", + "workflows/flux1-01.json": "84a2a8d6c8a6794c5cda2addf2931f96a1472cb95b7132a9731d7f932866fed2", + "workflows/pf-img100-05-t1-s43.json": "3b2e0e0d9fca0ca157071e3f649fc90261ce35356a2d0d07b611989d09670a38", + "workflows/pf-img100-02-t5-s42.json": "d4aa4fc5019fef34ef68d5b10e30ce9f945bf108c8e1f762531a3392e102a2d7", + "workflows/pf-img100-02-t4-s42.json": "cb9095dfea8829fafd413945edc59c86d77b0bdf16027108ddd3bb9c8d6b32b9", + "workflows/pf-img100-09-t1-s42.json": "0e87990cc6c626e8ed86f755f6143c4017af5a14048fea6c5a53b6aafd10cd10", + "workflows/qwen-02.json": "923c283c6c2f9230299d3c633deebb620aefc38f74e9fd1cc485e8e3511c209a", + "workflows/pf-img100-08-t3-s42.json": "9eaf604deeac014726befc983a1890c7a9f2466a756b6d76d80023ac9039ed9f", + "workflows/pf-img100-08-t2-s42.json": "e9415062042aebcc2b9256c4d08c449c84914ef40a901e9a0bfa98419f10882d" + }, + "old_record_file_bytes_preserved": { + "records/qwen-01.json": "743fee320a0a7792eb2cfc3ff9f757e2efb8ce15f502e01d73e5201f3d4b125f", + "records/flux3-01.json": "7fc7f127ea15e80b628bfc727186948b1c91d1de392d6ca78e5c4a5b3070bb05", + "records/flux3-02.json": "5e3f1b95d62ca0f3facc27471d775b56f82017be14e8e2cf0b919909366647c7", + "records/qwen-02.json": "fbc964c392dee9eaf91dc3d521b1cd3418a0769438dd9ec9c1c5eaa3294c2e3e", + "records/qwen-03.json": "055cf4602485eb771c87d027b39309c6eba706cb362091a47b69014779a87c3d", + "records/qwen-04.json": "066ac69bf74c1b5d1c68da31b1106c780d2654ec3740c7927b9ad9883b60274a", + "records/flux3-03.json": "78522294de00fd1dfc61b87da2cf36c76d96a96e6839fe69642f5d679d0661ab", + "records/flux3-04.json": "022a4fca28010f1f9c86bd090cff60ae3b606e276ba3d85887eb4fbb8b0a3047", + "records/flux1-01.json": "49e9e55d02fe1178c18b6e5eebc469bb98779bfc395a33da28f4846189b9c68f", + "records/flux1-02.json": "5289063c5d462e9228ac7f7c7e55a63db05d2d313746ba0c5e0ec1a8199572ed", + "records/yue2-01.json": "ecc3eabdf923cada72589e40d65110f67643cb25c5a588d3ae93d2d4077d066c", + "records/yue2-02.json": "087e23ef8e60ed140a75e8a2351374ebf6344aba888d460338dd98861ccf9e89", + "records/wan-01.json": "29feffe49c1ab54065883d891d4f42e0a85b74e37ce79bd86b18b2c879180bbd", + "records/wan-02.json": "e0b5fe76645860741d7864a750282c8499dc72d8824032e88a78e38dd59fa64b", + "records/wan-03.json": "b3834c59078281f1dc30d208f6eaa88637df5b00a48043a418eaae9f2a0de72a", + "records/ltx-01.json": "b1b5c7b4679876e64856b036e0bed34ed062e46ba7ee9762b946d86e7850c538", + "records/ltx-02.json": "51b183d893d28c76d5c8f97245bd34ee76f46965ca60b1e371e69a0b3027562d", + "records/ltx-03.json": "3a7c289c0fc632c60cdf4dbbeacc828f02b1bc9087dd1a815b150edb4657c867", + "records/ltx-04.json": "b75a8f130e51cb00bff7b662c4b6082ed79614de0fbb56438997c460ed37a1f7", + "records/cog-01.json": "de39838f2a0bfa956a8b5754a2ec8c94d19a762f3c8c16cb474d31aa32a37ff3", + "records/cog-02.json": "26129ae7d4bb8bb5cc07a241d9148cd20081536f82d900e7982f1fa28f105618", + "records/pf-img100-01-t1-s42.json": "d9b5870b07b0e5f4fb64d4ec56c9aa0ac8fa9789795a20766f890640032eb03f", + "records/pf-img100-01-t1-s43.json": "f394c83d962c39ec21503e2b4e108326ec00c29afaeb5f383afca8e6a708b4e5", + "records/pf-img100-02-t1-s42.json": "92bb5216208a0ea8269892f644f0b490b09d748d8d07a4709da7e044ee71a6cc", + "records/pf-img100-02-t1-s43.json": "d96d96a5663c115f02e0efbf88889179a581fa95e93365b81689fb26e4d0e884", + "records/pf-img100-03-t1-s42.json": "483258c981fd7002f3aca3aabdfe79614cac81881d1929a20b038a30441b53af", + "records/pf-img100-03-t1-s43.json": "e0760f8c1384507f1b060b187cbee003dcc7a9894906935c4feb9751e1dec29e", + "records/pf-img100-04-t1-s42.json": "31aefc76b379d3717b0c2de41c255c4faaf2c8089c33225e9fb24f9f0cc5d325", + "records/pf-img100-04-t1-s43.json": "1620da37b2e05281e6e7aa3541d6a738d14ceacd1ac9f552b9c883b3635a9f14", + "records/pf-img100-05-t1-s42.json": "7c3280ce1234e1c2022957f7ff53ad256aa1734037c1cd5d6407b952d72312bf", + "records/pf-img100-05-t1-s43.json": "d8c0acdde5cc62b9125b2ad9c252939f17d6335c0450ba564c18f3ace65f8194", + "records/pf-img100-06-t1-s42.json": "eee57aba85e4cc1f61de9c38d349ba998b556180ceccaf1402148947734e27b6", + "records/pf-img100-06-t1-s43.json": "2574f67e32160398d3bac75cdce2a9391310bbc46ecd3ceeec49ba63f695f248", + "records/pf-img100-08-t1-s42.json": "b92d198fde106601aeeefdba7b73056efcdf5ece22792edf6dd7645992c2c142", + "records/pf-img100-08-t1-s43.json": "d181dde1c5bde4088b025903935c7871f3c450fe9a4ca2d0270bbac6c8e5ecd3", + "records/pf-img100-09-t1-s42.json": "de364b2ba02fa4f96e7551a5af4ad758d307e51ad9e6a889ab95a621e2115c7a", + "records/pf-img100-09-t1-s43.json": "2e0d704d2dd711afd0c5758f0f38f09fa1cdb39cfec29fbe14b35459bb05b61c", + "records/pf-img100-16-t1-s42.json": "07aaf8bdc8028b57070b8549756adf93d1599440eff0198801635dcbbb425eea", + "records/pf-img100-16-t1-s43.json": "e139800b58d6330bd12b6a93c26735ab9fe96cc3699effe11335633dd3d9139b", + "records/pf-img100-20-t1-s42.json": "a21c5efda217dd3912b8b05e9ef3086f933f7570f8116ed25c2647aba4366027", + "records/pf-img100-20-t1-s43.json": "496a6c257d804dfd23fda9ac41caea01002590b09ccd2f0983fb2b67daac3690", + "records/pf-img100-01-t2-s42.json": "7e303d2569d6b22c29accd55155f4776c1baf0cbf0606b6887b069c84c07ced7", + "records/pf-img100-01-t2-s43.json": "98bc99d371081aab4eb0fee546dcaad4e996a4af7647483a0f9c53113b6e1c49", + "records/pf-img100-02-t2-s42.json": "ce22ac677eaee78e5c768e6022b3229f9e84c82d4e367a3a7ee169dae38ba99d", + "records/pf-img100-02-t2-s43.json": "3dc77450f91d7eb94348133bf7b3b10acf4affd6ac923df09147fc8838a5376f", + "records/pf-img100-03-t2-s42.json": "33a8ee489e4187c9c99c0c6866f181257cb6c1717549921ede1585197ac22b93", + "records/pf-img100-03-t2-s43.json": "0a77045399c3fc788becfd0f33f29d04aab5cd920d36b3065b2bc6924865006f", + "records/pf-img100-04-t2-s42.json": "7ace46e0e6ad426c413f90bbbb43a527268bf2730920cab77ae49ccaddb24963", + "records/pf-img100-04-t2-s43.json": "4073b3667d5a31c3b56b8eaa6b284ba792a0835f9187052109efc4f18b043308", + "records/pf-img100-05-t2-s42.json": "04a8c66c880c9a01c4d3817d62ee67fec40a592ca0400774bf20626ebfe9fde3", + "records/pf-img100-05-t2-s43.json": "f1488c04399076b3cacdff209be18116f613f4a1bedc1c408072786d1658ec19", + "records/pf-img100-06-t2-s42.json": "1fbc9afbf4b4d3f4484fa7601d693e9d7b2138dd8295cbe8696fbc21cad05bb0", + "records/pf-img100-06-t2-s43.json": "85fc64715ce07d0ab581960702bf96bc9e441032ef39d94e32cc4c9282f1149c", + "records/pf-img100-08-t2-s42.json": "764fccb7e5e2541a0d76cf472d76549147feb07c8cbd2dce7a7921de0ca39ab9", + "records/pf-img100-08-t2-s43.json": "27e42654cd750718dc853b51f2814153f4a92f4663c7e4f0b46ab12c3332090a", + "records/pf-img100-09-t2-s42.json": "aa14bb296388eeb623ece6015eaacc6aec6588af622178b191349d3da604c072", + "records/pf-img100-09-t2-s43.json": "a8b5a24a28ac3d00ff4fdcd1955d819230f94618483ea9c0817cd64ca3e31c2c", + "records/pf-img100-16-t2-s42.json": "380a850b991821a1d92fb4815652373f71f60971506ec2ec459091a888f2dbc8", + "records/pf-img100-16-t2-s43.json": "14c47a7ffa35a662e55e10fdce121ea69de0e68995074f3ec0d1837d18d60528", + "records/pf-img100-20-t2-s42.json": "6eac263be67f016b19fd6ab652edd95fdb26cfcdec45361307615819b0c405da", + "records/pf-img100-20-t2-s43.json": "8c8e7baa3558ce17c423908b5b86db9129128156266a217f2a4368bd45e836db", + "records/pf-img100-01-t3-s42.json": "2d0f48b22817cfe7eb1bd7ed2cb5be41955339047a1d4a2a99149b987f0ff237", + "records/pf-img100-01-t3-s43.json": "d72dd89c1d27b837c42cb6fa70840cf9251d3fef392cb5cfe0212f31da09b1bb", + "records/pf-img100-02-t3-s42.json": "320e1585d75e6da9b321e18505c050feff275297cf2f7decb2ee1156d7da21c0", + "records/pf-img100-02-t3-s43.json": "065f8d6ef12e3ed5a9364b2c98a1c1ab1a9d5e68f2e0f3c32203f18098ca1b71", + "records/pf-img100-03-t3-s42.json": "4103561b335a02113f4b1c3834b84d8f24fae4d80998876ace4ff49710f0fd72", + "records/pf-img100-03-t3-s43.json": "d4acde036531ae9d8de9c91d9401adb758757bd5c169b1ba9e7127a3238958b4", + "records/pf-img100-04-t3-s42.json": "5287d0afa8ebd2b8b52914594b285d39f068c416f5dca2f255eaa720f1754e3b", + "records/pf-img100-04-t3-s43.json": "9b240f99ab5077849620e88c4417dbf0af8a846bf6f580325e1ad1615bc463ae", + "records/pf-img100-05-t3-s42.json": "bcff9f36b8c2d449cd3d29f7221e84e24bdd975af1ad6998f6bce32870598d82", + "records/pf-img100-05-t3-s43.json": "79adfd16188d4f38f6ce2aa62e4b4aa58a94e72b33e1b4175255b895c99fc2c0", + "records/pf-img100-06-t3-s42.json": "ec475d29f226d147b405e1212979de59f4099087b1d0403d76f988c6cc53c3c4", + "records/pf-img100-06-t3-s43.json": "68c63888a0315697707f29a240b4c701c5bc6c913b663a25163b823c1d4da926", + "records/pf-img100-08-t3-s42.json": "286ea30177fd614d501871ea4a82e9f8bae26d3f424ad3e0ae02232429cd309f", + "records/pf-img100-08-t3-s43.json": "77e440828f9d218440d08c10c17415c4bc757f25f0a3375cccfdfca84f419081", + "records/pf-img100-09-t3-s42.json": "3b91daa0fa846cee981bdb47487010054f0f8145714f38cc501dfd1239121fa9", + "records/pf-img100-09-t3-s43.json": "0b0058e40e0dfccbedf0911f8a7baae47e06a2408b5610a84385a6c8f028bd72", + "records/pf-img100-16-t3-s42.json": "97f2b3c266f11daf76ee33bdf75ac8e994e8a38a230a1f0804f47505247b74c0", + "records/pf-img100-16-t3-s43.json": "92005f99637268513bccf2b68a24a863d193158d0e9b919d481cfc0fa50a9e79", + "records/pf-img100-20-t3-s42.json": "9a940ec88e231b1bb71aab3e5a0889efb1e3f9f5fcf2649624b6e78014cfe575", + "records/pf-img100-20-t3-s43.json": "f1aabcbfc1647c6c5d1b376f9e8c2b4575349037604de06c499e6b1f0c1b7a67", + "records/pf-img100-01-t4-s42.json": "5b6c9f8add5e65724dcebdb060e902eefa3ecd2f86ce77ea1d47913899cb4603", + "records/pf-img100-01-t4-s43.json": "8583c48c9e321807ad836f8ad228aab18799c4e631e7a490bab35143b5686482", + "records/pf-img100-02-t4-s42.json": "9855a3ea55e4fc1524531354921a205b9a664743a9cce5fbee0ae8842f3a1d72", + "records/pf-img100-02-t4-s43.json": "c5db703292afaef119aaff0c098ed53d6ed7966401b156b691ef0e2ba35e1505", + "records/pf-img100-03-t4-s42.json": "20f0d4c80ad68d898318f5362f3fdc0ca87ccc2303c3332e5ac82c1fb105d881", + "records/pf-img100-03-t4-s43.json": "e47a9f387ba0c12619ca93ed064e0582233c137fab28bc4ad0ecafbb9c42ced2", + "records/pf-img100-04-t4-s42.json": "f28363a2285c321372a046903565867575fe27bb9eaaf508d16a5fc235443020", + "records/pf-img100-04-t4-s43.json": "8d7ff7138c0bde064bf4d2ca09ce33534cbbd1221273099a5792472b13636002", + "records/pf-img100-05-t4-s42.json": "3ca0d0378ebdd3b999a9a037171fc5bd8db6378b23a29214dcaa84d8aa04ba3f", + "records/pf-img100-05-t4-s43.json": "534362f1c34881a5294358824274457c3a0f3b319316d990e52ee27c4d942e9c", + "records/pf-img100-06-t4-s42.json": "eb2265cda25601390762b26061e85cd7dbf5d7b70e2d42297abe8180bf4eb8af", + "records/pf-img100-06-t4-s43.json": "1c30d0de87fb2b6c145234688cb747ed9ebbc44a1fb747d8981b0da76b1aa9c8", + "records/pf-img100-08-t4-s42.json": "b6c6b0320e5c003e9e9d30dff4206fff353b153613954cce18416cb89d985865", + "records/pf-img100-08-t4-s43.json": "2bb9390f2abdb17051eb91600057fef030a7f74a164904af9a330815b00e0352", + "records/pf-img100-09-t4-s42.json": "84d1f7434bbc18294ecd4b78b1521bc007cec7de3583058467e59002d87eba6b", + "records/pf-img100-09-t4-s43.json": "1837b4640b1fcba60835cc5596ffb63e7b1fec4dee186ef11b81e139d922f8db", + "records/pf-img100-16-t4-s42.json": "cb67f33c44fbced7126cd58904dd10e326d2e287a29fc5dff8c2c59363006359", + "records/pf-img100-16-t4-s43.json": "b61cf5e6aa748db3e338baf8a513bb494f85204fa53dee5c4ef429d05e09c8cb", + "records/pf-img100-20-t4-s42.json": "efadc5157e1efb4ac3627c847b366550e3f0a5c23bacdabda7c4660b5be76f7a", + "records/pf-img100-20-t4-s43.json": "612a069aa9dccef794b157062168b09bae58eaa0a101ebc0f8d8930e2add8e17", + "records/pf-img100-01-t5-s42.json": "0f1b2a5138941587450b5590af69e18fbc58bc8a4b6229f8e64253d787755968", + "records/pf-img100-01-t5-s43.json": "cac98ed65b57bd4fb18eebaa276b18a5d81be971c6146b8978470463dad5f07a", + "records/pf-img100-02-t5-s42.json": "fe35e08e66f83fec54dce18de422ad0cc38a601527a26c3a62922b684d69e34a", + "records/pf-img100-02-t5-s43.json": "8276e1c2f4bc2444d79a40e14895eca45a998a9eae22ce923cc6236e8c6b145a", + "records/pf-img100-03-t5-s42.json": "d912f26e0e6019801eb13cbe6830583891d680521263e0fd2a7f0692baef73c5", + "records/pf-img100-03-t5-s43.json": "b8cc467b01867bb04e3b181467b1e81638b9fbd0b6ac2a98bd3c208d07bf8cf5", + "records/pf-img100-04-t5-s42.json": "d254795f925e49969dc8c96d70802f2a2046beb59ad4def0479d889e20114b8f", + "records/pf-img100-04-t5-s43.json": "51dfe3b16b40870019b22d9ab55183a187b7a62d6a0fd37df1e63166c16d6bf6", + "records/pf-img100-05-t5-s42.json": "eb245310e554418bd30c51cd4640a5330ec1328954ca42b623c5d2db8e8932de", + "records/pf-img100-05-t5-s43.json": "809b4fc9fa2c88c709db7e2f644f5d44ddf1bb6dce4100ecb546ec300c193a8b", + "records/pf-img100-06-t5-s42.json": "ed6d8d7453b3a6e640ba6db965f60bec7386c3856595d770597bc87fcfa3eb82", + "records/pf-img100-06-t5-s43.json": "7ad4b54cc8561a338bdb69c1030d7163920498b137e7a11b5e60ab59c51dcf7b", + "records/pf-img100-08-t5-s42.json": "362c28eda5ac7368bb32b1a28ed2daad36acf1055c73abab80f64471503886fe", + "records/pf-img100-08-t5-s43.json": "0bc6d4d7e27d4a53d24914f85b7f2ebb96df5909fa85111a9aca7ad0d96ab2da", + "records/pf-img100-09-t5-s42.json": "abb21456457f02aa2bc62afebb66e1cec8f0478750415e8cb44e6df28f05fd66", + "records/pf-img100-09-t5-s43.json": "3d43632a48fe7a94feaa4eb3af651099c59d2ace122d73f24836af7a8beda1ec", + "records/pf-img100-16-t5-s42.json": "9b734956b170cbda862d97a136e831e4bb1d366a421a8038370ec0acede83738", + "records/pf-img100-16-t5-s43.json": "dc28452555018b2407d39755251475e61c7c3e2060ff63ccc7b16672aee56afc", + "records/pf-img100-20-t5-s42.json": "7ff00b071732142055f47924680d56cd083cfe8c8a314d7ff921328be75922fd", + "records/pf-img100-20-t5-s43.json": "cba9bf4388d442189ce936eee336fd1d64af085391df29044f6d94795e7eecc8", + "records/cog-03.json": "6d8625d1dc49f2cee5a2f0220f4586ad6a534b3a7497ae81b8454d8393c18996", + "records/cog-04.json": "6418d81ad5446db349165644048cb4c6e5ea157ff3304efe2c4da2c49fc9d87e", + "records/wan-failure-01.json": "695c657835bdfcb9795b42c82eb52ddb8163633a07c6d402c6cd50d052a82ea2", + "records/ltx-failure-01.json": "afad4bfb484de5d7bed3b80418dc6653f0520059d99c899c0db83b6bac3c9d6d", + "records/ltx-failure-02.json": "8941e18c6f4c7d579e880e484ce6872af2a4d7a4a15fff550bb42e0495e78efb", + "records/cog-failure-01.json": "d46c537b0cdb57b25f2682d84e23f28fd6be566c9aa8858f2376650b215f9439", + "records/h3-legacy-failure.json": "bb86e67d68c24c8d73ab708fc0e1ac97d438f86ba53ac85815413a183dcc41e2", + "records/h3-dynamic-failure.json": "91a7d349da5e2b2d4ce14c1b04b44f865920b52d29a5c61c4841de43ae4c1642", + "records/yue2-cpu-cancelled.json": "c3d6bf893e709cbf862a7e020995440f7a5997e3ea40d250f8a8e15d38bfbe7c" + }, + "new_checks": [ + { + "id": "qwen-turbo-01", + "public_path": "media/qwen-turbo-01.png", + "source_sha256": "c4a363cb2d3fbcb5d51b02226bdb57facc1963d98cac306cefb3ad252700b3d0", + "public_sha256": "56129ce43e04bb2aca5849e23b9d6a0c9adfdee3047a9329f1fcc360c2d9eb2e", + "pixel_sha256": "5df320c6bdbdad6dc8dbc07af73f2dc513a55deecad9b2f49b659e85ba5ba2f4", + "full_decode": true, + "content_preserved": "Decoded pixels byte-identical", + "ancillary_metadata_removed": true + }, + { + "id": "qwen-turbo-02", + "public_path": "media/qwen-turbo-02.png", + "source_sha256": "f8bd3fcfba3462d120142e83fb0d7ec637a9eceebd25f6db11f61bda76254eb1", + "public_sha256": "238d33136a36561317196ed2a9c0096cbafe4109059ab2d0d8df5ec3079d74bd", + "pixel_sha256": "ac9f30aded9dc1c28b70d2b87cebd61a4df3d93e3dc62206d1ed31874410fe48", + "full_decode": true, + "content_preserved": "Decoded pixels byte-identical", + "ancillary_metadata_removed": true + }, + { + "id": "qwen-turbo-03", + "public_path": "media/qwen-turbo-03.png", + "source_sha256": "c5e2e3901c3129df9943c2f995d5b42c7d646c708eb6f9053295778ed77012e7", + "public_sha256": "57418cb5f47b5647b8773c0e012572e528e150b0f27f422e78f4f05d5a683a11", + "pixel_sha256": "ff22ec81f55d0251c62e225baca505db986d0d734728cb05f57b4d2087cd12d1", + "full_decode": true, + "content_preserved": "Decoded pixels byte-identical", + "ancillary_metadata_removed": true + }, + { + "id": "qwen-turbo-04", + "public_path": "media/qwen-turbo-04.png", + "source_sha256": "1c2b9ddeca53a1fd944073a596b050e43f8f9b1b4a7e2fee57608095395b97d3", + "public_sha256": "e5b3093fdb6bb39fcec4b803b1345733cf212bb274806338d27e140d3b882c3d", + "pixel_sha256": "2db9b0cf39ba6fdb3f0febf48a4f32d9fbd8012ed303d8d3ec711a4a074f8b81", + "full_decode": true, + "content_preserved": "Decoded pixels byte-identical", + "ancillary_metadata_removed": true + }, + { + "id": "qwen-turbo-05", + "public_path": "media/qwen-turbo-05.png", + "source_sha256": "5a9008f775161ef48b16915597d158f4a0cd80883a7c4d452cbf240bba049ce2", + "public_sha256": "94e64b09278ffff4feb04c3197545dd8d6371b281a514d0e3d1cf8550648d919", + "pixel_sha256": "18129fd4fc85277093ff6d330be029055bfde66f2a18af657ce735b6e928f894", + "full_decode": true, + "content_preserved": "Decoded pixels byte-identical", + "ancillary_metadata_removed": true + }, + { + "id": "qwen-turbo-06", + "public_path": "media/qwen-turbo-06.png", + "source_sha256": "bd1ac8888861a03d6b9a2436cede0ad7b3009f1535c11b360f794ad47a9c0321", + "public_sha256": "f1ae1c3294bcffe1c0d283e957bcc81c5d48df83fe6e0f316eb4182d57db2af1", + "pixel_sha256": "f74bd726756ed8f29d230b4c506533b708c21ecd53275092c3e54672dbb5a1cd", + "full_decode": true, + "content_preserved": "Decoded pixels byte-identical", + "ancillary_metadata_removed": true + }, + { + "id": "qwen-turbo-07", + "public_path": "media/qwen-turbo-07.png", + "source_sha256": "ed1e6c70637edfc74eb3f051c51632087649ce6e88c705b2c4f08f1ee81d756a", + "public_sha256": "7b4feea8f758374a5544ad0e9078c8455f32f5a48e8305256ab0839f2afd9472", + "pixel_sha256": "23fe51347e19cff881406c1152d713a2773a46cd9032f54cabc9dc4b5129c108", + "full_decode": true, + "content_preserved": "Decoded pixels byte-identical", + "ancillary_metadata_removed": true + }, + { + "id": "qwen-turbo-08", + "public_path": "media/qwen-turbo-08.png", + "source_sha256": "63abbaf5c49b2baca103bd65761fd4fed09f7e5f173406e91eaaf47d136694f2", + "public_sha256": "bd1bcc5079a48be9f59a4d72511eae46abb6e2aa788fe22f72bbac1cc4ba8b7a", + "pixel_sha256": "224125f9ca31c3c15dfa975dd61d42fa8f1703d9c012374876d2242054759658", + "full_decode": true, + "content_preserved": "Decoded pixels byte-identical", + "ancillary_metadata_removed": true + }, + { + "id": "qwen-turbo-09", + "public_path": "media/qwen-turbo-09.png", + "source_sha256": "5b992fa647b0461c1eb8514854a32b5f44cdd297a129a110f72b1abff6a830ef", + "public_sha256": "847ec6dd9be4503a94967f96ca4b91d223ffb7638749c4fd08fb318723f544e0", + "pixel_sha256": "ffaa52b367e79221e07aeed4066dcadc2039859372848f38c9bc028d623b7f50", + "full_decode": true, + "content_preserved": "Decoded pixels byte-identical", + "ancillary_metadata_removed": true + }, + { + "id": "qwen-turbo-10", + "public_path": "media/qwen-turbo-10.png", + "source_sha256": "f051ec1ab055f23d3fd6513e3428d807734a846f5e51881403cf419ccb33c161", + "public_sha256": "7bc1033039962507ee2aa153ebfacaa4cdd1f77e93e152bd7241ebae686b85bc", + "pixel_sha256": "9672ee9777924c1b99ccff8952362c0119f99a87bb1e22c7bc82a24b579ccb6f", + "full_decode": true, + "content_preserved": "Decoded pixels byte-identical", + "ancillary_metadata_removed": true + } + ], + "private_projection_scan_findings": [] +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/dataset.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/dataset.json index 671c1cb..c43937c 100644 --- a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/dataset.json +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/dataset.json @@ -26,19 +26,20 @@ } }, "coverage": { - "retained_output_files": 123, - "accepted_outputs": 122, - "images": 110, - "accepted_images": 109, + "retained_output_files": 133, + "accepted_outputs": 132, + "images": 120, + "accepted_images": 119, "image_complexity_suite": 100, "videos": 11, "music": 2, "contact_sheets": 11, "failed_delivery_outputs_with_media": 1, "failed_or_cancelled_submitted_jobs_without_media": 7, - "unique_original_media_sha256": 121, + "unique_original_media_sha256": 131, "diagnostic_fixture_exclusion": "Two synthetic codec preflight MP4s are fixtures, not model outputs; excluded from gallery and generated-output counts.", - "duplicate_media_note": "Two pairs of early accepted image files have byte-identical source hashes. Each delivery is retained;123files does not mean123unique contents." + "duplicate_media_note": "Two pairs of early accepted image files have byte-identical source hashes. Every delivery is retained; file count and unique original byte sequences are distinct.", + "qwen_turbo_ten_image_trial": 10 }, "methodology": [ "Same model, template and fixed generation settings; record any intentional prompt change and seed42 then43. A changed prompt forces fresh text encoding; a changed seed forces sampling.", @@ -181,7 +182,1376 @@ ] }, "quality": "All100 complexity-image thumbnails showed recognizable scenes; some lettering was malformed. The user described both YuE2 samples as usable music. Video results ranged from nearly static fox scenes and colored noise to recognizable scenes with poor framing and prompt adherence.", - "failed_jobs_preserved": 7 + "failed_jobs_preserved": 7, + "campaign_totals": { + "recorded_attempts": 140, + "retained_output_files": 133, + "images": 120, + "videos": 11, + "music": 2, + "accepted_outputs": 132, + "no_output_failures": 7, + "failed_delivery_outputs": 1 + }, + "original_overnight_scope": { + "recorded_attempts": 130, + "retained_output_files": 123, + "images": 110, + "image_complexity_suite": 100, + "videos": 11, + "music": 2 + }, + "qwen_turbo_followup_scope": { + "recorded_attempts": 10, + "retained_output_files": 10, + "images": 10, + "distinct_briefs": 9 + }, + "seed_variant_cache": { + "schema_version": "seed-variant-cache-1", + "scope": "Read-only reanalysis of the existing FLUX 100 images and completed ten-image Qwen Turbo trial. No additional generation for this analysis.", + "flux50_seed_pairs": { + "model": "Abiray/flux-3-image-klein-sd-GGUF:Q4_K_M", + "pairs": 50, + "observed_seed_variants_per_brief": 2, + "five_seed_variants_tested": false, + "configuration": { + "width": 512, + "height": 512, + "steps": 4, + "cfg": 1, + "generator_quant": "Q4_K_M", + "compute": "FP32", + "cpu_vae": true, + "lowvram": true + }, + "cache_counts": { + "second_positive_encoder_cached": 50, + "second_negative_encoder_cached": 50, + "second_sampler_cached": 0, + "same_actual_runtime_pair": 50, + "identical_graph_except_seed": 50, + "second_probe_is_first_probe": 50, + "distinct_output_hashes_per_pair": 50, + "first_positive_encoder_cached": 0 + }, + "all_pairs": { + "first_native_seconds": { + "n": 50, + "min": 64.204, + "median": 71.671, + "mean": 79.30762, + "max": 241.398, + "total": 3965.3810000000003 + }, + "second_native_seconds": { + "n": 50, + "min": 36.576, + "median": 38.397999999999996, + "mean": 39.505, + "max": 44.558, + "total": 1975.2499999999998 + }, + "native_seconds_saved": { + "n": 50, + "min": 27.309, + "median": 33.367000000000004, + "mean": 39.80262, + "max": 204.344, + "total": 1990.1310000000003 + }, + "native_speedup": { + "n": 50, + "min": 1.7400193433990652, + "median": 1.8708729216784072, + "mean": 2.002159338297874, + "max": 6.514762238894586, + "total": 100.10796691489372 + }, + "first_trace_seconds": { + "n": 50, + "min": 66.467404, + "median": 74.03037850000001, + "mean": 81.160127, + "max": 242.338818, + "total": 4058.0063499999997 + }, + "second_trace_seconds": { + "n": 50, + "min": 39.273643, + "median": 39.397068000000004, + "mean": 41.17113458, + "max": 45.413099, + "total": 2058.556729 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 50, + "min": 10.422, + "median": 22.38, + "mean": 21.21556, + "max": 31.712, + "total": 1060.778 + }, + "first_positive_encoding_seconds": { + "n": 50, + "min": 27.51600000000326, + "median": 31.68800000002375, + "mean": 36.278460000006014, + "max": 56.32800000003772, + "total": 1813.9230000003008 + } + }, + "exclude_initial_load_pair": { + "first_native_seconds": { + "n": 49, + "min": 64.204, + "median": 70.046, + "mean": 75.99965306122449, + "max": 101.131, + "total": 3723.983 + }, + "second_native_seconds": { + "n": 49, + "min": 36.576, + "median": 38.491, + "mean": 39.555020408163266, + "max": 44.558, + "total": 1938.196 + }, + "native_seconds_saved": { + "n": 49, + "min": 27.309, + "median": 32.125, + "mean": 36.44463265306123, + "max": 56.761, + "total": 1785.7870000000003 + }, + "native_speedup": { + "n": 49, + "min": 1.7400193433990652, + "median": 1.8471559294322406, + "mean": 1.9100654015510024, + "max": 2.279265269326121, + "total": 93.59320467599913 + }, + "first_trace_seconds": { + "n": 49, + "min": 66.467404, + "median": 72.559234, + "mean": 77.87076595918367, + "max": 102.832423, + "total": 3815.6675319999995 + }, + "second_trace_seconds": { + "n": 49, + "min": 39.273643, + "median": 39.3951, + "mean": 41.20691226530612, + "max": 45.413099, + "total": 2019.138701 + }, + "first_positive_encoding_seconds": { + "n": 49, + "min": 27.51600000000326, + "median": 31.985000000044238, + "mean": 36.452836734700085, + "max": 56.32800000003772, + "total": 1786.189000000304 + } + }, + "by_tier": { + "1": { + "first_native_seconds": { + "n": 10, + "min": 64.57, + "median": 64.88550000000001, + "mean": 82.5835, + "max": 241.398, + "total": 825.835 + }, + "second_native_seconds": { + "n": 10, + "min": 36.576, + "median": 36.931, + "mean": 36.9348, + "max": 37.258, + "total": 369.348 + }, + "native_seconds_saved": { + "n": 10, + "min": 27.545, + "median": 28.106, + "mean": 45.6487, + "max": 204.344, + "total": 456.487 + }, + "first_trace_seconds": { + "n": 10, + "min": 66.467404, + "median": 66.59116850000001, + "mean": 84.1501359, + "max": 242.338818, + "total": 841.5013589999999 + }, + "second_trace_seconds": { + "n": 10, + "min": 39.273643, + "median": 39.350615, + "mean": 39.362302899999996, + "max": 39.476116, + "total": 393.623029 + }, + "first_positive_encoding_seconds": { + "n": 10, + "min": 27.73399999999674, + "median": 27.859499999991385, + "mean": 27.93749999999418, + "max": 28.485000000044238, + "total": 279.3749999999418 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 10, + "min": 10.422, + "median": 26.314999999999998, + "mean": 21.0136, + "max": 27.701, + "total": 210.136 + } + }, + "2": { + "first_native_seconds": { + "n": 10, + "min": 64.204, + "median": 64.979, + "mean": 66.5586, + "max": 70.046, + "total": 665.5860000000001 + }, + "second_native_seconds": { + "n": 10, + "min": 36.682, + "median": 37.0485, + "mean": 37.4005, + "max": 38.292, + "total": 374.005 + }, + "native_seconds_saved": { + "n": 10, + "min": 27.309, + "median": 28.152, + "mean": 29.1581, + "max": 32.125, + "total": 291.581 + }, + "first_trace_seconds": { + "n": 10, + "min": 66.470952, + "median": 66.60241149999999, + "mean": 68.0339363, + "max": 72.466618, + "total": 680.3393630000002 + }, + "second_trace_seconds": { + "n": 10, + "min": 39.288523, + "median": 39.369251000000006, + "mean": 39.3625179, + "max": 39.4423, + "total": 393.625179 + }, + "first_positive_encoding_seconds": { + "n": 10, + "min": 27.51600000000326, + "median": 28.094000000011874, + "mean": 29.17840000001015, + "max": 31.985000000044238, + "total": 291.7840000001015 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 10, + "min": 11.778, + "median": 24.357, + "mean": 23.0208, + "max": 31.712, + "total": 230.20799999999997 + } + }, + "3": { + "first_native_seconds": { + "n": 10, + "min": 69.077, + "median": 69.54599999999999, + "mean": 70.9772, + "max": 73.464, + "total": 709.7719999999999 + }, + "second_native_seconds": { + "n": 10, + "min": 38.008, + "median": 38.397999999999996, + "mean": 38.4107, + "max": 38.705, + "total": 384.107 + }, + "native_seconds_saved": { + "n": 10, + "min": 30.839, + "median": 31.393, + "mean": 32.5665, + "max": 34.852, + "total": 325.66499999999996 + }, + "first_trace_seconds": { + "n": 10, + "min": 69.48063, + "median": 72.549484, + "mean": 72.8756681, + "max": 75.63889, + "total": 728.7566810000001 + }, + "second_trace_seconds": { + "n": 10, + "min": 39.287177, + "median": 39.335657499999996, + "mean": 39.3440559, + "max": 39.400407, + "total": 393.440559 + }, + "first_positive_encoding_seconds": { + "n": 10, + "min": 30.90600000001723, + "median": 31.38300000000163, + "mean": 32.56720000000787, + "max": 34.78099999995902, + "total": 325.6720000000787 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 10, + "min": 18.393, + "median": 21.78, + "mean": 21.6223, + "max": 26.858, + "total": 216.223 + } + }, + "4": { + "first_native_seconds": { + "n": 10, + "min": 77.877, + "median": 78.6135, + "mean": 78.4776, + "max": 78.736, + "total": 784.7760000000001 + }, + "second_native_seconds": { + "n": 10, + "min": 40.423, + "median": 40.4815, + "mean": 40.5054, + "max": 40.68, + "total": 405.05400000000003 + }, + "native_seconds_saved": { + "n": 10, + "min": 37.423, + "median": 37.981, + "mean": 37.9722, + "max": 38.269, + "total": 379.72200000000004 + }, + "first_trace_seconds": { + "n": 10, + "min": 78.600529, + "median": 81.565219, + "mean": 80.7096113, + "max": 81.696329, + "total": 807.0961130000001 + }, + "second_trace_seconds": { + "n": 10, + "min": 42.349654, + "median": 42.403067, + "mean": 42.4041422, + "max": 42.532345, + "total": 424.041422 + }, + "first_positive_encoding_seconds": { + "n": 10, + "min": 37.65600000001723, + "median": 38.10950000002049, + "mean": 38.05140000000247, + "max": 38.32799999997951, + "total": 380.5140000000247 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 10, + "min": 13.12, + "median": 14.9955, + "mean": 15.7422, + "max": 21.868, + "total": 157.42200000000003 + } + }, + "5": { + "first_native_seconds": { + "n": 10, + "min": 96.483, + "median": 97.2595, + "mean": 97.94120000000001, + "max": 101.131, + "total": 979.4119999999999 + }, + "second_native_seconds": { + "n": 10, + "min": 44.121, + "median": 44.253, + "mean": 44.2736, + "max": 44.558, + "total": 442.73600000000005 + }, + "native_seconds_saved": { + "n": 10, + "min": 52.362, + "median": 53.0135, + "mean": 53.6676, + "max": 56.761, + "total": 536.676 + }, + "first_trace_seconds": { + "n": 10, + "min": 96.80878, + "median": 99.68121, + "mean": 100.0312834, + "max": 102.832423, + "total": 1000.3128339999998 + }, + "second_trace_seconds": { + "n": 10, + "min": 45.341639, + "median": 45.3771555, + "mean": 45.382654, + "max": 45.413099, + "total": 453.82654 + }, + "first_positive_encoding_seconds": { + "n": 10, + "min": 52.40600000001723, + "median": 53.0625, + "mean": 53.65780000001541, + "max": 56.32800000003772, + "total": 536.5780000001541 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 10, + "min": 12.666, + "median": 27.817, + "mean": 24.6789, + "max": 29.406, + "total": 246.78900000000004 + } + } + }, + "pair_evidence": [ + { + "first_record": "pf-img100-01-t1-s42", + "second_record": "pf-img100-01-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 241.398, + "second_native_seconds": 37.054, + "native_inter_run_gap_seconds": 14.139, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-02-t1-s42", + "second_record": "pf-img100-02-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 65.35, + "second_native_seconds": 37.258, + "native_inter_run_gap_seconds": 11.281, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-03-t1-s42", + "second_record": "pf-img100-03-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.767, + "second_native_seconds": 37.222, + "native_inter_run_gap_seconds": 26.604, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-04-t1-s42", + "second_record": "pf-img100-04-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.679, + "second_native_seconds": 37.016, + "native_inter_run_gap_seconds": 26.543, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-05-t1-s42", + "second_record": "pf-img100-05-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.57, + "second_native_seconds": 36.908, + "native_inter_run_gap_seconds": 10.422, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-06-t1-s42", + "second_record": "pf-img100-06-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.943, + "second_native_seconds": 36.896, + "native_inter_run_gap_seconds": 27.239, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-08-t1-s42", + "second_record": "pf-img100-08-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.791, + "second_native_seconds": 36.576, + "native_inter_run_gap_seconds": 13.335, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-09-t1-s42", + "second_record": "pf-img100-09-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.828, + "second_native_seconds": 36.708, + "native_inter_run_gap_seconds": 27.701, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-16-t1-s42", + "second_record": "pf-img100-16-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 65.354, + "second_native_seconds": 36.954, + "native_inter_run_gap_seconds": 26.087, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-20-t1-s42", + "second_record": "pf-img100-20-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 65.155, + "second_native_seconds": 36.756, + "native_inter_run_gap_seconds": 26.785, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-01-t2-s42", + "second_record": "pf-img100-01-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.204, + "second_native_seconds": 36.895, + "native_inter_run_gap_seconds": 11.778, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-02-t2-s42", + "second_record": "pf-img100-02-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.965, + "second_native_seconds": 36.897, + "native_inter_run_gap_seconds": 31.712, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-03-t2-s42", + "second_record": "pf-img100-03-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.918, + "second_native_seconds": 36.682, + "native_inter_run_gap_seconds": 26.206, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-04-t2-s42", + "second_record": "pf-img100-04-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.993, + "second_native_seconds": 37.056, + "native_inter_run_gap_seconds": 27.134, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-05-t2-s42", + "second_record": "pf-img100-05-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.762, + "second_native_seconds": 37.041, + "native_inter_run_gap_seconds": 11.901, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-06-t2-s42", + "second_record": "pf-img100-06-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 70.046, + "second_native_seconds": 37.921, + "native_inter_run_gap_seconds": 22.36, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-08-t2-s42", + "second_record": "pf-img100-08-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.192, + "second_native_seconds": 38.292, + "native_inter_run_gap_seconds": 22.4, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-09-t2-s42", + "second_record": "pf-img100-09-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.379, + "second_native_seconds": 36.876, + "native_inter_run_gap_seconds": 28.003, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-16-t2-s42", + "second_record": "pf-img100-16-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 68.971, + "second_native_seconds": 38.169, + "native_inter_run_gap_seconds": 24.357, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-20-t2-s42", + "second_record": "pf-img100-20-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.156, + "second_native_seconds": 38.176, + "native_inter_run_gap_seconds": 24.357, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-01-t3-s42", + "second_record": "pf-img100-01-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.601, + "second_native_seconds": 38.305, + "native_inter_run_gap_seconds": 23.193, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-02-t3-s42", + "second_record": "pf-img100-02-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.491, + "second_native_seconds": 38.008, + "native_inter_run_gap_seconds": 22.005, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-03-t3-s42", + "second_record": "pf-img100-03-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.451, + "second_native_seconds": 38.148, + "native_inter_run_gap_seconds": 22.876, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-04-t3-s42", + "second_record": "pf-img100-04-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.125, + "second_native_seconds": 38.262, + "native_inter_run_gap_seconds": 26.858, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-05-t3-s42", + "second_record": "pf-img100-05-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.388, + "second_native_seconds": 38.491, + "native_inter_run_gap_seconds": 24.166, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-06-t3-s42", + "second_record": "pf-img100-06-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 73.296, + "second_native_seconds": 38.687, + "native_inter_run_gap_seconds": 18.854, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-08-t3-s42", + "second_record": "pf-img100-08-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 73.437, + "second_native_seconds": 38.705, + "native_inter_run_gap_seconds": 18.393, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-09-t3-s42", + "second_record": "pf-img100-09-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.077, + "second_native_seconds": 38.238, + "native_inter_run_gap_seconds": 21.555, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-16-t3-s42", + "second_record": "pf-img100-16-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 73.442, + "second_native_seconds": 38.651, + "native_inter_run_gap_seconds": 19.37, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-20-t3-s42", + "second_record": "pf-img100-20-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 73.464, + "second_native_seconds": 38.612, + "native_inter_run_gap_seconds": 18.953, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-01-t4-s42", + "second_record": "pf-img100-01-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.215, + "second_native_seconds": 40.46, + "native_inter_run_gap_seconds": 21.868, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-02-t4-s42", + "second_record": "pf-img100-02-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.617, + "second_native_seconds": 40.423, + "native_inter_run_gap_seconds": 13.502, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-03-t4-s42", + "second_record": "pf-img100-03-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.656, + "second_native_seconds": 40.474, + "native_inter_run_gap_seconds": 14.668, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-04-t4-s42", + "second_record": "pf-img100-04-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.617, + "second_native_seconds": 40.612, + "native_inter_run_gap_seconds": 15.534, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-05-t4-s42", + "second_record": "pf-img100-05-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.467, + "second_native_seconds": 40.51, + "native_inter_run_gap_seconds": 15.323, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-06-t4-s42", + "second_record": "pf-img100-06-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 77.877, + "second_native_seconds": 40.454, + "native_inter_run_gap_seconds": 18.954, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-08-t4-s42", + "second_record": "pf-img100-08-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.727, + "second_native_seconds": 40.458, + "native_inter_run_gap_seconds": 15.354, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-09-t4-s42", + "second_record": "pf-img100-09-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.254, + "second_native_seconds": 40.489, + "native_inter_run_gap_seconds": 14.643, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-16-t4-s42", + "second_record": "pf-img100-16-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.61, + "second_native_seconds": 40.68, + "native_inter_run_gap_seconds": 13.12, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-20-t4-s42", + "second_record": "pf-img100-20-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.736, + "second_native_seconds": 40.494, + "native_inter_run_gap_seconds": 14.456, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-01-t5-s42", + "second_record": "pf-img100-01-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.733, + "second_native_seconds": 44.245, + "native_inter_run_gap_seconds": 27.851, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-02-t5-s42", + "second_record": "pf-img100-02-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.058, + "second_native_seconds": 44.319, + "native_inter_run_gap_seconds": 13.702, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-03-t5-s42", + "second_record": "pf-img100-03-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.156, + "second_native_seconds": 44.279, + "native_inter_run_gap_seconds": 28.917, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-04-t5-s42", + "second_record": "pf-img100-04-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 96.93, + "second_native_seconds": 44.162, + "native_inter_run_gap_seconds": 28.792, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-05-t5-s42", + "second_record": "pf-img100-05-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.26, + "second_native_seconds": 44.231, + "native_inter_run_gap_seconds": 28.36, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-06-t5-s42", + "second_record": "pf-img100-06-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 101.049, + "second_native_seconds": 44.558, + "native_inter_run_gap_seconds": 24.123, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-08-t5-s42", + "second_record": "pf-img100-08-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.259, + "second_native_seconds": 44.261, + "native_inter_run_gap_seconds": 29.406, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-09-t5-s42", + "second_record": "pf-img100-09-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.353, + "second_native_seconds": 44.19, + "native_inter_run_gap_seconds": 27.783, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-16-t5-s42", + "second_record": "pf-img100-16-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 96.483, + "second_native_seconds": 44.121, + "native_inter_run_gap_seconds": 12.666, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-20-t5-s42", + "second_record": "pf-img100-20-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 101.131, + "second_native_seconds": 44.37, + "native_inter_run_gap_seconds": 25.189, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + } + ], + "interpretation": "All 50 second requests reused positive and negative conditioning. Sampling was not cached, and every pair produced distinct outputs. The initial-load pair is excluded from the main steady-runtime comparison. No five-variant group was tested.", + "limits": [ + "Same42/43 seed ordering; no randomized cache-disabled control.", + "Node intervals include preparation/event delivery, not isolated kernels.", + "Observed cache survives the measured gaps but is local to one retained Comfy runtime; no persistence guarantee across model switches, restarts, factory changes or eviction." + ] + }, + "qwen_turbo_ten": { + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "generated_outputs": 10, + "first_native_start_utc": "2026-10-21T08:41:18.067000+00:00", + "distinct_briefs": 9, + "measured_seed_variants_in_first_brief": 2, + "five_seed_variants_tested": false, + "initial_native_seconds": 320.246, + "same_brief_variant_native_seconds": 95.234, + "native_inter_run_gap_seconds": 34.322, + "same_runtime_retained": true, + "only_graph_difference": [ + "/9/inputs/noise_seed" + ], + "second_encoder_cached": true, + "sampling_and_decoding_repeated": true, + "new_briefs_in_retained_runtime": { + "requests": 8, + "native_seconds": [ + 194.865, + 196.417, + 199.64, + 199.412, + 199.784, + 210.023, + 208.566, + 223.135 + ], + "minimum_seconds": 194.865, + "median_seconds": 199.712, + "maximum_seconds": 223.135, + "sum_seconds": 1631.842 + }, + "limits": [ + "The initial request includes loading plus text encoding, so the first/second delta is a combined startup/model/encoding benefit.", + "Only two same-brief seed outputs were measured; do not present five-variant performance as measured.", + "Fresh-prompt latency is a separate workload from cached same-brief variation." + ] + }, + "production_guidance": "Keep the complete composed brief/materials/settings identical and vary only the seed in consecutive requests to the same factory/profile/Comfy runtime. Cache state is local to that runtime and can be lost after restarts, model switches or eviction.", + "five_variant_group_measured": false + } }, "models": [ { @@ -458,7 +1828,7 @@ "See pinned upstream model terms." ], "findings": [ - "Exact0.9.8distilled locallyQ8converted; four clips105\u2013181s, but motion/anatomy/adherence inconsistent. One base pass, not official multiscale upscale pipeline." + "Exact0.9.8distilled locallyQ8converted; four clips105–181s, but motion/anatomy/adherence inconsistent. One base pass, not official multiscale upscale pipeline." ], "local_conversion": { "source_sha256": "76aa8c4786af752fa6f951947129d5290c3c6c0b2fadcadea6b5e114ae2cad8f", @@ -535,12 +1905,54 @@ "conversion_seconds": 41.84399999998277, "architecture": "CogVideoX2Bnonrotary" } + }, + { + "id": "qwen-turbo", + "label": "Qwen Image 2.1 Turbo Q4_K_M", + "source": "https://huggingface.co/unsloth/Qwen-Image-2.1-Turbo-GGUF", + "revision": "5d1f9f759ea877ae556cda88dbec958b62ad3809", + "weights": [ + { + "filename": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "revision": "5d1f9f759ea877ae556cda88dbec958b62ad3809", + "sha256": "8940e8b522af886e0c0f5afddf720e82dd223cb9a43580b5f61c4d740452ac20", + "url": "https://huggingface.co/unsloth/Qwen-Image-2.1-Turbo-GGUF/resolve/5d1f9f759ea877ae556cda88dbec958b62ad3809/qwen-image-2.1-turbo-Q4_K_M.gguf", + "repository": "unsloth/Qwen-Image-2.1-Turbo-GGUF", + "size_bytes": 4199565120 + }, + { + "filename": "text_encoders/qwen3vl_8b_int8_convrot.safetensors", + "revision": "a8342359c61979b06a8dafb1a47991fe1f30e994", + "sha256": "8bfd0f6e12abf2d2d697ecc888e5e90b0d6741d6708f05799f53afa560452e8f", + "url": "https://huggingface.co/Comfy-Org/Qwen-Image-2.1/resolve/a8342359c61979b06a8dafb1a47991fe1f30e994/text_encoders/qwen3vl_8b_int8_convrot.safetensors", + "repository": "Comfy-Org/Qwen-Image-2.1", + "size_bytes": 9350798360 + }, + { + "filename": "vae/qwen_image_2.1_vae_bf16.safetensors", + "revision": "a8342359c61979b06a8dafb1a47991fe1f30e994", + "sha256": "bb21f7473051e1ac368515dd3f2e15cd44d7a11748ee8823e1ddca3e4876b7c9", + "url": "https://huggingface.co/Comfy-Org/Qwen-Image-2.1/resolve/a8342359c61979b06a8dafb1a47991fe1f30e994/vae/qwen_image_2.1_vae_bf16.safetensors", + "repository": "Comfy-Org/Qwen-Image-2.1", + "size_bytes": 675509688 + } + ], + "licenses": [ + "Qwen research license; not Apache-2.0. https://huggingface.co/Qwen/Qwen-Image-2.1-Turbo/blob/fc13ba1986fcb905854dc8de2f52eee395f49b5b/LICENSE" + ], + "findings": [ + "Distinct Turbo checkpoint tested with the exact eight-step ManualSigmas schedule, CFG1 and Euler. It is separate from the earlier non-Turbo25-step Qwen trials.", + "Q4_K_M describes generator storage; the existing INT8 ConvRot encoder was on CPU, with separately staged VAE and FP32-compatible runtime settings. A checkpoint smaller than 6 GB does not establish complete dedicated-VRAM residency.", + "Ten requests over nine briefs completed; first two used the same brief and seeds 42/43. Five variants per brief remain untested." + ], + "workflow_source": "https://github.com/Comfy-Org/workflow_templates/blob/0cf483d1dcb548b8b8d3f38ddf0ff41c89896cdb/templates/image_qwen_image_2_1_turbo_t2i.json", + "workflow_notice": "qwen-turbo-workflow-NOTICE.md" } ], "runs": [ { "id": "qwen-01", - "label": "Qwen Image 2.1 Q4_K_M \u00b7 seed 43", + "label": "Qwen Image 2.1 Q4_K_M · seed 43", "group": "qwen-initial", "kind": "image", "model_id": "qwen", @@ -642,7 +2054,7 @@ }, { "id": "flux3-01", - "label": "FLUX.3 image Klein Q4_K_M \u00b7 seed 42", + "label": "FLUX.3 image Klein Q4_K_M · seed 42", "group": "flux3-initial", "kind": "image", "model_id": "flux3", @@ -734,7 +2146,7 @@ }, { "id": "flux3-02", - "label": "FLUX.3 image Klein Q4_K_M \u00b7 seed 42", + "label": "FLUX.3 image Klein Q4_K_M · seed 42", "group": "flux3-initial", "kind": "image", "model_id": "flux3", @@ -826,7 +2238,7 @@ }, { "id": "qwen-02", - "label": "Qwen Image 2.1 Q4_K_M \u00b7 seed 42", + "label": "Qwen Image 2.1 Q4_K_M · seed 42", "group": "qwen-initial", "kind": "image", "model_id": "qwen", @@ -924,7 +2336,7 @@ }, { "id": "qwen-03", - "label": "Qwen Image 2.1 Q4_K_M \u00b7 seed 42", + "label": "Qwen Image 2.1 Q4_K_M · seed 42", "group": "qwen-initial", "kind": "image", "model_id": "qwen", @@ -1042,7 +2454,7 @@ }, { "id": "qwen-04", - "label": "Qwen Image 2.1 Q4_K_M \u00b7 seed 43", + "label": "Qwen Image 2.1 Q4_K_M · seed 43", "group": "qwen-initial", "kind": "image", "model_id": "qwen", @@ -1166,7 +2578,7 @@ }, { "id": "flux3-03", - "label": "FLUX.3 image Klein Q4_K_M \u00b7 seed 42", + "label": "FLUX.3 image Klein Q4_K_M · seed 42", "group": "flux3-initial", "kind": "image", "model_id": "flux3", @@ -1288,7 +2700,7 @@ }, { "id": "flux3-04", - "label": "FLUX.3 image Klein Q4_K_M \u00b7 seed 43", + "label": "FLUX.3 image Klein Q4_K_M · seed 43", "group": "flux3-initial", "kind": "image", "model_id": "flux3", @@ -1416,7 +2828,7 @@ }, { "id": "flux1-01", - "label": "FLUX.1 dev Q4_0 \u00b7 seed 42", + "label": "FLUX.1 dev Q4_0 · seed 42", "group": "flux1-initial", "kind": "image", "model_id": "flux1", @@ -1544,7 +2956,7 @@ }, { "id": "flux1-02", - "label": "FLUX.1 dev Q4_0 \u00b7 seed 43", + "label": "FLUX.1 dev Q4_0 · seed 43", "group": "flux1-initial", "kind": "image", "model_id": "flux1", @@ -1677,7 +3089,7 @@ }, { "id": "yue2-01", - "label": "YuE2 INT8 ConvRot \u00b7 seed 42", + "label": "YuE2 INT8 ConvRot · seed 42", "group": "yue2-pair", "kind": "audio", "model_id": "yue2", @@ -1784,7 +3196,7 @@ }, { "id": "yue2-02", - "label": "YuE2 INT8 ConvRot \u00b7 seed 43", + "label": "YuE2 INT8 ConvRot · seed 43", "group": "yue2-pair", "kind": "audio", "model_id": "yue2", @@ -1894,7 +3306,7 @@ }, { "id": "wan-01", - "label": "Wan2.1 T2V 1.3B \u00b7 seed 42", + "label": "Wan2.1 T2V 1.3B · seed 42", "group": "wan21-small-v1", "kind": "video", "model_id": "wan", @@ -2004,7 +3416,7 @@ }, { "id": "wan-02", - "label": "Wan2.1 T2V 1.3B \u00b7 seed 42", + "label": "Wan2.1 T2V 1.3B · seed 42", "group": "wan21-q8-whole-stage-v1", "kind": "video", "model_id": "wan", @@ -2147,7 +3559,7 @@ }, { "id": "wan-03", - "label": "Wan2.1 T2V 1.3B \u00b7 seed 43", + "label": "Wan2.1 T2V 1.3B · seed 43", "group": "wan21-q8-whole-stage-v1", "kind": "video", "model_id": "wan", @@ -2272,7 +3684,7 @@ }, { "id": "ltx-01", - "label": "LTX-Video 2B 0.9.8 distilled Q8_0 \u00b7 seed 42", + "label": "LTX-Video 2B 0.9.8 distilled Q8_0 · seed 42", "group": "ltxv2b-q8-text-v3", "kind": "video", "model_id": "ltx", @@ -2427,7 +3839,7 @@ }, { "id": "ltx-02", - "label": "LTX-Video 2B 0.9.8 distilled Q8_0 \u00b7 seed 43", + "label": "LTX-Video 2B 0.9.8 distilled Q8_0 · seed 43", "group": "ltxv2b-q8-text-v3", "kind": "video", "model_id": "ltx", @@ -2583,7 +3995,7 @@ }, { "id": "ltx-03", - "label": "LTX-Video 2B 0.9.8 distilled Q8_0 \u00b7 seed 42", + "label": "LTX-Video 2B 0.9.8 distilled Q8_0 · seed 42", "group": "ltxv2b-q8-image-v1", "kind": "video", "model_id": "ltx", @@ -2759,7 +4171,7 @@ }, { "id": "ltx-04", - "label": "LTX-Video 2B 0.9.8 distilled Q8_0 \u00b7 seed 43", + "label": "LTX-Video 2B 0.9.8 distilled Q8_0 · seed 43", "group": "ltxv2b-q8-image-v1", "kind": "video", "model_id": "ltx", @@ -2922,7 +4334,7 @@ }, { "id": "cog-01", - "label": "CogVideoX 2B local Q8_0 \u00b7 seed 42", + "label": "CogVideoX 2B local Q8_0 · seed 42", "group": "cogvideox2b-q8-subquad-text-v2", "kind": "video", "model_id": "cog", @@ -3064,7 +4476,7 @@ }, { "id": "cog-02", - "label": "CogVideoX 2B local Q8_0 \u00b7 seed 43", + "label": "CogVideoX 2B local Q8_0 · seed 43", "group": "cogvideox2b-q8-subquad-text-v2", "kind": "video", "model_id": "cog", @@ -18301,13 +19713,13 @@ "seed": 42, "status": "succeeded", "accepted_result": true, - "prompt": "A manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body\u2019s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.", + "prompt": "A manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body’s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.", "prompt_details": { "conditioning_texts": { - "4.text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body\u2019s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.", + "4.text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body’s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.", "5.text": "" }, - "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body\u2019s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.", + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body’s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.", "scene_id": "coffee-grinder", "scene_title": "Exploded mechanical coffee grinder", "tier": 4, @@ -18509,13 +19921,13 @@ "seed": 43, "status": "succeeded", "accepted_result": true, - "prompt": "A manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body\u2019s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.", + "prompt": "A manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body’s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.", "prompt_details": { "conditioning_texts": { - "4.text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body\u2019s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.", + "4.text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body’s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.", "5.text": "" }, - "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body\u2019s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.", + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body’s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.", "scene_id": "coffee-grinder", "scene_title": "Exploded mechanical coffee grinder", "tier": 4, @@ -22066,13 +23478,13 @@ "seed": 42, "status": "succeeded", "accepted_result": true, - "prompt": "Design a midnight blue jazz poster titled NIGHT GARDEN, featuring three musicians surrounded by stylized leaves.\n\nUse original geometric illustration with a warm cream title and coral accents. Show a trumpet player, upright bassist, and drummer. Place the title at the top and the performers across the lower half.\n\nThe exact headline is NIGHT GARDEN in large uppercase letters. Under it, set the smaller line LIVE JAZZ. Keep both phrases clearly separated from the instruments and foliage. Use flat screen printed shapes with a slight grain texture, strong negative space, and a limited palette of midnight blue, cream, coral, and muted green. The three musicians are anonymous invented figures with simplified faces and distinct poses.\n\nArrange the trumpet player on the left, facing inward; the bassist in the center, holding an upright bass vertically; and the drummer on the right, seated behind a compact kit. Instruments must remain recognizable even in the geometric style. Leaf shapes curve around the outer edges and stop before the headline. The poster has a narrow cream border and a small centered circle near the bottom containing a simple leaf symbol. Align the typography on a clear central axis, with the main title spanning two lines if necessary. Maintain enough contrast for comfortable reading, and reserve breathing room around every letter rather than filling the entire page with decoration.\n\nTreat this as a finished cultural poster with deliberate hierarchy and controlled printing texture. NIGHT GARDEN should be the largest element, followed by the three musician silhouettes and then LIVE JAZZ. Do not add dates, venues, sponsor marks, or any other words. The headline can use a sturdy condensed serif style with modest variations in letter width, but each letter should have a clean interior shape and remain legible. Give the trumpet a flared bell aimed toward the center and attach it naturally to the musician\u2019s hands and mouth. The upright bass has one continuous neck, a curved body, bridge, and four suggested strings; the bassist\u2019s bow or plucking hand should not cut through the instrument. The drummer holds two separate sticks above a snare and one cymbal, with a small kick drum visible below. Their simplified limbs should remain intentional and connected. Use coral for a few instrument highlights and clothing panels, while cream defines faces, hands, and select leaf edges. Keep green leaves visually subordinate to the performers. Apply grain as a fine variation inside printed shapes, without eroding letterforms or adding random scratches across faces. A few slight registration offsets can suggest traditional printing, but the title should stay crisp. The result should work as a single portrait oriented sheet, balancing lively music energy with a calm nocturnal botanical identity.", + "prompt": "Design a midnight blue jazz poster titled NIGHT GARDEN, featuring three musicians surrounded by stylized leaves.\n\nUse original geometric illustration with a warm cream title and coral accents. Show a trumpet player, upright bassist, and drummer. Place the title at the top and the performers across the lower half.\n\nThe exact headline is NIGHT GARDEN in large uppercase letters. Under it, set the smaller line LIVE JAZZ. Keep both phrases clearly separated from the instruments and foliage. Use flat screen printed shapes with a slight grain texture, strong negative space, and a limited palette of midnight blue, cream, coral, and muted green. The three musicians are anonymous invented figures with simplified faces and distinct poses.\n\nArrange the trumpet player on the left, facing inward; the bassist in the center, holding an upright bass vertically; and the drummer on the right, seated behind a compact kit. Instruments must remain recognizable even in the geometric style. Leaf shapes curve around the outer edges and stop before the headline. The poster has a narrow cream border and a small centered circle near the bottom containing a simple leaf symbol. Align the typography on a clear central axis, with the main title spanning two lines if necessary. Maintain enough contrast for comfortable reading, and reserve breathing room around every letter rather than filling the entire page with decoration.\n\nTreat this as a finished cultural poster with deliberate hierarchy and controlled printing texture. NIGHT GARDEN should be the largest element, followed by the three musician silhouettes and then LIVE JAZZ. Do not add dates, venues, sponsor marks, or any other words. The headline can use a sturdy condensed serif style with modest variations in letter width, but each letter should have a clean interior shape and remain legible. Give the trumpet a flared bell aimed toward the center and attach it naturally to the musician’s hands and mouth. The upright bass has one continuous neck, a curved body, bridge, and four suggested strings; the bassist’s bow or plucking hand should not cut through the instrument. The drummer holds two separate sticks above a snare and one cymbal, with a small kick drum visible below. Their simplified limbs should remain intentional and connected. Use coral for a few instrument highlights and clothing panels, while cream defines faces, hands, and select leaf edges. Keep green leaves visually subordinate to the performers. Apply grain as a fine variation inside printed shapes, without eroding letterforms or adding random scratches across faces. A few slight registration offsets can suggest traditional printing, but the title should stay crisp. The result should work as a single portrait oriented sheet, balancing lively music energy with a calm nocturnal botanical identity.", "prompt_details": { "conditioning_texts": { - "4.text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nDesign a midnight blue jazz poster titled NIGHT GARDEN, featuring three musicians surrounded by stylized leaves.\n\nUse original geometric illustration with a warm cream title and coral accents. Show a trumpet player, upright bassist, and drummer. Place the title at the top and the performers across the lower half.\n\nThe exact headline is NIGHT GARDEN in large uppercase letters. Under it, set the smaller line LIVE JAZZ. Keep both phrases clearly separated from the instruments and foliage. Use flat screen printed shapes with a slight grain texture, strong negative space, and a limited palette of midnight blue, cream, coral, and muted green. The three musicians are anonymous invented figures with simplified faces and distinct poses.\n\nArrange the trumpet player on the left, facing inward; the bassist in the center, holding an upright bass vertically; and the drummer on the right, seated behind a compact kit. Instruments must remain recognizable even in the geometric style. Leaf shapes curve around the outer edges and stop before the headline. The poster has a narrow cream border and a small centered circle near the bottom containing a simple leaf symbol. Align the typography on a clear central axis, with the main title spanning two lines if necessary. Maintain enough contrast for comfortable reading, and reserve breathing room around every letter rather than filling the entire page with decoration.\n\nTreat this as a finished cultural poster with deliberate hierarchy and controlled printing texture. NIGHT GARDEN should be the largest element, followed by the three musician silhouettes and then LIVE JAZZ. Do not add dates, venues, sponsor marks, or any other words. The headline can use a sturdy condensed serif style with modest variations in letter width, but each letter should have a clean interior shape and remain legible. Give the trumpet a flared bell aimed toward the center and attach it naturally to the musician\u2019s hands and mouth. The upright bass has one continuous neck, a curved body, bridge, and four suggested strings; the bassist\u2019s bow or plucking hand should not cut through the instrument. The drummer holds two separate sticks above a snare and one cymbal, with a small kick drum visible below. Their simplified limbs should remain intentional and connected. Use coral for a few instrument highlights and clothing panels, while cream defines faces, hands, and select leaf edges. Keep green leaves visually subordinate to the performers. Apply grain as a fine variation inside printed shapes, without eroding letterforms or adding random scratches across faces. A few slight registration offsets can suggest traditional printing, but the title should stay crisp. The result should work as a single portrait oriented sheet, balancing lively music energy with a calm nocturnal botanical identity.", + "4.text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nDesign a midnight blue jazz poster titled NIGHT GARDEN, featuring three musicians surrounded by stylized leaves.\n\nUse original geometric illustration with a warm cream title and coral accents. Show a trumpet player, upright bassist, and drummer. Place the title at the top and the performers across the lower half.\n\nThe exact headline is NIGHT GARDEN in large uppercase letters. Under it, set the smaller line LIVE JAZZ. Keep both phrases clearly separated from the instruments and foliage. Use flat screen printed shapes with a slight grain texture, strong negative space, and a limited palette of midnight blue, cream, coral, and muted green. The three musicians are anonymous invented figures with simplified faces and distinct poses.\n\nArrange the trumpet player on the left, facing inward; the bassist in the center, holding an upright bass vertically; and the drummer on the right, seated behind a compact kit. Instruments must remain recognizable even in the geometric style. Leaf shapes curve around the outer edges and stop before the headline. The poster has a narrow cream border and a small centered circle near the bottom containing a simple leaf symbol. Align the typography on a clear central axis, with the main title spanning two lines if necessary. Maintain enough contrast for comfortable reading, and reserve breathing room around every letter rather than filling the entire page with decoration.\n\nTreat this as a finished cultural poster with deliberate hierarchy and controlled printing texture. NIGHT GARDEN should be the largest element, followed by the three musician silhouettes and then LIVE JAZZ. Do not add dates, venues, sponsor marks, or any other words. The headline can use a sturdy condensed serif style with modest variations in letter width, but each letter should have a clean interior shape and remain legible. Give the trumpet a flared bell aimed toward the center and attach it naturally to the musician’s hands and mouth. The upright bass has one continuous neck, a curved body, bridge, and four suggested strings; the bassist’s bow or plucking hand should not cut through the instrument. The drummer holds two separate sticks above a snare and one cymbal, with a small kick drum visible below. Their simplified limbs should remain intentional and connected. Use coral for a few instrument highlights and clothing panels, while cream defines faces, hands, and select leaf edges. Keep green leaves visually subordinate to the performers. Apply grain as a fine variation inside printed shapes, without eroding letterforms or adding random scratches across faces. A few slight registration offsets can suggest traditional printing, but the title should stay crisp. The result should work as a single portrait oriented sheet, balancing lively music energy with a calm nocturnal botanical identity.", "5.text": "" }, - "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nDesign a midnight blue jazz poster titled NIGHT GARDEN, featuring three musicians surrounded by stylized leaves.\n\nUse original geometric illustration with a warm cream title and coral accents. Show a trumpet player, upright bassist, and drummer. Place the title at the top and the performers across the lower half.\n\nThe exact headline is NIGHT GARDEN in large uppercase letters. Under it, set the smaller line LIVE JAZZ. Keep both phrases clearly separated from the instruments and foliage. Use flat screen printed shapes with a slight grain texture, strong negative space, and a limited palette of midnight blue, cream, coral, and muted green. The three musicians are anonymous invented figures with simplified faces and distinct poses.\n\nArrange the trumpet player on the left, facing inward; the bassist in the center, holding an upright bass vertically; and the drummer on the right, seated behind a compact kit. Instruments must remain recognizable even in the geometric style. Leaf shapes curve around the outer edges and stop before the headline. The poster has a narrow cream border and a small centered circle near the bottom containing a simple leaf symbol. Align the typography on a clear central axis, with the main title spanning two lines if necessary. Maintain enough contrast for comfortable reading, and reserve breathing room around every letter rather than filling the entire page with decoration.\n\nTreat this as a finished cultural poster with deliberate hierarchy and controlled printing texture. NIGHT GARDEN should be the largest element, followed by the three musician silhouettes and then LIVE JAZZ. Do not add dates, venues, sponsor marks, or any other words. The headline can use a sturdy condensed serif style with modest variations in letter width, but each letter should have a clean interior shape and remain legible. Give the trumpet a flared bell aimed toward the center and attach it naturally to the musician\u2019s hands and mouth. The upright bass has one continuous neck, a curved body, bridge, and four suggested strings; the bassist\u2019s bow or plucking hand should not cut through the instrument. The drummer holds two separate sticks above a snare and one cymbal, with a small kick drum visible below. Their simplified limbs should remain intentional and connected. Use coral for a few instrument highlights and clothing panels, while cream defines faces, hands, and select leaf edges. Keep green leaves visually subordinate to the performers. Apply grain as a fine variation inside printed shapes, without eroding letterforms or adding random scratches across faces. A few slight registration offsets can suggest traditional printing, but the title should stay crisp. The result should work as a single portrait oriented sheet, balancing lively music energy with a calm nocturnal botanical identity.", + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nDesign a midnight blue jazz poster titled NIGHT GARDEN, featuring three musicians surrounded by stylized leaves.\n\nUse original geometric illustration with a warm cream title and coral accents. Show a trumpet player, upright bassist, and drummer. Place the title at the top and the performers across the lower half.\n\nThe exact headline is NIGHT GARDEN in large uppercase letters. Under it, set the smaller line LIVE JAZZ. Keep both phrases clearly separated from the instruments and foliage. Use flat screen printed shapes with a slight grain texture, strong negative space, and a limited palette of midnight blue, cream, coral, and muted green. The three musicians are anonymous invented figures with simplified faces and distinct poses.\n\nArrange the trumpet player on the left, facing inward; the bassist in the center, holding an upright bass vertically; and the drummer on the right, seated behind a compact kit. Instruments must remain recognizable even in the geometric style. Leaf shapes curve around the outer edges and stop before the headline. The poster has a narrow cream border and a small centered circle near the bottom containing a simple leaf symbol. Align the typography on a clear central axis, with the main title spanning two lines if necessary. Maintain enough contrast for comfortable reading, and reserve breathing room around every letter rather than filling the entire page with decoration.\n\nTreat this as a finished cultural poster with deliberate hierarchy and controlled printing texture. NIGHT GARDEN should be the largest element, followed by the three musician silhouettes and then LIVE JAZZ. Do not add dates, venues, sponsor marks, or any other words. The headline can use a sturdy condensed serif style with modest variations in letter width, but each letter should have a clean interior shape and remain legible. Give the trumpet a flared bell aimed toward the center and attach it naturally to the musician’s hands and mouth. The upright bass has one continuous neck, a curved body, bridge, and four suggested strings; the bassist’s bow or plucking hand should not cut through the instrument. The drummer holds two separate sticks above a snare and one cymbal, with a small kick drum visible below. Their simplified limbs should remain intentional and connected. Use coral for a few instrument highlights and clothing panels, while cream defines faces, hands, and select leaf edges. Keep green leaves visually subordinate to the performers. Apply grain as a fine variation inside printed shapes, without eroding letterforms or adding random scratches across faces. A few slight registration offsets can suggest traditional printing, but the title should stay crisp. The result should work as a single portrait oriented sheet, balancing lively music energy with a calm nocturnal botanical identity.", "scene_id": "night-garden-poster", "scene_title": "Night Garden jazz poster", "tier": 5, @@ -22277,13 +23689,13 @@ "seed": 43, "status": "succeeded", "accepted_result": true, - "prompt": "Design a midnight blue jazz poster titled NIGHT GARDEN, featuring three musicians surrounded by stylized leaves.\n\nUse original geometric illustration with a warm cream title and coral accents. Show a trumpet player, upright bassist, and drummer. Place the title at the top and the performers across the lower half.\n\nThe exact headline is NIGHT GARDEN in large uppercase letters. Under it, set the smaller line LIVE JAZZ. Keep both phrases clearly separated from the instruments and foliage. Use flat screen printed shapes with a slight grain texture, strong negative space, and a limited palette of midnight blue, cream, coral, and muted green. The three musicians are anonymous invented figures with simplified faces and distinct poses.\n\nArrange the trumpet player on the left, facing inward; the bassist in the center, holding an upright bass vertically; and the drummer on the right, seated behind a compact kit. Instruments must remain recognizable even in the geometric style. Leaf shapes curve around the outer edges and stop before the headline. The poster has a narrow cream border and a small centered circle near the bottom containing a simple leaf symbol. Align the typography on a clear central axis, with the main title spanning two lines if necessary. Maintain enough contrast for comfortable reading, and reserve breathing room around every letter rather than filling the entire page with decoration.\n\nTreat this as a finished cultural poster with deliberate hierarchy and controlled printing texture. NIGHT GARDEN should be the largest element, followed by the three musician silhouettes and then LIVE JAZZ. Do not add dates, venues, sponsor marks, or any other words. The headline can use a sturdy condensed serif style with modest variations in letter width, but each letter should have a clean interior shape and remain legible. Give the trumpet a flared bell aimed toward the center and attach it naturally to the musician\u2019s hands and mouth. The upright bass has one continuous neck, a curved body, bridge, and four suggested strings; the bassist\u2019s bow or plucking hand should not cut through the instrument. The drummer holds two separate sticks above a snare and one cymbal, with a small kick drum visible below. Their simplified limbs should remain intentional and connected. Use coral for a few instrument highlights and clothing panels, while cream defines faces, hands, and select leaf edges. Keep green leaves visually subordinate to the performers. Apply grain as a fine variation inside printed shapes, without eroding letterforms or adding random scratches across faces. A few slight registration offsets can suggest traditional printing, but the title should stay crisp. The result should work as a single portrait oriented sheet, balancing lively music energy with a calm nocturnal botanical identity.", + "prompt": "Design a midnight blue jazz poster titled NIGHT GARDEN, featuring three musicians surrounded by stylized leaves.\n\nUse original geometric illustration with a warm cream title and coral accents. Show a trumpet player, upright bassist, and drummer. Place the title at the top and the performers across the lower half.\n\nThe exact headline is NIGHT GARDEN in large uppercase letters. Under it, set the smaller line LIVE JAZZ. Keep both phrases clearly separated from the instruments and foliage. Use flat screen printed shapes with a slight grain texture, strong negative space, and a limited palette of midnight blue, cream, coral, and muted green. The three musicians are anonymous invented figures with simplified faces and distinct poses.\n\nArrange the trumpet player on the left, facing inward; the bassist in the center, holding an upright bass vertically; and the drummer on the right, seated behind a compact kit. Instruments must remain recognizable even in the geometric style. Leaf shapes curve around the outer edges and stop before the headline. The poster has a narrow cream border and a small centered circle near the bottom containing a simple leaf symbol. Align the typography on a clear central axis, with the main title spanning two lines if necessary. Maintain enough contrast for comfortable reading, and reserve breathing room around every letter rather than filling the entire page with decoration.\n\nTreat this as a finished cultural poster with deliberate hierarchy and controlled printing texture. NIGHT GARDEN should be the largest element, followed by the three musician silhouettes and then LIVE JAZZ. Do not add dates, venues, sponsor marks, or any other words. The headline can use a sturdy condensed serif style with modest variations in letter width, but each letter should have a clean interior shape and remain legible. Give the trumpet a flared bell aimed toward the center and attach it naturally to the musician’s hands and mouth. The upright bass has one continuous neck, a curved body, bridge, and four suggested strings; the bassist’s bow or plucking hand should not cut through the instrument. The drummer holds two separate sticks above a snare and one cymbal, with a small kick drum visible below. Their simplified limbs should remain intentional and connected. Use coral for a few instrument highlights and clothing panels, while cream defines faces, hands, and select leaf edges. Keep green leaves visually subordinate to the performers. Apply grain as a fine variation inside printed shapes, without eroding letterforms or adding random scratches across faces. A few slight registration offsets can suggest traditional printing, but the title should stay crisp. The result should work as a single portrait oriented sheet, balancing lively music energy with a calm nocturnal botanical identity.", "prompt_details": { "conditioning_texts": { - "4.text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nDesign a midnight blue jazz poster titled NIGHT GARDEN, featuring three musicians surrounded by stylized leaves.\n\nUse original geometric illustration with a warm cream title and coral accents. Show a trumpet player, upright bassist, and drummer. Place the title at the top and the performers across the lower half.\n\nThe exact headline is NIGHT GARDEN in large uppercase letters. Under it, set the smaller line LIVE JAZZ. Keep both phrases clearly separated from the instruments and foliage. Use flat screen printed shapes with a slight grain texture, strong negative space, and a limited palette of midnight blue, cream, coral, and muted green. The three musicians are anonymous invented figures with simplified faces and distinct poses.\n\nArrange the trumpet player on the left, facing inward; the bassist in the center, holding an upright bass vertically; and the drummer on the right, seated behind a compact kit. Instruments must remain recognizable even in the geometric style. Leaf shapes curve around the outer edges and stop before the headline. The poster has a narrow cream border and a small centered circle near the bottom containing a simple leaf symbol. Align the typography on a clear central axis, with the main title spanning two lines if necessary. Maintain enough contrast for comfortable reading, and reserve breathing room around every letter rather than filling the entire page with decoration.\n\nTreat this as a finished cultural poster with deliberate hierarchy and controlled printing texture. NIGHT GARDEN should be the largest element, followed by the three musician silhouettes and then LIVE JAZZ. Do not add dates, venues, sponsor marks, or any other words. The headline can use a sturdy condensed serif style with modest variations in letter width, but each letter should have a clean interior shape and remain legible. Give the trumpet a flared bell aimed toward the center and attach it naturally to the musician\u2019s hands and mouth. The upright bass has one continuous neck, a curved body, bridge, and four suggested strings; the bassist\u2019s bow or plucking hand should not cut through the instrument. The drummer holds two separate sticks above a snare and one cymbal, with a small kick drum visible below. Their simplified limbs should remain intentional and connected. Use coral for a few instrument highlights and clothing panels, while cream defines faces, hands, and select leaf edges. Keep green leaves visually subordinate to the performers. Apply grain as a fine variation inside printed shapes, without eroding letterforms or adding random scratches across faces. A few slight registration offsets can suggest traditional printing, but the title should stay crisp. The result should work as a single portrait oriented sheet, balancing lively music energy with a calm nocturnal botanical identity.", + "4.text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nDesign a midnight blue jazz poster titled NIGHT GARDEN, featuring three musicians surrounded by stylized leaves.\n\nUse original geometric illustration with a warm cream title and coral accents. Show a trumpet player, upright bassist, and drummer. Place the title at the top and the performers across the lower half.\n\nThe exact headline is NIGHT GARDEN in large uppercase letters. Under it, set the smaller line LIVE JAZZ. Keep both phrases clearly separated from the instruments and foliage. Use flat screen printed shapes with a slight grain texture, strong negative space, and a limited palette of midnight blue, cream, coral, and muted green. The three musicians are anonymous invented figures with simplified faces and distinct poses.\n\nArrange the trumpet player on the left, facing inward; the bassist in the center, holding an upright bass vertically; and the drummer on the right, seated behind a compact kit. Instruments must remain recognizable even in the geometric style. Leaf shapes curve around the outer edges and stop before the headline. The poster has a narrow cream border and a small centered circle near the bottom containing a simple leaf symbol. Align the typography on a clear central axis, with the main title spanning two lines if necessary. Maintain enough contrast for comfortable reading, and reserve breathing room around every letter rather than filling the entire page with decoration.\n\nTreat this as a finished cultural poster with deliberate hierarchy and controlled printing texture. NIGHT GARDEN should be the largest element, followed by the three musician silhouettes and then LIVE JAZZ. Do not add dates, venues, sponsor marks, or any other words. The headline can use a sturdy condensed serif style with modest variations in letter width, but each letter should have a clean interior shape and remain legible. Give the trumpet a flared bell aimed toward the center and attach it naturally to the musician’s hands and mouth. The upright bass has one continuous neck, a curved body, bridge, and four suggested strings; the bassist’s bow or plucking hand should not cut through the instrument. The drummer holds two separate sticks above a snare and one cymbal, with a small kick drum visible below. Their simplified limbs should remain intentional and connected. Use coral for a few instrument highlights and clothing panels, while cream defines faces, hands, and select leaf edges. Keep green leaves visually subordinate to the performers. Apply grain as a fine variation inside printed shapes, without eroding letterforms or adding random scratches across faces. A few slight registration offsets can suggest traditional printing, but the title should stay crisp. The result should work as a single portrait oriented sheet, balancing lively music energy with a calm nocturnal botanical identity.", "5.text": "" }, - "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nDesign a midnight blue jazz poster titled NIGHT GARDEN, featuring three musicians surrounded by stylized leaves.\n\nUse original geometric illustration with a warm cream title and coral accents. Show a trumpet player, upright bassist, and drummer. Place the title at the top and the performers across the lower half.\n\nThe exact headline is NIGHT GARDEN in large uppercase letters. Under it, set the smaller line LIVE JAZZ. Keep both phrases clearly separated from the instruments and foliage. Use flat screen printed shapes with a slight grain texture, strong negative space, and a limited palette of midnight blue, cream, coral, and muted green. The three musicians are anonymous invented figures with simplified faces and distinct poses.\n\nArrange the trumpet player on the left, facing inward; the bassist in the center, holding an upright bass vertically; and the drummer on the right, seated behind a compact kit. Instruments must remain recognizable even in the geometric style. Leaf shapes curve around the outer edges and stop before the headline. The poster has a narrow cream border and a small centered circle near the bottom containing a simple leaf symbol. Align the typography on a clear central axis, with the main title spanning two lines if necessary. Maintain enough contrast for comfortable reading, and reserve breathing room around every letter rather than filling the entire page with decoration.\n\nTreat this as a finished cultural poster with deliberate hierarchy and controlled printing texture. NIGHT GARDEN should be the largest element, followed by the three musician silhouettes and then LIVE JAZZ. Do not add dates, venues, sponsor marks, or any other words. The headline can use a sturdy condensed serif style with modest variations in letter width, but each letter should have a clean interior shape and remain legible. Give the trumpet a flared bell aimed toward the center and attach it naturally to the musician\u2019s hands and mouth. The upright bass has one continuous neck, a curved body, bridge, and four suggested strings; the bassist\u2019s bow or plucking hand should not cut through the instrument. The drummer holds two separate sticks above a snare and one cymbal, with a small kick drum visible below. Their simplified limbs should remain intentional and connected. Use coral for a few instrument highlights and clothing panels, while cream defines faces, hands, and select leaf edges. Keep green leaves visually subordinate to the performers. Apply grain as a fine variation inside printed shapes, without eroding letterforms or adding random scratches across faces. A few slight registration offsets can suggest traditional printing, but the title should stay crisp. The result should work as a single portrait oriented sheet, balancing lively music energy with a calm nocturnal botanical identity.", + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nDesign a midnight blue jazz poster titled NIGHT GARDEN, featuring three musicians surrounded by stylized leaves.\n\nUse original geometric illustration with a warm cream title and coral accents. Show a trumpet player, upright bassist, and drummer. Place the title at the top and the performers across the lower half.\n\nThe exact headline is NIGHT GARDEN in large uppercase letters. Under it, set the smaller line LIVE JAZZ. Keep both phrases clearly separated from the instruments and foliage. Use flat screen printed shapes with a slight grain texture, strong negative space, and a limited palette of midnight blue, cream, coral, and muted green. The three musicians are anonymous invented figures with simplified faces and distinct poses.\n\nArrange the trumpet player on the left, facing inward; the bassist in the center, holding an upright bass vertically; and the drummer on the right, seated behind a compact kit. Instruments must remain recognizable even in the geometric style. Leaf shapes curve around the outer edges and stop before the headline. The poster has a narrow cream border and a small centered circle near the bottom containing a simple leaf symbol. Align the typography on a clear central axis, with the main title spanning two lines if necessary. Maintain enough contrast for comfortable reading, and reserve breathing room around every letter rather than filling the entire page with decoration.\n\nTreat this as a finished cultural poster with deliberate hierarchy and controlled printing texture. NIGHT GARDEN should be the largest element, followed by the three musician silhouettes and then LIVE JAZZ. Do not add dates, venues, sponsor marks, or any other words. The headline can use a sturdy condensed serif style with modest variations in letter width, but each letter should have a clean interior shape and remain legible. Give the trumpet a flared bell aimed toward the center and attach it naturally to the musician’s hands and mouth. The upright bass has one continuous neck, a curved body, bridge, and four suggested strings; the bassist’s bow or plucking hand should not cut through the instrument. The drummer holds two separate sticks above a snare and one cymbal, with a small kick drum visible below. Their simplified limbs should remain intentional and connected. Use coral for a few instrument highlights and clothing panels, while cream defines faces, hands, and select leaf edges. Keep green leaves visually subordinate to the performers. Apply grain as a fine variation inside printed shapes, without eroding letterforms or adding random scratches across faces. A few slight registration offsets can suggest traditional printing, but the title should stay crisp. The result should work as a single portrait oriented sheet, balancing lively music energy with a calm nocturnal botanical identity.", "scene_id": "night-garden-poster", "scene_title": "Night Garden jazz poster", "tier": 5, @@ -22487,13 +23899,13 @@ "seed": 42, "status": "succeeded", "accepted_result": true, - "prompt": "A manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body\u2019s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.\n\nMake the image explain a plausible manual mechanism through geometry and material detail. The crank is a short bent steel arm attached to a comfortable rounded wooden knob, with a visible mounting hole at its inner end. The top cap has a central bearing opening and a smooth rim that fits the body. A slender drive shaft descends from that bearing into the inner burr, while the outer burr forms a matching toothed ring. Suggest cutting teeth with crisp repeated grooves rather than decorative random spikes. Keep the inner and outer burrs proportionate so one can nest inside the other during assembly. The body is a brushed steel cylinder with a narrow rubber grip band, a clean top opening, and a threaded lower collar. The jar is a simple clear vessel with subtle refraction, its circular mouth becoming an ellipse in perspective. Show a few coffee beans beside the assembled axis near the surface as a scale reference, but do not scatter grounds across the floating components. Preserve consistent perspective and the same light direction on every piece. The background should remain neutral and unbranded, allowing fine reflections and edge separations to carry the image. Avoid arrows, diagram callouts, extra screws without a visible role, or impossible parts passing through each other. The final composition should read like a carefully art directed product explanation suitable for understanding how a small durable object is put together.", + "prompt": "A manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body’s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.\n\nMake the image explain a plausible manual mechanism through geometry and material detail. The crank is a short bent steel arm attached to a comfortable rounded wooden knob, with a visible mounting hole at its inner end. The top cap has a central bearing opening and a smooth rim that fits the body. A slender drive shaft descends from that bearing into the inner burr, while the outer burr forms a matching toothed ring. Suggest cutting teeth with crisp repeated grooves rather than decorative random spikes. Keep the inner and outer burrs proportionate so one can nest inside the other during assembly. The body is a brushed steel cylinder with a narrow rubber grip band, a clean top opening, and a threaded lower collar. The jar is a simple clear vessel with subtle refraction, its circular mouth becoming an ellipse in perspective. Show a few coffee beans beside the assembled axis near the surface as a scale reference, but do not scatter grounds across the floating components. Preserve consistent perspective and the same light direction on every piece. The background should remain neutral and unbranded, allowing fine reflections and edge separations to carry the image. Avoid arrows, diagram callouts, extra screws without a visible role, or impossible parts passing through each other. The final composition should read like a carefully art directed product explanation suitable for understanding how a small durable object is put together.", "prompt_details": { "conditioning_texts": { - "4.text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body\u2019s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.\n\nMake the image explain a plausible manual mechanism through geometry and material detail. The crank is a short bent steel arm attached to a comfortable rounded wooden knob, with a visible mounting hole at its inner end. The top cap has a central bearing opening and a smooth rim that fits the body. A slender drive shaft descends from that bearing into the inner burr, while the outer burr forms a matching toothed ring. Suggest cutting teeth with crisp repeated grooves rather than decorative random spikes. Keep the inner and outer burrs proportionate so one can nest inside the other during assembly. The body is a brushed steel cylinder with a narrow rubber grip band, a clean top opening, and a threaded lower collar. The jar is a simple clear vessel with subtle refraction, its circular mouth becoming an ellipse in perspective. Show a few coffee beans beside the assembled axis near the surface as a scale reference, but do not scatter grounds across the floating components. Preserve consistent perspective and the same light direction on every piece. The background should remain neutral and unbranded, allowing fine reflections and edge separations to carry the image. Avoid arrows, diagram callouts, extra screws without a visible role, or impossible parts passing through each other. The final composition should read like a carefully art directed product explanation suitable for understanding how a small durable object is put together.", + "4.text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body’s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.\n\nMake the image explain a plausible manual mechanism through geometry and material detail. The crank is a short bent steel arm attached to a comfortable rounded wooden knob, with a visible mounting hole at its inner end. The top cap has a central bearing opening and a smooth rim that fits the body. A slender drive shaft descends from that bearing into the inner burr, while the outer burr forms a matching toothed ring. Suggest cutting teeth with crisp repeated grooves rather than decorative random spikes. Keep the inner and outer burrs proportionate so one can nest inside the other during assembly. The body is a brushed steel cylinder with a narrow rubber grip band, a clean top opening, and a threaded lower collar. The jar is a simple clear vessel with subtle refraction, its circular mouth becoming an ellipse in perspective. Show a few coffee beans beside the assembled axis near the surface as a scale reference, but do not scatter grounds across the floating components. Preserve consistent perspective and the same light direction on every piece. The background should remain neutral and unbranded, allowing fine reflections and edge separations to carry the image. Avoid arrows, diagram callouts, extra screws without a visible role, or impossible parts passing through each other. The final composition should read like a carefully art directed product explanation suitable for understanding how a small durable object is put together.", "5.text": "" }, - "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body\u2019s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.\n\nMake the image explain a plausible manual mechanism through geometry and material detail. The crank is a short bent steel arm attached to a comfortable rounded wooden knob, with a visible mounting hole at its inner end. The top cap has a central bearing opening and a smooth rim that fits the body. A slender drive shaft descends from that bearing into the inner burr, while the outer burr forms a matching toothed ring. Suggest cutting teeth with crisp repeated grooves rather than decorative random spikes. Keep the inner and outer burrs proportionate so one can nest inside the other during assembly. The body is a brushed steel cylinder with a narrow rubber grip band, a clean top opening, and a threaded lower collar. The jar is a simple clear vessel with subtle refraction, its circular mouth becoming an ellipse in perspective. Show a few coffee beans beside the assembled axis near the surface as a scale reference, but do not scatter grounds across the floating components. Preserve consistent perspective and the same light direction on every piece. The background should remain neutral and unbranded, allowing fine reflections and edge separations to carry the image. Avoid arrows, diagram callouts, extra screws without a visible role, or impossible parts passing through each other. The final composition should read like a carefully art directed product explanation suitable for understanding how a small durable object is put together.", + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body’s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.\n\nMake the image explain a plausible manual mechanism through geometry and material detail. The crank is a short bent steel arm attached to a comfortable rounded wooden knob, with a visible mounting hole at its inner end. The top cap has a central bearing opening and a smooth rim that fits the body. A slender drive shaft descends from that bearing into the inner burr, while the outer burr forms a matching toothed ring. Suggest cutting teeth with crisp repeated grooves rather than decorative random spikes. Keep the inner and outer burrs proportionate so one can nest inside the other during assembly. The body is a brushed steel cylinder with a narrow rubber grip band, a clean top opening, and a threaded lower collar. The jar is a simple clear vessel with subtle refraction, its circular mouth becoming an ellipse in perspective. Show a few coffee beans beside the assembled axis near the surface as a scale reference, but do not scatter grounds across the floating components. Preserve consistent perspective and the same light direction on every piece. The background should remain neutral and unbranded, allowing fine reflections and edge separations to carry the image. Avoid arrows, diagram callouts, extra screws without a visible role, or impossible parts passing through each other. The final composition should read like a carefully art directed product explanation suitable for understanding how a small durable object is put together.", "scene_id": "coffee-grinder", "scene_title": "Exploded mechanical coffee grinder", "tier": 5, @@ -22698,13 +24110,13 @@ "seed": 43, "status": "succeeded", "accepted_result": true, - "prompt": "A manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body\u2019s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.\n\nMake the image explain a plausible manual mechanism through geometry and material detail. The crank is a short bent steel arm attached to a comfortable rounded wooden knob, with a visible mounting hole at its inner end. The top cap has a central bearing opening and a smooth rim that fits the body. A slender drive shaft descends from that bearing into the inner burr, while the outer burr forms a matching toothed ring. Suggest cutting teeth with crisp repeated grooves rather than decorative random spikes. Keep the inner and outer burrs proportionate so one can nest inside the other during assembly. The body is a brushed steel cylinder with a narrow rubber grip band, a clean top opening, and a threaded lower collar. The jar is a simple clear vessel with subtle refraction, its circular mouth becoming an ellipse in perspective. Show a few coffee beans beside the assembled axis near the surface as a scale reference, but do not scatter grounds across the floating components. Preserve consistent perspective and the same light direction on every piece. The background should remain neutral and unbranded, allowing fine reflections and edge separations to carry the image. Avoid arrows, diagram callouts, extra screws without a visible role, or impossible parts passing through each other. The final composition should read like a carefully art directed product explanation suitable for understanding how a small durable object is put together.", + "prompt": "A manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body’s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.\n\nMake the image explain a plausible manual mechanism through geometry and material detail. The crank is a short bent steel arm attached to a comfortable rounded wooden knob, with a visible mounting hole at its inner end. The top cap has a central bearing opening and a smooth rim that fits the body. A slender drive shaft descends from that bearing into the inner burr, while the outer burr forms a matching toothed ring. Suggest cutting teeth with crisp repeated grooves rather than decorative random spikes. Keep the inner and outer burrs proportionate so one can nest inside the other during assembly. The body is a brushed steel cylinder with a narrow rubber grip band, a clean top opening, and a threaded lower collar. The jar is a simple clear vessel with subtle refraction, its circular mouth becoming an ellipse in perspective. Show a few coffee beans beside the assembled axis near the surface as a scale reference, but do not scatter grounds across the floating components. Preserve consistent perspective and the same light direction on every piece. The background should remain neutral and unbranded, allowing fine reflections and edge separations to carry the image. Avoid arrows, diagram callouts, extra screws without a visible role, or impossible parts passing through each other. The final composition should read like a carefully art directed product explanation suitable for understanding how a small durable object is put together.", "prompt_details": { "conditioning_texts": { - "4.text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body\u2019s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.\n\nMake the image explain a plausible manual mechanism through geometry and material detail. The crank is a short bent steel arm attached to a comfortable rounded wooden knob, with a visible mounting hole at its inner end. The top cap has a central bearing opening and a smooth rim that fits the body. A slender drive shaft descends from that bearing into the inner burr, while the outer burr forms a matching toothed ring. Suggest cutting teeth with crisp repeated grooves rather than decorative random spikes. Keep the inner and outer burrs proportionate so one can nest inside the other during assembly. The body is a brushed steel cylinder with a narrow rubber grip band, a clean top opening, and a threaded lower collar. The jar is a simple clear vessel with subtle refraction, its circular mouth becoming an ellipse in perspective. Show a few coffee beans beside the assembled axis near the surface as a scale reference, but do not scatter grounds across the floating components. Preserve consistent perspective and the same light direction on every piece. The background should remain neutral and unbranded, allowing fine reflections and edge separations to carry the image. Avoid arrows, diagram callouts, extra screws without a visible role, or impossible parts passing through each other. The final composition should read like a carefully art directed product explanation suitable for understanding how a small durable object is put together.", + "4.text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body’s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.\n\nMake the image explain a plausible manual mechanism through geometry and material detail. The crank is a short bent steel arm attached to a comfortable rounded wooden knob, with a visible mounting hole at its inner end. The top cap has a central bearing opening and a smooth rim that fits the body. A slender drive shaft descends from that bearing into the inner burr, while the outer burr forms a matching toothed ring. Suggest cutting teeth with crisp repeated grooves rather than decorative random spikes. Keep the inner and outer burrs proportionate so one can nest inside the other during assembly. The body is a brushed steel cylinder with a narrow rubber grip band, a clean top opening, and a threaded lower collar. The jar is a simple clear vessel with subtle refraction, its circular mouth becoming an ellipse in perspective. Show a few coffee beans beside the assembled axis near the surface as a scale reference, but do not scatter grounds across the floating components. Preserve consistent perspective and the same light direction on every piece. The background should remain neutral and unbranded, allowing fine reflections and edge separations to carry the image. Avoid arrows, diagram callouts, extra screws without a visible role, or impossible parts passing through each other. The final composition should read like a carefully art directed product explanation suitable for understanding how a small durable object is put together.", "5.text": "" }, - "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body\u2019s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.\n\nMake the image explain a plausible manual mechanism through geometry and material detail. The crank is a short bent steel arm attached to a comfortable rounded wooden knob, with a visible mounting hole at its inner end. The top cap has a central bearing opening and a smooth rim that fits the body. A slender drive shaft descends from that bearing into the inner burr, while the outer burr forms a matching toothed ring. Suggest cutting teeth with crisp repeated grooves rather than decorative random spikes. Keep the inner and outer burrs proportionate so one can nest inside the other during assembly. The body is a brushed steel cylinder with a narrow rubber grip band, a clean top opening, and a threaded lower collar. The jar is a simple clear vessel with subtle refraction, its circular mouth becoming an ellipse in perspective. Show a few coffee beans beside the assembled axis near the surface as a scale reference, but do not scatter grounds across the floating components. Preserve consistent perspective and the same light direction on every piece. The background should remain neutral and unbranded, allowing fine reflections and edge separations to carry the image. Avoid arrows, diagram callouts, extra screws without a visible role, or impossible parts passing through each other. The final composition should read like a carefully art directed product explanation suitable for understanding how a small durable object is put together.", + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA manual coffee grinder appears in an exploded product view, with wooden handle, steel burrs, and glass jar.\n\nArrange the parts along one vertical assembly axis above a pale gray studio surface. Use warm walnut, brushed steel, clear glass, and a small amount of matte black rubber.\n\nThe crank and wooden knob float above the top cap, followed by the central shaft, paired burr pieces, main body, and collection jar. Every part remains separate and aligned so the assembly order is understandable. Give metal edges narrow highlights, walnut a fine visible grain, and the jar a thick transparent base. Add a soft contact shadow beneath the lowest component, with no labels or dimension text.\n\nUse a slightly elevated three quarter viewpoint so the tops and sides of the parts can be inspected. Keep the overall product compact, with clear gaps between successive pieces and no overlapping silhouettes that hide the burr shapes. The shaft passes conceptually through the center of the cap and body, with matching openings visible. Place a small threaded retaining ring beside its correct position on the axis, offset only enough to show its profile. The glass jar sits directly below the grinder body, its mouth aligned with the body’s lower threads. Lighting comes from a large soft panel at the upper left and a weaker fill on the right.\n\nMake the image explain a plausible manual mechanism through geometry and material detail. The crank is a short bent steel arm attached to a comfortable rounded wooden knob, with a visible mounting hole at its inner end. The top cap has a central bearing opening and a smooth rim that fits the body. A slender drive shaft descends from that bearing into the inner burr, while the outer burr forms a matching toothed ring. Suggest cutting teeth with crisp repeated grooves rather than decorative random spikes. Keep the inner and outer burrs proportionate so one can nest inside the other during assembly. The body is a brushed steel cylinder with a narrow rubber grip band, a clean top opening, and a threaded lower collar. The jar is a simple clear vessel with subtle refraction, its circular mouth becoming an ellipse in perspective. Show a few coffee beans beside the assembled axis near the surface as a scale reference, but do not scatter grounds across the floating components. Preserve consistent perspective and the same light direction on every piece. The background should remain neutral and unbranded, allowing fine reflections and edge separations to carry the image. Avoid arrows, diagram callouts, extra screws without a visible role, or impossible parts passing through each other. The final composition should read like a carefully art directed product explanation suitable for understanding how a small durable object is put together.", "scene_id": "coffee-grinder", "scene_title": "Exploded mechanical coffee grinder", "tier": 5, @@ -22908,13 +24320,13 @@ "seed": 42, "status": "succeeded", "accepted_result": true, - "prompt": "An older artisan sews a handmade book at a wooden workbench, focused on the needle and folded pages.\n\nPhotograph an invented anonymous person in a quiet bookbinding workshop. They wear a plain linen apron over a blue shirt. Shelves of paper, cloth, and unfinished bindings fill the softly lit background.\n\nShow both hands clearly: one steadies the folded paper sections, while the other guides a threaded needle through a prepared hole. A small wooden pressing frame sits nearby. Window light falls across the face, knuckles, paper edges, and worn tabletop. Keep skin texture natural, with fine wrinkles and realistic nails. The book has cream pages and a muted rust cloth cover waiting beside it.\n\nUse a three quarter viewpoint from across the workbench, framing the artisan from the waist upward with enough table space to understand the task. Their gaze follows the needle rather than the camera. The hands remain separated, with five plausible fingers on each where visible, and the thread follows one continuous path. Place a bone folder, ruler, and two small paper offcuts along the near edge, aligned with ordinary working habits rather than a decorative display. The book sections rest on the table, not suspended in midair. Let the background shelves recede gently into shadow while preserving a sense of a tidy, regularly used craft space.\n\nConvey concentration through posture, touch, and the relationship between tools and materials. The artisan leans forward slightly with relaxed shoulders, the elbows supported comfortably near the bench edge. The supporting hand presses the paper stack firmly enough to keep the sections aligned, while the sewing hand holds the needle between thumb and forefinger with a believable grip. A short length of waxed thread arcs from the needle to the last visible stitch, and the remaining strand lies loosely across the table without tangling through fingers. The paper folds have consistent thickness, with small stitch holes along the spine. The waiting cloth cover has turned edges and two flat boards beneath the fabric, suggesting its eventual construction. Add a modest spool of thread near the pressing frame and one closed notebook farther back, with no readable brand or title. The person\u2019s expression is calm and absorbed; avoid an exaggerated smile or a glamorous studio pose. Their hair, clothing folds, and skin should respond to the same soft daylight. Use warm timber tones, cool blue fabric, cream paper, and subdued rust for the cover, keeping the face and hands the clearest points of attention. The final portrait should respect the craft through precise physical actions and ordinary workshop detail, with a natural photographic depth of field that keeps the sewing operation legible.", + "prompt": "An older artisan sews a handmade book at a wooden workbench, focused on the needle and folded pages.\n\nPhotograph an invented anonymous person in a quiet bookbinding workshop. They wear a plain linen apron over a blue shirt. Shelves of paper, cloth, and unfinished bindings fill the softly lit background.\n\nShow both hands clearly: one steadies the folded paper sections, while the other guides a threaded needle through a prepared hole. A small wooden pressing frame sits nearby. Window light falls across the face, knuckles, paper edges, and worn tabletop. Keep skin texture natural, with fine wrinkles and realistic nails. The book has cream pages and a muted rust cloth cover waiting beside it.\n\nUse a three quarter viewpoint from across the workbench, framing the artisan from the waist upward with enough table space to understand the task. Their gaze follows the needle rather than the camera. The hands remain separated, with five plausible fingers on each where visible, and the thread follows one continuous path. Place a bone folder, ruler, and two small paper offcuts along the near edge, aligned with ordinary working habits rather than a decorative display. The book sections rest on the table, not suspended in midair. Let the background shelves recede gently into shadow while preserving a sense of a tidy, regularly used craft space.\n\nConvey concentration through posture, touch, and the relationship between tools and materials. The artisan leans forward slightly with relaxed shoulders, the elbows supported comfortably near the bench edge. The supporting hand presses the paper stack firmly enough to keep the sections aligned, while the sewing hand holds the needle between thumb and forefinger with a believable grip. A short length of waxed thread arcs from the needle to the last visible stitch, and the remaining strand lies loosely across the table without tangling through fingers. The paper folds have consistent thickness, with small stitch holes along the spine. The waiting cloth cover has turned edges and two flat boards beneath the fabric, suggesting its eventual construction. Add a modest spool of thread near the pressing frame and one closed notebook farther back, with no readable brand or title. The person’s expression is calm and absorbed; avoid an exaggerated smile or a glamorous studio pose. Their hair, clothing folds, and skin should respond to the same soft daylight. Use warm timber tones, cool blue fabric, cream paper, and subdued rust for the cover, keeping the face and hands the clearest points of attention. The final portrait should respect the craft through precise physical actions and ordinary workshop detail, with a natural photographic depth of field that keeps the sewing operation legible.", "prompt_details": { "conditioning_texts": { - "4.text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn older artisan sews a handmade book at a wooden workbench, focused on the needle and folded pages.\n\nPhotograph an invented anonymous person in a quiet bookbinding workshop. They wear a plain linen apron over a blue shirt. Shelves of paper, cloth, and unfinished bindings fill the softly lit background.\n\nShow both hands clearly: one steadies the folded paper sections, while the other guides a threaded needle through a prepared hole. A small wooden pressing frame sits nearby. Window light falls across the face, knuckles, paper edges, and worn tabletop. Keep skin texture natural, with fine wrinkles and realistic nails. The book has cream pages and a muted rust cloth cover waiting beside it.\n\nUse a three quarter viewpoint from across the workbench, framing the artisan from the waist upward with enough table space to understand the task. Their gaze follows the needle rather than the camera. The hands remain separated, with five plausible fingers on each where visible, and the thread follows one continuous path. Place a bone folder, ruler, and two small paper offcuts along the near edge, aligned with ordinary working habits rather than a decorative display. The book sections rest on the table, not suspended in midair. Let the background shelves recede gently into shadow while preserving a sense of a tidy, regularly used craft space.\n\nConvey concentration through posture, touch, and the relationship between tools and materials. The artisan leans forward slightly with relaxed shoulders, the elbows supported comfortably near the bench edge. The supporting hand presses the paper stack firmly enough to keep the sections aligned, while the sewing hand holds the needle between thumb and forefinger with a believable grip. A short length of waxed thread arcs from the needle to the last visible stitch, and the remaining strand lies loosely across the table without tangling through fingers. The paper folds have consistent thickness, with small stitch holes along the spine. The waiting cloth cover has turned edges and two flat boards beneath the fabric, suggesting its eventual construction. Add a modest spool of thread near the pressing frame and one closed notebook farther back, with no readable brand or title. The person\u2019s expression is calm and absorbed; avoid an exaggerated smile or a glamorous studio pose. Their hair, clothing folds, and skin should respond to the same soft daylight. Use warm timber tones, cool blue fabric, cream paper, and subdued rust for the cover, keeping the face and hands the clearest points of attention. The final portrait should respect the craft through precise physical actions and ordinary workshop detail, with a natural photographic depth of field that keeps the sewing operation legible.", + "4.text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn older artisan sews a handmade book at a wooden workbench, focused on the needle and folded pages.\n\nPhotograph an invented anonymous person in a quiet bookbinding workshop. They wear a plain linen apron over a blue shirt. Shelves of paper, cloth, and unfinished bindings fill the softly lit background.\n\nShow both hands clearly: one steadies the folded paper sections, while the other guides a threaded needle through a prepared hole. A small wooden pressing frame sits nearby. Window light falls across the face, knuckles, paper edges, and worn tabletop. Keep skin texture natural, with fine wrinkles and realistic nails. The book has cream pages and a muted rust cloth cover waiting beside it.\n\nUse a three quarter viewpoint from across the workbench, framing the artisan from the waist upward with enough table space to understand the task. Their gaze follows the needle rather than the camera. The hands remain separated, with five plausible fingers on each where visible, and the thread follows one continuous path. Place a bone folder, ruler, and two small paper offcuts along the near edge, aligned with ordinary working habits rather than a decorative display. The book sections rest on the table, not suspended in midair. Let the background shelves recede gently into shadow while preserving a sense of a tidy, regularly used craft space.\n\nConvey concentration through posture, touch, and the relationship between tools and materials. The artisan leans forward slightly with relaxed shoulders, the elbows supported comfortably near the bench edge. The supporting hand presses the paper stack firmly enough to keep the sections aligned, while the sewing hand holds the needle between thumb and forefinger with a believable grip. A short length of waxed thread arcs from the needle to the last visible stitch, and the remaining strand lies loosely across the table without tangling through fingers. The paper folds have consistent thickness, with small stitch holes along the spine. The waiting cloth cover has turned edges and two flat boards beneath the fabric, suggesting its eventual construction. Add a modest spool of thread near the pressing frame and one closed notebook farther back, with no readable brand or title. The person’s expression is calm and absorbed; avoid an exaggerated smile or a glamorous studio pose. Their hair, clothing folds, and skin should respond to the same soft daylight. Use warm timber tones, cool blue fabric, cream paper, and subdued rust for the cover, keeping the face and hands the clearest points of attention. The final portrait should respect the craft through precise physical actions and ordinary workshop detail, with a natural photographic depth of field that keeps the sewing operation legible.", "5.text": "" }, - "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn older artisan sews a handmade book at a wooden workbench, focused on the needle and folded pages.\n\nPhotograph an invented anonymous person in a quiet bookbinding workshop. They wear a plain linen apron over a blue shirt. Shelves of paper, cloth, and unfinished bindings fill the softly lit background.\n\nShow both hands clearly: one steadies the folded paper sections, while the other guides a threaded needle through a prepared hole. A small wooden pressing frame sits nearby. Window light falls across the face, knuckles, paper edges, and worn tabletop. Keep skin texture natural, with fine wrinkles and realistic nails. The book has cream pages and a muted rust cloth cover waiting beside it.\n\nUse a three quarter viewpoint from across the workbench, framing the artisan from the waist upward with enough table space to understand the task. Their gaze follows the needle rather than the camera. The hands remain separated, with five plausible fingers on each where visible, and the thread follows one continuous path. Place a bone folder, ruler, and two small paper offcuts along the near edge, aligned with ordinary working habits rather than a decorative display. The book sections rest on the table, not suspended in midair. Let the background shelves recede gently into shadow while preserving a sense of a tidy, regularly used craft space.\n\nConvey concentration through posture, touch, and the relationship between tools and materials. The artisan leans forward slightly with relaxed shoulders, the elbows supported comfortably near the bench edge. The supporting hand presses the paper stack firmly enough to keep the sections aligned, while the sewing hand holds the needle between thumb and forefinger with a believable grip. A short length of waxed thread arcs from the needle to the last visible stitch, and the remaining strand lies loosely across the table without tangling through fingers. The paper folds have consistent thickness, with small stitch holes along the spine. The waiting cloth cover has turned edges and two flat boards beneath the fabric, suggesting its eventual construction. Add a modest spool of thread near the pressing frame and one closed notebook farther back, with no readable brand or title. The person\u2019s expression is calm and absorbed; avoid an exaggerated smile or a glamorous studio pose. Their hair, clothing folds, and skin should respond to the same soft daylight. Use warm timber tones, cool blue fabric, cream paper, and subdued rust for the cover, keeping the face and hands the clearest points of attention. The final portrait should respect the craft through precise physical actions and ordinary workshop detail, with a natural photographic depth of field that keeps the sewing operation legible.", + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn older artisan sews a handmade book at a wooden workbench, focused on the needle and folded pages.\n\nPhotograph an invented anonymous person in a quiet bookbinding workshop. They wear a plain linen apron over a blue shirt. Shelves of paper, cloth, and unfinished bindings fill the softly lit background.\n\nShow both hands clearly: one steadies the folded paper sections, while the other guides a threaded needle through a prepared hole. A small wooden pressing frame sits nearby. Window light falls across the face, knuckles, paper edges, and worn tabletop. Keep skin texture natural, with fine wrinkles and realistic nails. The book has cream pages and a muted rust cloth cover waiting beside it.\n\nUse a three quarter viewpoint from across the workbench, framing the artisan from the waist upward with enough table space to understand the task. Their gaze follows the needle rather than the camera. The hands remain separated, with five plausible fingers on each where visible, and the thread follows one continuous path. Place a bone folder, ruler, and two small paper offcuts along the near edge, aligned with ordinary working habits rather than a decorative display. The book sections rest on the table, not suspended in midair. Let the background shelves recede gently into shadow while preserving a sense of a tidy, regularly used craft space.\n\nConvey concentration through posture, touch, and the relationship between tools and materials. The artisan leans forward slightly with relaxed shoulders, the elbows supported comfortably near the bench edge. The supporting hand presses the paper stack firmly enough to keep the sections aligned, while the sewing hand holds the needle between thumb and forefinger with a believable grip. A short length of waxed thread arcs from the needle to the last visible stitch, and the remaining strand lies loosely across the table without tangling through fingers. The paper folds have consistent thickness, with small stitch holes along the spine. The waiting cloth cover has turned edges and two flat boards beneath the fabric, suggesting its eventual construction. Add a modest spool of thread near the pressing frame and one closed notebook farther back, with no readable brand or title. The person’s expression is calm and absorbed; avoid an exaggerated smile or a glamorous studio pose. Their hair, clothing folds, and skin should respond to the same soft daylight. Use warm timber tones, cool blue fabric, cream paper, and subdued rust for the cover, keeping the face and hands the clearest points of attention. The final portrait should respect the craft through precise physical actions and ordinary workshop detail, with a natural photographic depth of field that keeps the sewing operation legible.", "scene_id": "bookbinder-portrait", "scene_title": "An artisan binding a book", "tier": 5, @@ -23119,13 +24531,13 @@ "seed": 43, "status": "succeeded", "accepted_result": true, - "prompt": "An older artisan sews a handmade book at a wooden workbench, focused on the needle and folded pages.\n\nPhotograph an invented anonymous person in a quiet bookbinding workshop. They wear a plain linen apron over a blue shirt. Shelves of paper, cloth, and unfinished bindings fill the softly lit background.\n\nShow both hands clearly: one steadies the folded paper sections, while the other guides a threaded needle through a prepared hole. A small wooden pressing frame sits nearby. Window light falls across the face, knuckles, paper edges, and worn tabletop. Keep skin texture natural, with fine wrinkles and realistic nails. The book has cream pages and a muted rust cloth cover waiting beside it.\n\nUse a three quarter viewpoint from across the workbench, framing the artisan from the waist upward with enough table space to understand the task. Their gaze follows the needle rather than the camera. The hands remain separated, with five plausible fingers on each where visible, and the thread follows one continuous path. Place a bone folder, ruler, and two small paper offcuts along the near edge, aligned with ordinary working habits rather than a decorative display. The book sections rest on the table, not suspended in midair. Let the background shelves recede gently into shadow while preserving a sense of a tidy, regularly used craft space.\n\nConvey concentration through posture, touch, and the relationship between tools and materials. The artisan leans forward slightly with relaxed shoulders, the elbows supported comfortably near the bench edge. The supporting hand presses the paper stack firmly enough to keep the sections aligned, while the sewing hand holds the needle between thumb and forefinger with a believable grip. A short length of waxed thread arcs from the needle to the last visible stitch, and the remaining strand lies loosely across the table without tangling through fingers. The paper folds have consistent thickness, with small stitch holes along the spine. The waiting cloth cover has turned edges and two flat boards beneath the fabric, suggesting its eventual construction. Add a modest spool of thread near the pressing frame and one closed notebook farther back, with no readable brand or title. The person\u2019s expression is calm and absorbed; avoid an exaggerated smile or a glamorous studio pose. Their hair, clothing folds, and skin should respond to the same soft daylight. Use warm timber tones, cool blue fabric, cream paper, and subdued rust for the cover, keeping the face and hands the clearest points of attention. The final portrait should respect the craft through precise physical actions and ordinary workshop detail, with a natural photographic depth of field that keeps the sewing operation legible.", + "prompt": "An older artisan sews a handmade book at a wooden workbench, focused on the needle and folded pages.\n\nPhotograph an invented anonymous person in a quiet bookbinding workshop. They wear a plain linen apron over a blue shirt. Shelves of paper, cloth, and unfinished bindings fill the softly lit background.\n\nShow both hands clearly: one steadies the folded paper sections, while the other guides a threaded needle through a prepared hole. A small wooden pressing frame sits nearby. Window light falls across the face, knuckles, paper edges, and worn tabletop. Keep skin texture natural, with fine wrinkles and realistic nails. The book has cream pages and a muted rust cloth cover waiting beside it.\n\nUse a three quarter viewpoint from across the workbench, framing the artisan from the waist upward with enough table space to understand the task. Their gaze follows the needle rather than the camera. The hands remain separated, with five plausible fingers on each where visible, and the thread follows one continuous path. Place a bone folder, ruler, and two small paper offcuts along the near edge, aligned with ordinary working habits rather than a decorative display. The book sections rest on the table, not suspended in midair. Let the background shelves recede gently into shadow while preserving a sense of a tidy, regularly used craft space.\n\nConvey concentration through posture, touch, and the relationship between tools and materials. The artisan leans forward slightly with relaxed shoulders, the elbows supported comfortably near the bench edge. The supporting hand presses the paper stack firmly enough to keep the sections aligned, while the sewing hand holds the needle between thumb and forefinger with a believable grip. A short length of waxed thread arcs from the needle to the last visible stitch, and the remaining strand lies loosely across the table without tangling through fingers. The paper folds have consistent thickness, with small stitch holes along the spine. The waiting cloth cover has turned edges and two flat boards beneath the fabric, suggesting its eventual construction. Add a modest spool of thread near the pressing frame and one closed notebook farther back, with no readable brand or title. The person’s expression is calm and absorbed; avoid an exaggerated smile or a glamorous studio pose. Their hair, clothing folds, and skin should respond to the same soft daylight. Use warm timber tones, cool blue fabric, cream paper, and subdued rust for the cover, keeping the face and hands the clearest points of attention. The final portrait should respect the craft through precise physical actions and ordinary workshop detail, with a natural photographic depth of field that keeps the sewing operation legible.", "prompt_details": { "conditioning_texts": { - "4.text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn older artisan sews a handmade book at a wooden workbench, focused on the needle and folded pages.\n\nPhotograph an invented anonymous person in a quiet bookbinding workshop. They wear a plain linen apron over a blue shirt. Shelves of paper, cloth, and unfinished bindings fill the softly lit background.\n\nShow both hands clearly: one steadies the folded paper sections, while the other guides a threaded needle through a prepared hole. A small wooden pressing frame sits nearby. Window light falls across the face, knuckles, paper edges, and worn tabletop. Keep skin texture natural, with fine wrinkles and realistic nails. The book has cream pages and a muted rust cloth cover waiting beside it.\n\nUse a three quarter viewpoint from across the workbench, framing the artisan from the waist upward with enough table space to understand the task. Their gaze follows the needle rather than the camera. The hands remain separated, with five plausible fingers on each where visible, and the thread follows one continuous path. Place a bone folder, ruler, and two small paper offcuts along the near edge, aligned with ordinary working habits rather than a decorative display. The book sections rest on the table, not suspended in midair. Let the background shelves recede gently into shadow while preserving a sense of a tidy, regularly used craft space.\n\nConvey concentration through posture, touch, and the relationship between tools and materials. The artisan leans forward slightly with relaxed shoulders, the elbows supported comfortably near the bench edge. The supporting hand presses the paper stack firmly enough to keep the sections aligned, while the sewing hand holds the needle between thumb and forefinger with a believable grip. A short length of waxed thread arcs from the needle to the last visible stitch, and the remaining strand lies loosely across the table without tangling through fingers. The paper folds have consistent thickness, with small stitch holes along the spine. The waiting cloth cover has turned edges and two flat boards beneath the fabric, suggesting its eventual construction. Add a modest spool of thread near the pressing frame and one closed notebook farther back, with no readable brand or title. The person\u2019s expression is calm and absorbed; avoid an exaggerated smile or a glamorous studio pose. Their hair, clothing folds, and skin should respond to the same soft daylight. Use warm timber tones, cool blue fabric, cream paper, and subdued rust for the cover, keeping the face and hands the clearest points of attention. The final portrait should respect the craft through precise physical actions and ordinary workshop detail, with a natural photographic depth of field that keeps the sewing operation legible.", + "4.text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn older artisan sews a handmade book at a wooden workbench, focused on the needle and folded pages.\n\nPhotograph an invented anonymous person in a quiet bookbinding workshop. They wear a plain linen apron over a blue shirt. Shelves of paper, cloth, and unfinished bindings fill the softly lit background.\n\nShow both hands clearly: one steadies the folded paper sections, while the other guides a threaded needle through a prepared hole. A small wooden pressing frame sits nearby. Window light falls across the face, knuckles, paper edges, and worn tabletop. Keep skin texture natural, with fine wrinkles and realistic nails. The book has cream pages and a muted rust cloth cover waiting beside it.\n\nUse a three quarter viewpoint from across the workbench, framing the artisan from the waist upward with enough table space to understand the task. Their gaze follows the needle rather than the camera. The hands remain separated, with five plausible fingers on each where visible, and the thread follows one continuous path. Place a bone folder, ruler, and two small paper offcuts along the near edge, aligned with ordinary working habits rather than a decorative display. The book sections rest on the table, not suspended in midair. Let the background shelves recede gently into shadow while preserving a sense of a tidy, regularly used craft space.\n\nConvey concentration through posture, touch, and the relationship between tools and materials. The artisan leans forward slightly with relaxed shoulders, the elbows supported comfortably near the bench edge. The supporting hand presses the paper stack firmly enough to keep the sections aligned, while the sewing hand holds the needle between thumb and forefinger with a believable grip. A short length of waxed thread arcs from the needle to the last visible stitch, and the remaining strand lies loosely across the table without tangling through fingers. The paper folds have consistent thickness, with small stitch holes along the spine. The waiting cloth cover has turned edges and two flat boards beneath the fabric, suggesting its eventual construction. Add a modest spool of thread near the pressing frame and one closed notebook farther back, with no readable brand or title. The person’s expression is calm and absorbed; avoid an exaggerated smile or a glamorous studio pose. Their hair, clothing folds, and skin should respond to the same soft daylight. Use warm timber tones, cool blue fabric, cream paper, and subdued rust for the cover, keeping the face and hands the clearest points of attention. The final portrait should respect the craft through precise physical actions and ordinary workshop detail, with a natural photographic depth of field that keeps the sewing operation legible.", "5.text": "" }, - "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn older artisan sews a handmade book at a wooden workbench, focused on the needle and folded pages.\n\nPhotograph an invented anonymous person in a quiet bookbinding workshop. They wear a plain linen apron over a blue shirt. Shelves of paper, cloth, and unfinished bindings fill the softly lit background.\n\nShow both hands clearly: one steadies the folded paper sections, while the other guides a threaded needle through a prepared hole. A small wooden pressing frame sits nearby. Window light falls across the face, knuckles, paper edges, and worn tabletop. Keep skin texture natural, with fine wrinkles and realistic nails. The book has cream pages and a muted rust cloth cover waiting beside it.\n\nUse a three quarter viewpoint from across the workbench, framing the artisan from the waist upward with enough table space to understand the task. Their gaze follows the needle rather than the camera. The hands remain separated, with five plausible fingers on each where visible, and the thread follows one continuous path. Place a bone folder, ruler, and two small paper offcuts along the near edge, aligned with ordinary working habits rather than a decorative display. The book sections rest on the table, not suspended in midair. Let the background shelves recede gently into shadow while preserving a sense of a tidy, regularly used craft space.\n\nConvey concentration through posture, touch, and the relationship between tools and materials. The artisan leans forward slightly with relaxed shoulders, the elbows supported comfortably near the bench edge. The supporting hand presses the paper stack firmly enough to keep the sections aligned, while the sewing hand holds the needle between thumb and forefinger with a believable grip. A short length of waxed thread arcs from the needle to the last visible stitch, and the remaining strand lies loosely across the table without tangling through fingers. The paper folds have consistent thickness, with small stitch holes along the spine. The waiting cloth cover has turned edges and two flat boards beneath the fabric, suggesting its eventual construction. Add a modest spool of thread near the pressing frame and one closed notebook farther back, with no readable brand or title. The person\u2019s expression is calm and absorbed; avoid an exaggerated smile or a glamorous studio pose. Their hair, clothing folds, and skin should respond to the same soft daylight. Use warm timber tones, cool blue fabric, cream paper, and subdued rust for the cover, keeping the face and hands the clearest points of attention. The final portrait should respect the craft through precise physical actions and ordinary workshop detail, with a natural photographic depth of field that keeps the sewing operation legible.", + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn older artisan sews a handmade book at a wooden workbench, focused on the needle and folded pages.\n\nPhotograph an invented anonymous person in a quiet bookbinding workshop. They wear a plain linen apron over a blue shirt. Shelves of paper, cloth, and unfinished bindings fill the softly lit background.\n\nShow both hands clearly: one steadies the folded paper sections, while the other guides a threaded needle through a prepared hole. A small wooden pressing frame sits nearby. Window light falls across the face, knuckles, paper edges, and worn tabletop. Keep skin texture natural, with fine wrinkles and realistic nails. The book has cream pages and a muted rust cloth cover waiting beside it.\n\nUse a three quarter viewpoint from across the workbench, framing the artisan from the waist upward with enough table space to understand the task. Their gaze follows the needle rather than the camera. The hands remain separated, with five plausible fingers on each where visible, and the thread follows one continuous path. Place a bone folder, ruler, and two small paper offcuts along the near edge, aligned with ordinary working habits rather than a decorative display. The book sections rest on the table, not suspended in midair. Let the background shelves recede gently into shadow while preserving a sense of a tidy, regularly used craft space.\n\nConvey concentration through posture, touch, and the relationship between tools and materials. The artisan leans forward slightly with relaxed shoulders, the elbows supported comfortably near the bench edge. The supporting hand presses the paper stack firmly enough to keep the sections aligned, while the sewing hand holds the needle between thumb and forefinger with a believable grip. A short length of waxed thread arcs from the needle to the last visible stitch, and the remaining strand lies loosely across the table without tangling through fingers. The paper folds have consistent thickness, with small stitch holes along the spine. The waiting cloth cover has turned edges and two flat boards beneath the fabric, suggesting its eventual construction. Add a modest spool of thread near the pressing frame and one closed notebook farther back, with no readable brand or title. The person’s expression is calm and absorbed; avoid an exaggerated smile or a glamorous studio pose. Their hair, clothing folds, and skin should respond to the same soft daylight. Use warm timber tones, cool blue fabric, cream paper, and subdued rust for the cover, keeping the face and hands the clearest points of attention. The final portrait should respect the craft through precise physical actions and ordinary workshop detail, with a natural photographic depth of field that keeps the sewing operation legible.", "scene_id": "bookbinder-portrait", "scene_title": "An artisan binding a book", "tier": 5, @@ -23742,7 +25154,7 @@ }, { "id": "cog-03", - "label": "CogVideoX 2B local Q8_0 \u00b7 seed 42", + "label": "CogVideoX 2B local Q8_0 · seed 42", "group": "cogvideox2b-q8-reference-ddim-text-v1", "kind": "video", "model_id": "cog", @@ -23883,7 +25295,7 @@ }, { "id": "cog-04", - "label": "CogVideoX 2B local Q8_0 \u00b7 seed 43", + "label": "CogVideoX 2B local Q8_0 · seed 43", "group": "cogvideox2b-q8-reference-ddim-text-v1", "kind": "video", "model_id": "cog", @@ -24017,12 +25429,1461 @@ } ], "source_workflow_canonical_sha256": "ab6c1546ca57009496c946973c7c6bdf9cf3359119cd420cfe00cb42ad4a606e" + }, + { + "id": "qwen-turbo-01", + "label": "Qwen Image 2.1 Turbo Q4_K_M · Initial request including model loading · seed 42", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 42, + "status": "succeeded", + "accepted_result": true, + "prompt": "A small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph.", + "prompt_details": { + "word_count": 19, + "complexity": "Very short / retained runtime pair", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 42 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 334.589337, + "comfy_server_seconds": 320.246, + "stages_seconds": { + "node_4_observer_receive_interval": 4.436999999976251, + "node_2_observer_receive_interval": 6.48399999999674, + "node_3_observer_receive_interval": 104.32900000002701, + "node_12_observer_receive_interval": 0.031000000017229468, + "node_11_observer_receive_interval": 0.014999999955762178, + "node_1_observer_receive_interval": 1.6570000000065193, + "node_5_observer_receive_interval": 0.10899999999674037, + "node_10_observer_receive_interval": 0, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 159.85899999999674, + "node_7_observer_receive_interval": 40.75, + "node_8_observer_receive_interval": 2.047000000020489 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [], + "same_runtime_retained": null, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "Initial request including model loading", + "native_inter_run_gap_seconds": null, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": false, + "cold_start_scope": "The first request includes initial/model work; OS disk-cache state was not isolated." + } + }, + "resources": { + "sampled_summary": { + "sample_count": 162, + "inflight_sample_count": 161, + "minimum_ram_available_bytes": 1912422400, + "minimum_commit_available_bytes": 11296174080, + "minimum_system_drive_free_bytes": 18670813184, + "maximum_sampled_device_gpu_used_mib": 3983 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "A coherent fox figurine standing on a pale wooden desk against a cream background; recognizably matches the simple scene.", + "Orange fox, desk, cream background and soft studio light are present.", + "Fine fur-like ridges dominate the body; ceramic material is not convincingly conveyed.", + "Legs, ears and tail are largely plausible for a stylized figurine. Face is slightly toy-like.", + "No text requested or visibly generated.", + "Ceramic material appears more like textured wood, fur or a molded toy." + ], + "media": [ + { + "path": "media/qwen-turbo-01.png", + "mime_type": "image/png", + "size_bytes": 261046, + "sha256": "56129ce43e04bb2aca5849e23b9d6a0c9adfdee3047a9329f1fcc360c2d9eb2e", + "source_sha256": "c4a363cb2d3fbcb5d51b02226bdb57facc1963d98cac306cefb3ad252700b3d0", + "width": 512, + "height": 512, + "pixel_sha256": "5df320c6bdbdad6dc8dbc07af73f2dc513a55deecad9b2f49b659e85ba5ba2f4", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-01.json", + "workflow_sha256": "4369239d25533a14c3b5bb3a15d663b4049d3285e9804cc52483e7ffccaf4852", + "source_workflow_canonical_sha256": "1ff73d5851a428f07381767c01cfdf4675c6c0b6f37e9df58ff08a9c74de77bb" + }, + { + "id": "qwen-turbo-02", + "label": "Qwen Image 2.1 Turbo Q4_K_M · Same brief, next seed · seed 43", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 43, + "status": "succeeded", + "accepted_result": true, + "prompt": "A small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph.", + "prompt_details": { + "word_count": 19, + "complexity": "Very short / retained runtime pair", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 43 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 96.670077, + "comfy_server_seconds": 95.234, + "stages_seconds": { + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 67.14100000000326, + "node_7_observer_receive_interval": 27.82800000003772, + "node_8_observer_receive_interval": 0.06299999996554106 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "10", + "11", + "12", + "2", + "3", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": true, + "context": { + "workload": "Same brief, next seed", + "native_inter_run_gap_seconds": 34.322, + "only_seed_changed_from_predecessor": true, + "text_changed_from_predecessor": false, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 48, + "inflight_sample_count": 47, + "minimum_ram_available_bytes": 8770334720, + "minimum_commit_available_bytes": 11553648640, + "minimum_system_drive_free_bytes": 18671108096, + "maximum_sampled_device_gpu_used_mib": 4315 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": true, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "A coherent front-facing seated fox figurine in the same basic studio setup, with a visibly different pose and face.", + "Orange fox, light wood and cream background are present; composition follows the brief.", + "Matte textured surface with fur-like ridges; not strongly ceramic.", + "Stylized anatomy is mostly coherent. Eye appearance is uneven, with a pale/cyan mark around the left visible eye.", + "No text requested or visibly generated.", + "Ceramic appearance is weak.", + "Small asymmetric eye detail may be a visual artifact." + ], + "media": [ + { + "path": "media/qwen-turbo-02.png", + "mime_type": "image/png", + "size_bytes": 283217, + "sha256": "238d33136a36561317196ed2a9c0096cbafe4109059ab2d0d8df5ec3079d74bd", + "source_sha256": "f8bd3fcfba3462d120142e83fb0d7ec637a9eceebd25f6db11f61bda76254eb1", + "width": 512, + "height": 512, + "pixel_sha256": "ac9f30aded9dc1c28b70d2b87cebd61a4df3d93e3dc62206d1ed31874410fe48", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-02.json", + "workflow_sha256": "a48c407f70a6f7ffb6722d35f80c1d304116d197f2307961e7f88bbd31456fa4", + "source_workflow_canonical_sha256": "84fabe49d8035de7dc9124d16ea9729ca181a0fdbca0f9e681520453dbb87ef2" + }, + { + "id": "qwen-turbo-03", + "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 44", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 44, + "status": "succeeded", + "accepted_result": true, + "prompt": "A turquoise alpine lake beneath snowy mountains at sunrise, a yellow canoe near the shore, crisp natural landscape photograph.", + "prompt_details": { + "word_count": 19, + "complexity": "Short landscape", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA turquoise alpine lake beneath snowy mountains at sunrise, a yellow canoe near the shore, crisp natural landscape photograph." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA turquoise alpine lake beneath snowy mountains at sunrise, a yellow canoe near the shore, crisp natural landscape photograph." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 44 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 195.952306, + "comfy_server_seconds": 194.865, + "stages_seconds": { + "node_3_observer_receive_interval": 100.20300000003772, + "node_10_observer_receive_interval": 0.014999999955762178, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 66.93800000002375, + "node_7_observer_receive_interval": 27.64100000000326, + "node_8_observer_receive_interval": 0.045999999972991645 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "11", + "12", + "2", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "New brief, retained model runtime", + "native_inter_run_gap_seconds": 22.977, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": true, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 96, + "inflight_sample_count": 95, + "minimum_ram_available_bytes": 4403757056, + "minimum_commit_available_bytes": 11454955520, + "minimum_system_drive_free_bytes": 18670780416, + "maximum_sampled_device_gpu_used_mib": 4315 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "Recognizable and visually coherent mountain-lake landscape with a yellow canoe-like object at the near shoreline.", + "Turquoise water, snowy mountains and warm sunlight on peaks are present. The yellow canoe is close to shore.", + "Water reflections and rock/snow separation are convincing at this size; fine granular texture is visible.", + "The canoe is heavily cropped by the lower edge, showing mostly its bow/top; full canoe geometry cannot be assessed.", + "No text requested or visibly generated.", + "Canoe is strongly cropped and less clearly legible than a full near-shore canoe.", + "Fine speckled texture is visible on the mountain surfaces." + ], + "media": [ + { + "path": "media/qwen-turbo-03.png", + "mime_type": "image/png", + "size_bytes": 474011, + "sha256": "57418cb5f47b5647b8773c0e012572e528e150b0f27f422e78f4f05d5a683a11", + "source_sha256": "c5e2e3901c3129df9943c2f995d5b42c7d646c708eb6f9053295778ed77012e7", + "width": 512, + "height": 512, + "pixel_sha256": "ff22ec81f55d0251c62e225baca505db986d0d734728cb05f57b4d2087cd12d1", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-03.json", + "workflow_sha256": "d26655caaebfa0bd74939c8f5f1977d3759141277d816b6eb1790a1c18f853c1", + "source_workflow_canonical_sha256": "96b30904e6a2fa858c6bc001bf5a0b3e4168fbcbc36fcf53b796c7ec9f07cfd2" + }, + { + "id": "qwen-turbo-04", + "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 45", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 45, + "status": "succeeded", + "accepted_result": true, + "prompt": "Product photograph of a matte black ceramic teapot with a warm copper handle, resting on dark slate beside two pale linen napkins. Three-quarter view, softbox lighting, crisp material detail, uncluttered background.", + "prompt_details": { + "word_count": 31, + "complexity": "Product materials", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nProduct photograph of a matte black ceramic teapot with a warm copper handle, resting on dark slate beside two pale linen napkins. Three-quarter view, softbox lighting, crisp material detail, uncluttered background." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nProduct photograph of a matte black ceramic teapot with a warm copper handle, resting on dark slate beside two pale linen napkins. Three-quarter view, softbox lighting, crisp material detail, uncluttered background." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 45 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 198.952217, + "comfy_server_seconds": 196.417, + "stages_seconds": { + "node_3_observer_receive_interval": 100.10899999999674, + "node_10_observer_receive_interval": 0.01600000000325963, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 67.98399999999674, + "node_7_observer_receive_interval": 28.21899999998277, + "node_8_observer_receive_interval": 0.061999999976251274 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "11", + "12", + "2", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "New brief, retained model runtime", + "native_inter_run_gap_seconds": 16.828, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": true, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 98, + "inflight_sample_count": 97, + "minimum_ram_available_bytes": 4163956736, + "minimum_commit_available_bytes": 11323981824, + "minimum_system_drive_free_bytes": 18670002176, + "maximum_sampled_device_gpu_used_mib": 4351 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "A coherent, usable-looking product composition of a black teapot, copper-colored handle, dark slate and pale linen.", + "Main object, materials, slate and three-quarter product lighting are present. Only one clearly distinguishable rolled/folded napkin appears, rather than two separate napkins.", + "Black matte body, metallic handle and woven linen are well differentiated.", + "Teapot body, lid, spout and handle joins are broadly plausible; no obvious broken geometry at 512px.", + "No text requested or visibly generated.", + "Requested count of two napkins is not clearly met." + ], + "media": [ + { + "path": "media/qwen-turbo-04.png", + "mime_type": "image/png", + "size_bytes": 280344, + "sha256": "e5b3093fdb6bb39fcec4b803b1345733cf212bb274806338d27e140d3b882c3d", + "source_sha256": "1c2b9ddeca53a1fd944073a596b050e43f8f9b1b4a7e2fee57608095395b97d3", + "width": 512, + "height": 512, + "pixel_sha256": "2db9b0cf39ba6fdb3f0febf48a4f32d9fbd8012ed303d8d3ec711a4a074f8b81", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-04.json", + "workflow_sha256": "434243d6ae778adde215a2487cc10e1ee4123a6826f0f50c21e1b41238930d75", + "source_workflow_canonical_sha256": "7592e027ed1ef1f6903c6ca2bdae9faad31a02bf99cd7f67c71689cc3427c0ce" + }, + { + "id": "qwen-turbo-05", + "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 46", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 46, + "status": "succeeded", + "accepted_result": true, + "prompt": "A cozy neighborhood bakery storefront on a rainy evening. The main sign clearly reads \"MORNING BREAD\" in cream block letters. Warm light glows through the windows, a bicycle leans beside the door, wet paving reflects the orange interior, realistic street photograph.", + "prompt_details": { + "word_count": 41, + "complexity": "Readable sign", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA cozy neighborhood bakery storefront on a rainy evening. The main sign clearly reads \"MORNING BREAD\" in cream block letters. Warm light glows through the windows, a bicycle leans beside the door, wet paving reflects the orange interior, realistic street photograph." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA cozy neighborhood bakery storefront on a rainy evening. The main sign clearly reads \"MORNING BREAD\" in cream block letters. Warm light glows through the windows, a bicycle leans beside the door, wet paving reflects the orange interior, realistic street photograph." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 46 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 201.997778, + "comfy_server_seconds": 199.64, + "stages_seconds": { + "node_3_observer_receive_interval": 102.10899999999674, + "node_10_observer_receive_interval": 0.01600000000325963, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 68.81200000003446, + "node_7_observer_receive_interval": 28.625, + "node_8_observer_receive_interval": 0.06299999996554106 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "11", + "12", + "2", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "New brief, retained model runtime", + "native_inter_run_gap_seconds": 16.283, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": true, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 99, + "inflight_sample_count": 98, + "minimum_ram_available_bytes": 4295344128, + "minimum_commit_available_bytes": 11412852736, + "minimum_system_drive_free_bytes": 18669809664, + "maximum_sampled_device_gpu_used_mib": 4351 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "A coherent rainy-evening bakery storefront with a legible sign, bicycle, warm windows and wet paving.", + "The main sign visibly reads exactly MORNING BREAD in large cream block letters. Bicycle, orange interior and reflected light are present.", + "Brick, glazing and wet paving read distinctly; the dark storefront feels photograph-like.", + "Storefront perspective and bicycle placement are broadly plausible at this scale; fine bicycle detail is somewhat soft.", + "Exact requested wording MORNING BREAD is visibly correct. This is a subjective visual transcription, not OCR or an automated text benchmark.", + "Fine bicycle/spoke details are soft at the supplied resolution." + ], + "media": [ + { + "path": "media/qwen-turbo-05.png", + "mime_type": "image/png", + "size_bytes": 434268, + "sha256": "94e64b09278ffff4feb04c3197545dd8d6371b281a514d0e3d1cf8550648d919", + "source_sha256": "5a9008f775161ef48b16915597d158f4a0cd80883a7c4d452cbf240bba049ce2", + "width": 512, + "height": 512, + "pixel_sha256": "18129fd4fc85277093ff6d330be029055bfde66f2a18af657ce735b6e928f894", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-05.json", + "workflow_sha256": "8768f247abc378db4fd3057cb67987743253489f11b42e75bef673f1b6c39658", + "source_workflow_canonical_sha256": "874d22b62ef466732130e9e79fa45e12bf32d748a3b1124d21b16aedac7abf59" + }, + { + "id": "qwen-turbo-06", + "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 47", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 47, + "status": "succeeded", + "accepted_result": true, + "prompt": "A bright Victorian glass greenhouse filled with lush ferns, hanging orchids and terracotta pots. A narrow brick path curves from the bottom left toward an open green door at the far end. Morning sunlight makes geometric shadows through the iron window frames. Include a single brass watering can beside the path and misty trees beyond the glass. Architectural photograph, balanced natural colors, no people or text.", + "prompt_details": { + "word_count": 66, + "complexity": "Botanical architecture", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA bright Victorian glass greenhouse filled with lush ferns, hanging orchids and terracotta pots. A narrow brick path curves from the bottom left toward an open green door at the far end. Morning sunlight makes geometric shadows through the iron window frames. Include a single brass watering can beside the path and misty trees beyond the glass. Architectural photograph, balanced natural colors, no people or text." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA bright Victorian glass greenhouse filled with lush ferns, hanging orchids and terracotta pots. A narrow brick path curves from the bottom left toward an open green door at the far end. Morning sunlight makes geometric shadows through the iron window frames. Include a single brass watering can beside the path and misty trees beyond the glass. Architectural photograph, balanced natural colors, no people or text." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 47 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 201.931, + "comfy_server_seconds": 199.412, + "stages_seconds": { + "node_3_observer_receive_interval": 102.79699999996228, + "node_10_observer_receive_interval": 0.015000000013969839, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 68.43800000002375, + "node_7_observer_receive_interval": 28.04699999996228, + "node_8_observer_receive_interval": 0.062000000034458935 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "11", + "12", + "2", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "New brief, retained model runtime", + "native_inter_run_gap_seconds": 21.517, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": true, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 99, + "inflight_sample_count": 98, + "minimum_ram_available_bytes": 4470677504, + "minimum_commit_available_bytes": 11370139648, + "minimum_system_drive_free_bytes": 18651901952, + "maximum_sampled_device_gpu_used_mib": 4351 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "A coherent glass greenhouse with dense foliage, terracotta pots, a brick walkway, an open green door and a brass-colored watering can.", + "Most named scene elements are present. The path leads toward an open far door; ferns and hanging plants dominate. Distinct hanging orchid flowers are not obvious, and geometric sunlight shadows are subtle.", + "Glass/metal frames, foliage, pots and brick are distinguishable, but foliage and frame textures have noticeable fine speckling.", + "Overall greenhouse perspective is convincing; dense plant/frame detail is less clean on close inspection. Watering can is partly occluded/near the left edge.", + "No people or text requested; none visibly apparent.", + "Requested orchids are not clearly identifiable.", + "Geometric cast shadows are weaker than the brief suggests.", + "Dense foliage and framing show grainy/speckled texture." + ], + "media": [ + { + "path": "media/qwen-turbo-06.png", + "mime_type": "image/png", + "size_bytes": 649441, + "sha256": "f1ae1c3294bcffe1c0d283e957bcc81c5d48df83fe6e0f316eb4182d57db2af1", + "source_sha256": "bd1ac8888861a03d6b9a2436cede0ad7b3009f1535c11b360f794ad47a9c0321", + "width": 512, + "height": 512, + "pixel_sha256": "f74bd726756ed8f29d230b4c506533b708c21ecd53275092c3e54672dbb5a1cd", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-06.json", + "workflow_sha256": "c48538fbde2bdf10bb5849c8f29aaa79f9f05da8b1e53518af7b56789aa396a0", + "source_workflow_canonical_sha256": "6cc4ea59bd3d96d0811844809d534eda117290b8b705f13f0069b45cb5d2da6b" + }, + { + "id": "qwen-turbo-07", + "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 48", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 48, + "status": "succeeded", + "accepted_result": true, + "prompt": "An editorial portrait of an adult female ceramic artist in her forties working in a sunlit pottery studio. She wears a blue linen apron over a cream shirt and calmly holds a small unfinished bowl in both hands, fingers naturally curled around the rim. Her face is clear and expressive, with a few silver strands in dark curly hair. Shelves of glazed pottery remain softly out of focus behind her. Frame from the waist upward, eye-level camera, gentle window light from the left, documentary realism.", + "prompt_details": { + "word_count": 85, + "complexity": "Portrait / hands", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn editorial portrait of an adult female ceramic artist in her forties working in a sunlit pottery studio. She wears a blue linen apron over a cream shirt and calmly holds a small unfinished bowl in both hands, fingers naturally curled around the rim. Her face is clear and expressive, with a few silver strands in dark curly hair. Shelves of glazed pottery remain softly out of focus behind her. Frame from the waist upward, eye-level camera, gentle window light from the left, documentary realism." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn editorial portrait of an adult female ceramic artist in her forties working in a sunlit pottery studio. She wears a blue linen apron over a cream shirt and calmly holds a small unfinished bowl in both hands, fingers naturally curled around the rim. Her face is clear and expressive, with a few silver strands in dark curly hair. Shelves of glazed pottery remain softly out of focus behind her. Frame from the waist upward, eye-level camera, gentle window light from the left, documentary realism." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 48 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 201.952675, + "comfy_server_seconds": 199.784, + "stages_seconds": { + "node_3_observer_receive_interval": 103.42200000002049, + "node_10_observer_receive_interval": 0.015000000013969839, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 68.59399999998277, + "node_7_observer_receive_interval": 27.65600000001723, + "node_8_observer_receive_interval": 0.0779999999795109 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "11", + "12", + "2", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "New brief, retained model runtime", + "native_inter_run_gap_seconds": 13.667, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": true, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 99, + "inflight_sample_count": 98, + "minimum_ram_available_bytes": 4176404480, + "minimum_commit_available_bytes": 11358388224, + "minimum_system_drive_free_bytes": 18671104000, + "maximum_sampled_device_gpu_used_mib": 4351 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "A coherent and convincing waist-up studio portrait, with a clear adult face and a bowl held in two hands.", + "Blue apron, cream shirt, dark curly hair, pottery shelves, eye-level framing and soft window light from the left all appear. Silver hair strands are not clearly visible. Hands support the bowl from the side and underneath rather than both curling around its rim.", + "Cloth texture, skin and unfinished pottery are visibly differentiated; the bowl has a plausible unglazed look.", + "Hand shapes, wrist connections and bowl contact look broadly plausible at 512px. Some fingertips overlap or are occluded, so an exact finger-count claim is unwarranted. No obvious extra limb.", + "No text requested or visibly generated.", + "Specific silver hair detail is not clear.", + "The precise requested fingers-around-rim pose is only partly followed." + ], + "media": [ + { + "path": "media/qwen-turbo-07.png", + "mime_type": "image/png", + "size_bytes": 418100, + "sha256": "7b4feea8f758374a5544ad0e9078c8455f32f5a48e8305256ab0839f2afd9472", + "source_sha256": "ed1e6c70637edfc74eb3f051c51632087649ce6e88c705b2c4f08f1ee81d756a", + "width": 512, + "height": 512, + "pixel_sha256": "23fe51347e19cff881406c1152d713a2773a46cd9032f54cabc9dc4b5129c108", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-07.json", + "workflow_sha256": "23cb0fb14fb4cf31e9a4a67434ac21ab3e74b9858411f7e5724aaf8abbbbb755", + "source_workflow_canonical_sha256": "fc32cc9a0f7f8172edfda1400d2b8083ecbf5c86b636b5db6db024c9d20d4142" + }, + { + "id": "qwen-turbo-08", + "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 49", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 49, + "status": "succeeded", + "accepted_result": true, + "prompt": "A charming small brass robot with a round glass head stands on a wooden step stool inside an enormous old library, carefully returning a red book to a high shelf. The robot has two articulated arms and three visible fingers on each hand. A tiny blue mechanical bird perches on its shoulder. Bookshelves curve toward a domed ceiling painted with constellations; the far aisle is illuminated by warm amber lamps. Render as a detailed hand-painted storybook illustration with teal shadows, golden highlights and a clear focal point. Keep the robot, stool and book visibly separate, with no extra limbs, no captions and no written lettering.", + "prompt_details": { + "word_count": 105, + "complexity": "Stylized story scene", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA charming small brass robot with a round glass head stands on a wooden step stool inside an enormous old library, carefully returning a red book to a high shelf. The robot has two articulated arms and three visible fingers on each hand. A tiny blue mechanical bird perches on its shoulder. Bookshelves curve toward a domed ceiling painted with constellations; the far aisle is illuminated by warm amber lamps. Render as a detailed hand-painted storybook illustration with teal shadows, golden highlights and a clear focal point. Keep the robot, stool and book visibly separate, with no extra limbs, no captions and no written lettering." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA charming small brass robot with a round glass head stands on a wooden step stool inside an enormous old library, carefully returning a red book to a high shelf. The robot has two articulated arms and three visible fingers on each hand. A tiny blue mechanical bird perches on its shoulder. Bookshelves curve toward a domed ceiling painted with constellations; the far aisle is illuminated by warm amber lamps. Render as a detailed hand-painted storybook illustration with teal shadows, golden highlights and a clear focal point. Keep the robot, stool and book visibly separate, with no extra limbs, no captions and no written lettering." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 49 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 210.875124, + "comfy_server_seconds": 210.023, + "stages_seconds": { + "node_3_observer_receive_interval": 107.75, + "node_10_observer_receive_interval": 0.01600000000325963, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 72.96900000004098, + "node_7_observer_receive_interval": 29.20299999997951, + "node_8_observer_receive_interval": 0.061999999976251274 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "11", + "12", + "2", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "New brief, retained model runtime", + "native_inter_run_gap_seconds": 14.441, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": true, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 104, + "inflight_sample_count": 103, + "minimum_ram_available_bytes": 4373700608, + "minimum_commit_available_bytes": 11362136064, + "minimum_system_drive_free_bytes": 18670817280, + "maximum_sampled_device_gpu_used_mib": 4351 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "A coherent and attractive storybook-style library scene: brass robot, glass-like head, wooden stool, red book, blue shoulder bird and long illuminated aisle are all recognizable.", + "The specific main objects, articulated arms, teal/gold palette, warm far lamps and ceiling decoration are strongly represented. The exact three-visible-fingers-per-hand requirement is hard to certify because several small digits overlap or turn away.", + "Brass, a reflective glass dome, wood and book covers read distinctly within the painted style.", + "Robot, stool, book and shelves remain visibly separate. Arm joints and stool perspective are broadly coherent. Finger-count correctness cannot be established confidently at the supplied resolution.", + "No legible caption, title or written lettering is apparent.", + "Exact requested finger count remains uncertain; small jointed digits overlap.", + "Glass head is mainly reflective/opaque rather than clearly transparent, but still reads as a glass dome." + ], + "media": [ + { + "path": "media/qwen-turbo-08.png", + "mime_type": "image/png", + "size_bytes": 518714, + "sha256": "bd1bcc5079a48be9f59a4d72511eae46abb6e2aa788fe22f72bbac1cc4ba8b7a", + "source_sha256": "63abbaf5c49b2baca103bd65761fd4fed09f7e5f173406e91eaaf47d136694f2", + "width": 512, + "height": 512, + "pixel_sha256": "224125f9ca31c3c15dfa975dd61d42fa8f1703d9c012374876d2242054759658", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-08.json", + "workflow_sha256": "000bdfef44a50c9cd6275031d4ab57a84f36f216784d1f8c78d637eee60c820a", + "source_workflow_canonical_sha256": "3a9b68b7e06cf4b2a375d516a8f5d38b555d7ba5e593354edce22542adf12865" + }, + { + "id": "qwen-turbo-09", + "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 50", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 50, + "status": "succeeded", + "accepted_result": true, + "prompt": "A detailed travel photograph of a lively Mediterranean open-air produce market in early morning. In the foreground, three wooden crates hold red tomatoes, striped eggplants and yellow lemons, each type kept in its own crate. An adult vendor wearing a plain white shirt stands behind the stall, handing a small paper bag to an adult customer in a green jacket. A striped blue-and-white canopy casts cool shade over the stall. Beyond them, a narrow pedestrian street slopes toward a distant harbor with two white sailboats. Warm sunlight touches the upper walls but the foreground remains softly shaded. Compose the scene in layers with natural human proportions, believable hands, clear separation between people, baskets and produce, and rich but realistic color. Avoid artificial lens flares, watermarks and unreadable signs.", + "prompt_details": { + "word_count": 128, + "complexity": "Crowded composition", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA detailed travel photograph of a lively Mediterranean open-air produce market in early morning. In the foreground, three wooden crates hold red tomatoes, striped eggplants and yellow lemons, each type kept in its own crate. An adult vendor wearing a plain white shirt stands behind the stall, handing a small paper bag to an adult customer in a green jacket. A striped blue-and-white canopy casts cool shade over the stall. Beyond them, a narrow pedestrian street slopes toward a distant harbor with two white sailboats. Warm sunlight touches the upper walls but the foreground remains softly shaded. Compose the scene in layers with natural human proportions, believable hands, clear separation between people, baskets and produce, and rich but realistic color. Avoid artificial lens flares, watermarks and unreadable signs." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA detailed travel photograph of a lively Mediterranean open-air produce market in early morning. In the foreground, three wooden crates hold red tomatoes, striped eggplants and yellow lemons, each type kept in its own crate. An adult vendor wearing a plain white shirt stands behind the stall, handing a small paper bag to an adult customer in a green jacket. A striped blue-and-white canopy casts cool shade over the stall. Beyond them, a narrow pedestrian street slopes toward a distant harbor with two white sailboats. Warm sunlight touches the upper walls but the foreground remains softly shaded. Compose the scene in layers with natural human proportions, believable hands, clear separation between people, baskets and produce, and rich but realistic color. Avoid artificial lens flares, watermarks and unreadable signs." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 50 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 210.965741, + "comfy_server_seconds": 208.566, + "stages_seconds": { + "node_3_observer_receive_interval": 107.82800000003772, + "node_10_observer_receive_interval": 0, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 72.20299999997951, + "node_7_observer_receive_interval": 28.43800000002375, + "node_8_observer_receive_interval": 0.061999999976251274 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "11", + "12", + "2", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "New brief, retained model runtime", + "native_inter_run_gap_seconds": 13.101, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": true, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 103, + "inflight_sample_count": 102, + "minimum_ram_available_bytes": 4364185600, + "minimum_commit_available_bytes": 11356860416, + "minimum_system_drive_free_bytes": 18670198784, + "maximum_sampled_device_gpu_used_mib": 4351 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "Recognizable market transaction under a striped canopy, with three separate produce crates and a harbor beyond. Main scene layers are coherent.", + "White-shirt vendor, green-jacket customer, blue-and-white canopy, separate tomatoes/eggplants/lemons and sunlit upper wall over shaded foreground are all present. The paper bag and precise handoff are visually ambiguous. Harbor sailboats are visible but the exact requested count of two is not certifiable at this size.", + "Vegetables, wood, clothing and canopy fabric are distinguishable. Fine background detail is smeared/brushy despite a photographic brief.", + "Main adult body proportions look plausible. The central touching hands are soft and partially merged/occluded; finger anatomy and grasp cannot be verified. Background masts and architecture become muddled.", + "No prominent legible sign, watermark or caption appears.", + "Paper-bag handoff is less clear than requested.", + "Hands and distant harbor details are imprecise.", + "Exact two-sailboat count remains uncertain." + ], + "media": [ + { + "path": "media/qwen-turbo-09.png", + "mime_type": "image/png", + "size_bytes": 437788, + "sha256": "847ec6dd9be4503a94967f96ca4b91d223ffb7638749c4fd08fb318723f544e0", + "source_sha256": "5b992fa647b0461c1eb8514854a32b5f44cdd297a129a110f72b1abff6a830ef", + "width": 512, + "height": 512, + "pixel_sha256": "ffaa52b367e79221e07aeed4066dcadc2039859372848f38c9bc028d623b7f50", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-09.json", + "workflow_sha256": "049b53d4c92125dbe083563cea3d836330cfd864f19317933f71cb8794bc7db6", + "source_workflow_canonical_sha256": "e1fd339039ed8d1dde53d4a0668665f4b78470cce872ff0c3f831156539dd328" + }, + { + "id": "qwen-turbo-10", + "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 51", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 51, + "status": "succeeded", + "accepted_result": true, + "prompt": "Create a highly detailed cinematic architectural illustration of a remote mountain observatory at blue hour after a light snowfall. The building occupies the right two-thirds of the square image: a circular pale-stone base supports one weathered copper dome, whose open viewing slit points toward the upper left. A narrow exterior staircase with dark iron handrails climbs from a small lower entrance to a balcony; every step has a thin undisturbed layer of snow. One adult astronomer in a dark red winter coat stands on the balcony with a small white notebook held closed in one gloved hand, looking toward the sky. Place a warm amber lamp immediately above the entrance, creating a pool of light on the stone without illuminating the whole landscape. In the lower left foreground, a crooked pine tree and two exposed granite rocks frame a winding path that leads toward the building. Behind the observatory, show distant mountain ridges fading from slate blue into violet mist. The sky contains scattered sharp stars and a faint natural Milky Way band, with no oversized planets or fantasy moons. Balance warm interior light against cold ambient snow light, use physically plausible perspective, careful separation of the staircase rails and the figure, and visible differences between copper, stone, glass, fabric and snow. The final scene should feel quiet, inhabited and believable. Do not add titles, labels, logos, watermarks, extra people, extra domes or duplicate staircases.", + "prompt_details": { + "word_count": 235, + "complexity": "Elaborate constraints", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nCreate a highly detailed cinematic architectural illustration of a remote mountain observatory at blue hour after a light snowfall. The building occupies the right two-thirds of the square image: a circular pale-stone base supports one weathered copper dome, whose open viewing slit points toward the upper left. A narrow exterior staircase with dark iron handrails climbs from a small lower entrance to a balcony; every step has a thin undisturbed layer of snow. One adult astronomer in a dark red winter coat stands on the balcony with a small white notebook held closed in one gloved hand, looking toward the sky. Place a warm amber lamp immediately above the entrance, creating a pool of light on the stone without illuminating the whole landscape. In the lower left foreground, a crooked pine tree and two exposed granite rocks frame a winding path that leads toward the building. Behind the observatory, show distant mountain ridges fading from slate blue into violet mist. The sky contains scattered sharp stars and a faint natural Milky Way band, with no oversized planets or fantasy moons. Balance warm interior light against cold ambient snow light, use physically plausible perspective, careful separation of the staircase rails and the figure, and visible differences between copper, stone, glass, fabric and snow. The final scene should feel quiet, inhabited and believable. Do not add titles, labels, logos, watermarks, extra people, extra domes or duplicate staircases." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nCreate a highly detailed cinematic architectural illustration of a remote mountain observatory at blue hour after a light snowfall. The building occupies the right two-thirds of the square image: a circular pale-stone base supports one weathered copper dome, whose open viewing slit points toward the upper left. A narrow exterior staircase with dark iron handrails climbs from a small lower entrance to a balcony; every step has a thin undisturbed layer of snow. One adult astronomer in a dark red winter coat stands on the balcony with a small white notebook held closed in one gloved hand, looking toward the sky. Place a warm amber lamp immediately above the entrance, creating a pool of light on the stone without illuminating the whole landscape. In the lower left foreground, a crooked pine tree and two exposed granite rocks frame a winding path that leads toward the building. Behind the observatory, show distant mountain ridges fading from slate blue into violet mist. The sky contains scattered sharp stars and a faint natural Milky Way band, with no oversized planets or fantasy moons. Balance warm interior light against cold ambient snow light, use physically plausible perspective, careful separation of the staircase rails and the figure, and visible differences between copper, stone, glass, fabric and snow. The final scene should feel quiet, inhabited and believable. Do not add titles, labels, logos, watermarks, extra people, extra domes or duplicate staircases." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 51 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 226.020651, + "comfy_server_seconds": 223.135, + "stages_seconds": { + "node_3_observer_receive_interval": 115.98399999999674, + "node_10_observer_receive_interval": 0.01600000000325963, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 78.40600000001723, + "node_7_observer_receive_interval": 28.639999999955762, + "node_8_observer_receive_interval": 0.07900000002700835 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "11", + "12", + "2", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "New brief, retained model runtime", + "native_inter_run_gap_seconds": 10.771, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": true, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 111, + "inflight_sample_count": 110, + "minimum_ram_available_bytes": 4460896256, + "minimum_commit_available_bytes": 11257696256, + "minimum_system_drive_free_bytes": 18669465600, + "maximum_sampled_device_gpu_used_mib": 4351 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "A coherent quiet blue-hour observatory scene with one large dome, one red-coated figure, one snow-covered staircase, foreground pine/rocks and distant layered mountains.", + "Building placement, open slit toward the upper left, red coat, small light-colored notebook, amber entrance lamp, snow on steps and cold/warm lighting contrast are represented. The figure appears to attend to the notebook rather than clearly look at the sky. Requested stars and faint Milky Way are not visibly present. The exact two-rock count is not cleanly met or certifiable among overlapping foreground boulders.", + "Dome reads as dark weathered metal, base as smooth pale masonry/stucco, and snow, coat and rails are distinct. Fine stone material detail is less explicit than the brief.", + "Overall building, balcony and staircase perspective are plausible; thin rail details and figure-to-railing separation are imperfect but readable. No obvious extra person, dome or duplicate staircase.", + "No title, label, logo or watermark is apparent.", + "Stars and Milky Way are absent from the visibly plain sky.", + "Figure does not clearly follow the requested skyward gaze.", + "Exact rock-count and some fine architectural/material details are weaker than the broad composition." + ], + "media": [ + { + "path": "media/qwen-turbo-10.png", + "mime_type": "image/png", + "size_bytes": 402891, + "sha256": "7bc1033039962507ee2aa153ebfacaa4cdd1f77e93e152bd7241ebae686b85bc", + "source_sha256": "f051ec1ab055f23d3fd6513e3428d807734a846f5e51881403cf419ccb33c161", + "width": 512, + "height": 512, + "pixel_sha256": "9672ee9777924c1b99ccff8952362c0119f99a87bb1e22c7bc82a24b579ccb6f", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-10.json", + "workflow_sha256": "11fe5261e6abff03fd6a82be45d93b3827c421e3cc3e684f03b72df24694e0b0", + "source_workflow_canonical_sha256": "7120b879556ee04ac859a6126f8d1ab24771a28bf208d58fff10b120bf50e68a" } ], "failures": [ { "id": "wan-failure-01", - "label": "Wan2.1 T2V 1.3B \u00b7 cancelled", + "label": "Wan2.1 T2V 1.3B · cancelled", "group": "wan21-small-v1", "kind": "video", "model_id": "wan", @@ -24076,7 +26937,7 @@ }, { "id": "ltx-failure-01", - "label": "LTX-Video 2B 0.9.8 distilled Q8_0 \u00b7 cancelled", + "label": "LTX-Video 2B 0.9.8 distilled Q8_0 · cancelled", "group": "ltxv2b-q8-text-v1", "kind": "video", "model_id": "ltx", @@ -24130,7 +26991,7 @@ }, { "id": "ltx-failure-02", - "label": "LTX-Video 2B 0.9.8 distilled Q8_0 \u00b7 cancelled", + "label": "LTX-Video 2B 0.9.8 distilled Q8_0 · cancelled", "group": "ltxv2b-q8-text-v2", "kind": "video", "model_id": "ltx", @@ -24184,7 +27045,7 @@ }, { "id": "cog-failure-01", - "label": "CogVideoX 2B local Q8_0 \u00b7 failed", + "label": "CogVideoX 2B local Q8_0 · failed", "group": "cogvideox2b-q8-text-v1", "kind": "video", "model_id": "cog", @@ -24233,7 +27094,7 @@ }, { "id": "h3-legacy-failure", - "label": "MiniMax H3 \u00b7 legacy loader failure", + "label": "MiniMax H3 · legacy loader failure", "group": "h3-legacy", "kind": "video", "model_id": "h3", @@ -24288,7 +27149,7 @@ }, { "id": "h3-dynamic-failure", - "label": "MiniMax H3 \u00b7 dynamic cancelled", + "label": "MiniMax H3 · dynamic cancelled", "group": "h3-dynamic", "kind": "video", "model_id": "h3", @@ -24343,7 +27204,7 @@ }, { "id": "yue2-cpu-cancelled", - "label": "YuE2 \u00b7 cancelled CPU semanticgeneration probe", + "label": "YuE2 · cancelled CPU semanticgeneration probe", "group": "yue2-cpu-probe", "kind": "audio", "model_id": "yue2", @@ -24365,7 +27226,7 @@ "factory_trace_seconds": 698.101266, "last_observer_elapsed_seconds": 834.844, "stages_seconds": {}, - "limits": "Closedcancelledattempttrace(11:35:19.587812\u201311:46:57.689078UTC), not completedmusicgenerationtime. Last native observer834.844s continued aftertraceclosure." + "limits": "Closedcancelledattempttrace(11:35:19.587812–11:46:57.689078UTC), not completedmusicgenerationtime. Last native observer834.844s continued aftertraceclosure." }, "cache": { "second_request_submitted": false @@ -24408,10 +27269,11 @@ ], "findings": [ "Quantized generator weights reduce storage/residency but do not guarantee fasterFP32 sampling. WanQ8 sampler remainedabout98secondsperstep.", - "LTX0.9.8distilled completed inroughly2\u20133minutes; this was onebasepass with8steps, nottheofficialmultiscale highqualitypipeline.", + "LTX0.9.8distilled completed inroughly2–3minutes; this was onebasepass with8steps, nottheofficialmultiscale highqualitypipeline.", "Native parameterCUDA assertions are notproof that everyallocation stayed inphysicalVRAM. Cog decoderTorchreserved memory exceededphysical6GB.", "TwoCdrive storageguard stops were real no-output failures. Movingmodel/tool/temp/cache storage and addingDdrive paging preceded four successfulLTXtrials. This is anoperationalchange, notisolatedcausalproof.", - "OldCog colorednoise was producedbyanincorrect custombridge; correctedpositions andrawVP referenceDDIM removednoise insampledframes but didnotestablish qualityusablegeneration." + "OldCog colorednoise was producedbyanincorrect custombridge; correctedpositions andrawVP referenceDDIM removednoise insampledframes but didnotestablish qualityusablegeneration.", + "Across 50 FLUX second-seed requests, positive and negative conditioning were reused while sampling ran again and outputs changed. Excluding initial loading, native medians were 70.05 s new brief and 38.49 s next seed. The Qwen Turbo pair took 320.246 s then 95.234 s, combining startup/model and encoding reuse; only two seed variants were measured, not five." ], "diagnostics": [ { @@ -24465,6 +27327,12 @@ "component": "correctedCogstageinstrumentation", "source_sha256": "056628db7d0f4def94d00ce6d10c7211ea51fe14fac1a025ab7646d44b0586cc", "scope": "Hash recorded by the postpair saved-evidence audit." + }, + { + "component": "Qwen Image 2.1 Turbo operator-approved template", + "source_sha256": "5dd58f0121dabb6cef5e588c3bdd39cfe51296938a26bab742251f1ebf4cbac1", + "scope": "Template configuration hash; exact public per-output graphs and their source canonical hashes are separate fields.", + "upstream_workflow": "https://github.com/Comfy-Org/workflow_templates/blob/0cf483d1dcb548b8b8d3f38ddf0ff41c89896cdb/templates/image_qwen_image_2_1_turbo_t2i.json" } ] } diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/manifest.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/manifest.json index b72eebf..233f630 100644 --- a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/manifest.json +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/manifest.json @@ -4,8 +4,8 @@ "source": "Packet Foundry retained evidence", "publication_url": "https://benchmarks.weeyuga.com/benchmarks/predator-media-20261021.html", "article_url": "https://benchmarks.weeyuga.com/findings/images-video-music-on-6gb.html", - "recorded_attempts": 130, - "retained_outputs": 123, + "recorded_attempts": 140, + "retained_outputs": 133, "private_operator_logs_excluded": true, "generated_media_terms": "See README.md and dataset.json model-source/terms links; generated media are excluded from any blanket archive data license." } diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-01.png b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-01.png new file mode 100644 index 0000000..2164524 Binary files /dev/null and b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-01.png differ diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-02.png b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-02.png new file mode 100644 index 0000000..4878a4c Binary files /dev/null and b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-02.png differ diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-03.png b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-03.png new file mode 100644 index 0000000..72d1979 Binary files /dev/null and b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-03.png differ diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-04.png b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-04.png new file mode 100644 index 0000000..6468447 Binary files /dev/null and b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-04.png differ diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-05.png b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-05.png new file mode 100644 index 0000000..623b016 Binary files /dev/null and b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-05.png differ diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-06.png b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-06.png new file mode 100644 index 0000000..c71c882 Binary files /dev/null and b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-06.png differ diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-07.png b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-07.png new file mode 100644 index 0000000..3f60e10 Binary files /dev/null and b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-07.png differ diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-08.png b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-08.png new file mode 100644 index 0000000..1e7d47c Binary files /dev/null and b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-08.png differ diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-09.png b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-09.png new file mode 100644 index 0000000..1df0e48 Binary files /dev/null and b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-09.png differ diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-10.png b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-10.png new file mode 100644 index 0000000..326ecaa Binary files /dev/null and b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/media/qwen-turbo-10.png differ diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/metadata.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/metadata.json index e4203c1..ebfe466 100644 --- a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/metadata.json +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/metadata.json @@ -25,19 +25,20 @@ } }, "coverage": { - "retained_output_files": 123, - "accepted_outputs": 122, - "images": 110, - "accepted_images": 109, + "retained_output_files": 133, + "accepted_outputs": 132, + "images": 120, + "accepted_images": 119, "image_complexity_suite": 100, "videos": 11, "music": 2, "contact_sheets": 11, "failed_delivery_outputs_with_media": 1, "failed_or_cancelled_submitted_jobs_without_media": 7, - "unique_original_media_sha256": 121, + "unique_original_media_sha256": 131, "diagnostic_fixture_exclusion": "Two synthetic codec preflight MP4s are fixtures, not model outputs; excluded from gallery and generated-output counts.", - "duplicate_media_note": "Two pairs of early accepted image files have byte-identical source hashes. Each delivery is retained;123files does not mean123unique contents." + "duplicate_media_note": "Two pairs of early accepted image files have byte-identical source hashes. Every delivery is retained; file count and unique original byte sequences are distinct.", + "qwen_turbo_ten_image_trial": 10 }, "methodology": [ "Same model, template and fixed generation settings; record any intentional prompt change and seed42 then43. A changed prompt forces fresh text encoding; a changed seed forces sampling.", @@ -180,7 +181,1376 @@ ] }, "quality": "All100 complexity-image thumbnails showed recognizable scenes; some lettering was malformed. The user described both YuE2 samples as usable music. Video results ranged from nearly static fox scenes and colored noise to recognizable scenes with poor framing and prompt adherence.", - "failed_jobs_preserved": 7 + "failed_jobs_preserved": 7, + "campaign_totals": { + "recorded_attempts": 140, + "retained_output_files": 133, + "images": 120, + "videos": 11, + "music": 2, + "accepted_outputs": 132, + "no_output_failures": 7, + "failed_delivery_outputs": 1 + }, + "original_overnight_scope": { + "recorded_attempts": 130, + "retained_output_files": 123, + "images": 110, + "image_complexity_suite": 100, + "videos": 11, + "music": 2 + }, + "qwen_turbo_followup_scope": { + "recorded_attempts": 10, + "retained_output_files": 10, + "images": 10, + "distinct_briefs": 9 + }, + "seed_variant_cache": { + "schema_version": "seed-variant-cache-1", + "scope": "Read-only reanalysis of the existing FLUX 100 images and completed ten-image Qwen Turbo trial. No additional generation for this analysis.", + "flux50_seed_pairs": { + "model": "Abiray/flux-3-image-klein-sd-GGUF:Q4_K_M", + "pairs": 50, + "observed_seed_variants_per_brief": 2, + "five_seed_variants_tested": false, + "configuration": { + "width": 512, + "height": 512, + "steps": 4, + "cfg": 1, + "generator_quant": "Q4_K_M", + "compute": "FP32", + "cpu_vae": true, + "lowvram": true + }, + "cache_counts": { + "second_positive_encoder_cached": 50, + "second_negative_encoder_cached": 50, + "second_sampler_cached": 0, + "same_actual_runtime_pair": 50, + "identical_graph_except_seed": 50, + "second_probe_is_first_probe": 50, + "distinct_output_hashes_per_pair": 50, + "first_positive_encoder_cached": 0 + }, + "all_pairs": { + "first_native_seconds": { + "n": 50, + "min": 64.204, + "median": 71.671, + "mean": 79.30762, + "max": 241.398, + "total": 3965.3810000000003 + }, + "second_native_seconds": { + "n": 50, + "min": 36.576, + "median": 38.397999999999996, + "mean": 39.505, + "max": 44.558, + "total": 1975.2499999999998 + }, + "native_seconds_saved": { + "n": 50, + "min": 27.309, + "median": 33.367000000000004, + "mean": 39.80262, + "max": 204.344, + "total": 1990.1310000000003 + }, + "native_speedup": { + "n": 50, + "min": 1.7400193433990652, + "median": 1.8708729216784072, + "mean": 2.002159338297874, + "max": 6.514762238894586, + "total": 100.10796691489372 + }, + "first_trace_seconds": { + "n": 50, + "min": 66.467404, + "median": 74.03037850000001, + "mean": 81.160127, + "max": 242.338818, + "total": 4058.0063499999997 + }, + "second_trace_seconds": { + "n": 50, + "min": 39.273643, + "median": 39.397068000000004, + "mean": 41.17113458, + "max": 45.413099, + "total": 2058.556729 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 50, + "min": 10.422, + "median": 22.38, + "mean": 21.21556, + "max": 31.712, + "total": 1060.778 + }, + "first_positive_encoding_seconds": { + "n": 50, + "min": 27.51600000000326, + "median": 31.68800000002375, + "mean": 36.278460000006014, + "max": 56.32800000003772, + "total": 1813.9230000003008 + } + }, + "exclude_initial_load_pair": { + "first_native_seconds": { + "n": 49, + "min": 64.204, + "median": 70.046, + "mean": 75.99965306122449, + "max": 101.131, + "total": 3723.983 + }, + "second_native_seconds": { + "n": 49, + "min": 36.576, + "median": 38.491, + "mean": 39.555020408163266, + "max": 44.558, + "total": 1938.196 + }, + "native_seconds_saved": { + "n": 49, + "min": 27.309, + "median": 32.125, + "mean": 36.44463265306123, + "max": 56.761, + "total": 1785.7870000000003 + }, + "native_speedup": { + "n": 49, + "min": 1.7400193433990652, + "median": 1.8471559294322406, + "mean": 1.9100654015510024, + "max": 2.279265269326121, + "total": 93.59320467599913 + }, + "first_trace_seconds": { + "n": 49, + "min": 66.467404, + "median": 72.559234, + "mean": 77.87076595918367, + "max": 102.832423, + "total": 3815.6675319999995 + }, + "second_trace_seconds": { + "n": 49, + "min": 39.273643, + "median": 39.3951, + "mean": 41.20691226530612, + "max": 45.413099, + "total": 2019.138701 + }, + "first_positive_encoding_seconds": { + "n": 49, + "min": 27.51600000000326, + "median": 31.985000000044238, + "mean": 36.452836734700085, + "max": 56.32800000003772, + "total": 1786.189000000304 + } + }, + "by_tier": { + "1": { + "first_native_seconds": { + "n": 10, + "min": 64.57, + "median": 64.88550000000001, + "mean": 82.5835, + "max": 241.398, + "total": 825.835 + }, + "second_native_seconds": { + "n": 10, + "min": 36.576, + "median": 36.931, + "mean": 36.9348, + "max": 37.258, + "total": 369.348 + }, + "native_seconds_saved": { + "n": 10, + "min": 27.545, + "median": 28.106, + "mean": 45.6487, + "max": 204.344, + "total": 456.487 + }, + "first_trace_seconds": { + "n": 10, + "min": 66.467404, + "median": 66.59116850000001, + "mean": 84.1501359, + "max": 242.338818, + "total": 841.5013589999999 + }, + "second_trace_seconds": { + "n": 10, + "min": 39.273643, + "median": 39.350615, + "mean": 39.362302899999996, + "max": 39.476116, + "total": 393.623029 + }, + "first_positive_encoding_seconds": { + "n": 10, + "min": 27.73399999999674, + "median": 27.859499999991385, + "mean": 27.93749999999418, + "max": 28.485000000044238, + "total": 279.3749999999418 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 10, + "min": 10.422, + "median": 26.314999999999998, + "mean": 21.0136, + "max": 27.701, + "total": 210.136 + } + }, + "2": { + "first_native_seconds": { + "n": 10, + "min": 64.204, + "median": 64.979, + "mean": 66.5586, + "max": 70.046, + "total": 665.5860000000001 + }, + "second_native_seconds": { + "n": 10, + "min": 36.682, + "median": 37.0485, + "mean": 37.4005, + "max": 38.292, + "total": 374.005 + }, + "native_seconds_saved": { + "n": 10, + "min": 27.309, + "median": 28.152, + "mean": 29.1581, + "max": 32.125, + "total": 291.581 + }, + "first_trace_seconds": { + "n": 10, + "min": 66.470952, + "median": 66.60241149999999, + "mean": 68.0339363, + "max": 72.466618, + "total": 680.3393630000002 + }, + "second_trace_seconds": { + "n": 10, + "min": 39.288523, + "median": 39.369251000000006, + "mean": 39.3625179, + "max": 39.4423, + "total": 393.625179 + }, + "first_positive_encoding_seconds": { + "n": 10, + "min": 27.51600000000326, + "median": 28.094000000011874, + "mean": 29.17840000001015, + "max": 31.985000000044238, + "total": 291.7840000001015 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 10, + "min": 11.778, + "median": 24.357, + "mean": 23.0208, + "max": 31.712, + "total": 230.20799999999997 + } + }, + "3": { + "first_native_seconds": { + "n": 10, + "min": 69.077, + "median": 69.54599999999999, + "mean": 70.9772, + "max": 73.464, + "total": 709.7719999999999 + }, + "second_native_seconds": { + "n": 10, + "min": 38.008, + "median": 38.397999999999996, + "mean": 38.4107, + "max": 38.705, + "total": 384.107 + }, + "native_seconds_saved": { + "n": 10, + "min": 30.839, + "median": 31.393, + "mean": 32.5665, + "max": 34.852, + "total": 325.66499999999996 + }, + "first_trace_seconds": { + "n": 10, + "min": 69.48063, + "median": 72.549484, + "mean": 72.8756681, + "max": 75.63889, + "total": 728.7566810000001 + }, + "second_trace_seconds": { + "n": 10, + "min": 39.287177, + "median": 39.335657499999996, + "mean": 39.3440559, + "max": 39.400407, + "total": 393.440559 + }, + "first_positive_encoding_seconds": { + "n": 10, + "min": 30.90600000001723, + "median": 31.38300000000163, + "mean": 32.56720000000787, + "max": 34.78099999995902, + "total": 325.6720000000787 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 10, + "min": 18.393, + "median": 21.78, + "mean": 21.6223, + "max": 26.858, + "total": 216.223 + } + }, + "4": { + "first_native_seconds": { + "n": 10, + "min": 77.877, + "median": 78.6135, + "mean": 78.4776, + "max": 78.736, + "total": 784.7760000000001 + }, + "second_native_seconds": { + "n": 10, + "min": 40.423, + "median": 40.4815, + "mean": 40.5054, + "max": 40.68, + "total": 405.05400000000003 + }, + "native_seconds_saved": { + "n": 10, + "min": 37.423, + "median": 37.981, + "mean": 37.9722, + "max": 38.269, + "total": 379.72200000000004 + }, + "first_trace_seconds": { + "n": 10, + "min": 78.600529, + "median": 81.565219, + "mean": 80.7096113, + "max": 81.696329, + "total": 807.0961130000001 + }, + "second_trace_seconds": { + "n": 10, + "min": 42.349654, + "median": 42.403067, + "mean": 42.4041422, + "max": 42.532345, + "total": 424.041422 + }, + "first_positive_encoding_seconds": { + "n": 10, + "min": 37.65600000001723, + "median": 38.10950000002049, + "mean": 38.05140000000247, + "max": 38.32799999997951, + "total": 380.5140000000247 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 10, + "min": 13.12, + "median": 14.9955, + "mean": 15.7422, + "max": 21.868, + "total": 157.42200000000003 + } + }, + "5": { + "first_native_seconds": { + "n": 10, + "min": 96.483, + "median": 97.2595, + "mean": 97.94120000000001, + "max": 101.131, + "total": 979.4119999999999 + }, + "second_native_seconds": { + "n": 10, + "min": 44.121, + "median": 44.253, + "mean": 44.2736, + "max": 44.558, + "total": 442.73600000000005 + }, + "native_seconds_saved": { + "n": 10, + "min": 52.362, + "median": 53.0135, + "mean": 53.6676, + "max": 56.761, + "total": 536.676 + }, + "first_trace_seconds": { + "n": 10, + "min": 96.80878, + "median": 99.68121, + "mean": 100.0312834, + "max": 102.832423, + "total": 1000.3128339999998 + }, + "second_trace_seconds": { + "n": 10, + "min": 45.341639, + "median": 45.3771555, + "mean": 45.382654, + "max": 45.413099, + "total": 453.82654 + }, + "first_positive_encoding_seconds": { + "n": 10, + "min": 52.40600000001723, + "median": 53.0625, + "mean": 53.65780000001541, + "max": 56.32800000003772, + "total": 536.5780000001541 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 10, + "min": 12.666, + "median": 27.817, + "mean": 24.6789, + "max": 29.406, + "total": 246.78900000000004 + } + } + }, + "pair_evidence": [ + { + "first_record": "pf-img100-01-t1-s42", + "second_record": "pf-img100-01-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 241.398, + "second_native_seconds": 37.054, + "native_inter_run_gap_seconds": 14.139, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-02-t1-s42", + "second_record": "pf-img100-02-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 65.35, + "second_native_seconds": 37.258, + "native_inter_run_gap_seconds": 11.281, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-03-t1-s42", + "second_record": "pf-img100-03-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.767, + "second_native_seconds": 37.222, + "native_inter_run_gap_seconds": 26.604, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-04-t1-s42", + "second_record": "pf-img100-04-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.679, + "second_native_seconds": 37.016, + "native_inter_run_gap_seconds": 26.543, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-05-t1-s42", + "second_record": "pf-img100-05-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.57, + "second_native_seconds": 36.908, + "native_inter_run_gap_seconds": 10.422, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-06-t1-s42", + "second_record": "pf-img100-06-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.943, + "second_native_seconds": 36.896, + "native_inter_run_gap_seconds": 27.239, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-08-t1-s42", + "second_record": "pf-img100-08-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.791, + "second_native_seconds": 36.576, + "native_inter_run_gap_seconds": 13.335, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-09-t1-s42", + "second_record": "pf-img100-09-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.828, + "second_native_seconds": 36.708, + "native_inter_run_gap_seconds": 27.701, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-16-t1-s42", + "second_record": "pf-img100-16-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 65.354, + "second_native_seconds": 36.954, + "native_inter_run_gap_seconds": 26.087, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-20-t1-s42", + "second_record": "pf-img100-20-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 65.155, + "second_native_seconds": 36.756, + "native_inter_run_gap_seconds": 26.785, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-01-t2-s42", + "second_record": "pf-img100-01-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.204, + "second_native_seconds": 36.895, + "native_inter_run_gap_seconds": 11.778, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-02-t2-s42", + "second_record": "pf-img100-02-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.965, + "second_native_seconds": 36.897, + "native_inter_run_gap_seconds": 31.712, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-03-t2-s42", + "second_record": "pf-img100-03-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.918, + "second_native_seconds": 36.682, + "native_inter_run_gap_seconds": 26.206, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-04-t2-s42", + "second_record": "pf-img100-04-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.993, + "second_native_seconds": 37.056, + "native_inter_run_gap_seconds": 27.134, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-05-t2-s42", + "second_record": "pf-img100-05-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.762, + "second_native_seconds": 37.041, + "native_inter_run_gap_seconds": 11.901, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-06-t2-s42", + "second_record": "pf-img100-06-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 70.046, + "second_native_seconds": 37.921, + "native_inter_run_gap_seconds": 22.36, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-08-t2-s42", + "second_record": "pf-img100-08-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.192, + "second_native_seconds": 38.292, + "native_inter_run_gap_seconds": 22.4, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-09-t2-s42", + "second_record": "pf-img100-09-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.379, + "second_native_seconds": 36.876, + "native_inter_run_gap_seconds": 28.003, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-16-t2-s42", + "second_record": "pf-img100-16-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 68.971, + "second_native_seconds": 38.169, + "native_inter_run_gap_seconds": 24.357, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-20-t2-s42", + "second_record": "pf-img100-20-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.156, + "second_native_seconds": 38.176, + "native_inter_run_gap_seconds": 24.357, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-01-t3-s42", + "second_record": "pf-img100-01-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.601, + "second_native_seconds": 38.305, + "native_inter_run_gap_seconds": 23.193, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-02-t3-s42", + "second_record": "pf-img100-02-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.491, + "second_native_seconds": 38.008, + "native_inter_run_gap_seconds": 22.005, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-03-t3-s42", + "second_record": "pf-img100-03-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.451, + "second_native_seconds": 38.148, + "native_inter_run_gap_seconds": 22.876, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-04-t3-s42", + "second_record": "pf-img100-04-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.125, + "second_native_seconds": 38.262, + "native_inter_run_gap_seconds": 26.858, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-05-t3-s42", + "second_record": "pf-img100-05-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.388, + "second_native_seconds": 38.491, + "native_inter_run_gap_seconds": 24.166, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-06-t3-s42", + "second_record": "pf-img100-06-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 73.296, + "second_native_seconds": 38.687, + "native_inter_run_gap_seconds": 18.854, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-08-t3-s42", + "second_record": "pf-img100-08-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 73.437, + "second_native_seconds": 38.705, + "native_inter_run_gap_seconds": 18.393, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-09-t3-s42", + "second_record": "pf-img100-09-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.077, + "second_native_seconds": 38.238, + "native_inter_run_gap_seconds": 21.555, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-16-t3-s42", + "second_record": "pf-img100-16-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 73.442, + "second_native_seconds": 38.651, + "native_inter_run_gap_seconds": 19.37, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-20-t3-s42", + "second_record": "pf-img100-20-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 73.464, + "second_native_seconds": 38.612, + "native_inter_run_gap_seconds": 18.953, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-01-t4-s42", + "second_record": "pf-img100-01-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.215, + "second_native_seconds": 40.46, + "native_inter_run_gap_seconds": 21.868, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-02-t4-s42", + "second_record": "pf-img100-02-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.617, + "second_native_seconds": 40.423, + "native_inter_run_gap_seconds": 13.502, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-03-t4-s42", + "second_record": "pf-img100-03-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.656, + "second_native_seconds": 40.474, + "native_inter_run_gap_seconds": 14.668, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-04-t4-s42", + "second_record": "pf-img100-04-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.617, + "second_native_seconds": 40.612, + "native_inter_run_gap_seconds": 15.534, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-05-t4-s42", + "second_record": "pf-img100-05-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.467, + "second_native_seconds": 40.51, + "native_inter_run_gap_seconds": 15.323, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-06-t4-s42", + "second_record": "pf-img100-06-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 77.877, + "second_native_seconds": 40.454, + "native_inter_run_gap_seconds": 18.954, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-08-t4-s42", + "second_record": "pf-img100-08-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.727, + "second_native_seconds": 40.458, + "native_inter_run_gap_seconds": 15.354, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-09-t4-s42", + "second_record": "pf-img100-09-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.254, + "second_native_seconds": 40.489, + "native_inter_run_gap_seconds": 14.643, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-16-t4-s42", + "second_record": "pf-img100-16-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.61, + "second_native_seconds": 40.68, + "native_inter_run_gap_seconds": 13.12, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-20-t4-s42", + "second_record": "pf-img100-20-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.736, + "second_native_seconds": 40.494, + "native_inter_run_gap_seconds": 14.456, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-01-t5-s42", + "second_record": "pf-img100-01-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.733, + "second_native_seconds": 44.245, + "native_inter_run_gap_seconds": 27.851, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-02-t5-s42", + "second_record": "pf-img100-02-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.058, + "second_native_seconds": 44.319, + "native_inter_run_gap_seconds": 13.702, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-03-t5-s42", + "second_record": "pf-img100-03-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.156, + "second_native_seconds": 44.279, + "native_inter_run_gap_seconds": 28.917, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-04-t5-s42", + "second_record": "pf-img100-04-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 96.93, + "second_native_seconds": 44.162, + "native_inter_run_gap_seconds": 28.792, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-05-t5-s42", + "second_record": "pf-img100-05-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.26, + "second_native_seconds": 44.231, + "native_inter_run_gap_seconds": 28.36, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-06-t5-s42", + "second_record": "pf-img100-06-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 101.049, + "second_native_seconds": 44.558, + "native_inter_run_gap_seconds": 24.123, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-08-t5-s42", + "second_record": "pf-img100-08-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.259, + "second_native_seconds": 44.261, + "native_inter_run_gap_seconds": 29.406, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-09-t5-s42", + "second_record": "pf-img100-09-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.353, + "second_native_seconds": 44.19, + "native_inter_run_gap_seconds": 27.783, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-16-t5-s42", + "second_record": "pf-img100-16-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 96.483, + "second_native_seconds": 44.121, + "native_inter_run_gap_seconds": 12.666, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-20-t5-s42", + "second_record": "pf-img100-20-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 101.131, + "second_native_seconds": 44.37, + "native_inter_run_gap_seconds": 25.189, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + } + ], + "interpretation": "All 50 second requests reused positive and negative conditioning. Sampling was not cached, and every pair produced distinct outputs. The initial-load pair is excluded from the main steady-runtime comparison. No five-variant group was tested.", + "limits": [ + "Same42/43 seed ordering; no randomized cache-disabled control.", + "Node intervals include preparation/event delivery, not isolated kernels.", + "Observed cache survives the measured gaps but is local to one retained Comfy runtime; no persistence guarantee across model switches, restarts, factory changes or eviction." + ] + }, + "qwen_turbo_ten": { + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "generated_outputs": 10, + "first_native_start_utc": "2026-10-21T08:41:18.067000+00:00", + "distinct_briefs": 9, + "measured_seed_variants_in_first_brief": 2, + "five_seed_variants_tested": false, + "initial_native_seconds": 320.246, + "same_brief_variant_native_seconds": 95.234, + "native_inter_run_gap_seconds": 34.322, + "same_runtime_retained": true, + "only_graph_difference": [ + "/9/inputs/noise_seed" + ], + "second_encoder_cached": true, + "sampling_and_decoding_repeated": true, + "new_briefs_in_retained_runtime": { + "requests": 8, + "native_seconds": [ + 194.865, + 196.417, + 199.64, + 199.412, + 199.784, + 210.023, + 208.566, + 223.135 + ], + "minimum_seconds": 194.865, + "median_seconds": 199.712, + "maximum_seconds": 223.135, + "sum_seconds": 1631.842 + }, + "limits": [ + "The initial request includes loading plus text encoding, so the first/second delta is a combined startup/model/encoding benefit.", + "Only two same-brief seed outputs were measured; do not present five-variant performance as measured.", + "Fresh-prompt latency is a separate workload from cached same-brief variation." + ] + }, + "production_guidance": "Keep the complete composed brief/materials/settings identical and vary only the seed in consecutive requests to the same factory/profile/Comfy runtime. Cache state is local to that runtime and can be lost after restarts, model switches or eviction.", + "five_variant_group_measured": false + } }, "limits": [ "A valid media container, successful relay, or LangSmith trace is not a quality score.", @@ -189,5 +1559,5 @@ "No model weights or private credentials are redistributed. Source hashes identify retained originals; public metadata-stripped copies have separate hashes." ], "archive_run_id": "3cc53336-2555-5e90-99c3-7b94dcb40acc", - "dataset_sha256": "e67fd94071a63873c156e97e4acd3230b38e4928dfc7c01f01bf79272988c451" + "dataset_sha256": "99a1e345996ce2ff8e47c4c50bf07f90b0f676081a09a1e3e9331f2bbf7be251" } diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/qwen-turbo-workflow-NOTICE.md b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/qwen-turbo-workflow-NOTICE.md new file mode 100644 index 0000000..7b7e9d5 --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/qwen-turbo-workflow-NOTICE.md @@ -0,0 +1,31 @@ +# Qwen Image 2.1 Turbo workflow provenance + +The operator-approved API graph adapts the stock ComfyUI Qwen2.1 workflow to a GGUF generator, CPU INT8 ConvRot text encoder, the already installed VAE and the Turbo checkpoint’s exact ManualSigmas schedule. It replaces the older KSampler with RandomNoise, CFGGuider, KSamplerSelect, ManualSigmas and SamplerCustomAdvanced, retaining one image per request. + +Upstream template: https://github.com/Comfy-Org/workflow_templates/blob/0cf483d1dcb548b8b8d3f38ddf0ff41c89896cdb/templates/image_qwen_image_2_1_turbo_t2i.json + +Model: https://huggingface.co/unsloth/Qwen-Image-2.1-Turbo-GGUF/tree/5d1f9f759ea877ae556cda88dbec958b62ad3809 + +This MIT notice covers the adapted workflow material. Model weights and outputs retain separate upstream terms; it does not replace the Qwen research license. + +MIT License + +Copyright (c) 2023-present Comfy Org + +Permission is hereby granted, free of charge, to any person obtaining a copy +of this software and associated documentation files (the "Software"), to deal +in the Software without restriction, including without limitation the rights +to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +copies of the Software, and to permit persons to whom the Software is +furnished to do so, subject to the following conditions: + +The above copyright notice and this permission notice shall be included in all +copies or substantial portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE +SOFTWARE. diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-01.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-01.json new file mode 100644 index 0000000..e5ec34d --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-01.json @@ -0,0 +1,143 @@ +{ + "id": "qwen-turbo-01", + "label": "Qwen Image 2.1 Turbo Q4_K_M · Initial request including model loading · seed 42", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 42, + "status": "succeeded", + "accepted_result": true, + "prompt": "A small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph.", + "prompt_details": { + "word_count": 19, + "complexity": "Very short / retained runtime pair", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 42 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 334.589337, + "comfy_server_seconds": 320.246, + "stages_seconds": { + "node_4_observer_receive_interval": 4.436999999976251, + "node_2_observer_receive_interval": 6.48399999999674, + "node_3_observer_receive_interval": 104.32900000002701, + "node_12_observer_receive_interval": 0.031000000017229468, + "node_11_observer_receive_interval": 0.014999999955762178, + "node_1_observer_receive_interval": 1.6570000000065193, + "node_5_observer_receive_interval": 0.10899999999674037, + "node_10_observer_receive_interval": 0, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 159.85899999999674, + "node_7_observer_receive_interval": 40.75, + "node_8_observer_receive_interval": 2.047000000020489 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [], + "same_runtime_retained": null, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "Initial request including model loading", + "native_inter_run_gap_seconds": null, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": false, + "cold_start_scope": "The first request includes initial/model work; OS disk-cache state was not isolated." + } + }, + "resources": { + "sampled_summary": { + "sample_count": 162, + "inflight_sample_count": 161, + "minimum_ram_available_bytes": 1912422400, + "minimum_commit_available_bytes": 11296174080, + "minimum_system_drive_free_bytes": 18670813184, + "maximum_sampled_device_gpu_used_mib": 3983 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "A coherent fox figurine standing on a pale wooden desk against a cream background; recognizably matches the simple scene.", + "Orange fox, desk, cream background and soft studio light are present.", + "Fine fur-like ridges dominate the body; ceramic material is not convincingly conveyed.", + "Legs, ears and tail are largely plausible for a stylized figurine. Face is slightly toy-like.", + "No text requested or visibly generated.", + "Ceramic material appears more like textured wood, fur or a molded toy." + ], + "media": [ + { + "path": "media/qwen-turbo-01.png", + "mime_type": "image/png", + "size_bytes": 261046, + "sha256": "56129ce43e04bb2aca5849e23b9d6a0c9adfdee3047a9329f1fcc360c2d9eb2e", + "source_sha256": "c4a363cb2d3fbcb5d51b02226bdb57facc1963d98cac306cefb3ad252700b3d0", + "width": 512, + "height": 512, + "pixel_sha256": "5df320c6bdbdad6dc8dbc07af73f2dc513a55deecad9b2f49b659e85ba5ba2f4", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-01.json", + "workflow_sha256": "4369239d25533a14c3b5bb3a15d663b4049d3285e9804cc52483e7ffccaf4852", + "source_workflow_canonical_sha256": "1ff73d5851a428f07381767c01cfdf4675c6c0b6f37e9df58ff08a9c74de77bb" +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-02.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-02.json new file mode 100644 index 0000000..7848b7a --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-02.json @@ -0,0 +1,145 @@ +{ + "id": "qwen-turbo-02", + "label": "Qwen Image 2.1 Turbo Q4_K_M · Same brief, next seed · seed 43", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 43, + "status": "succeeded", + "accepted_result": true, + "prompt": "A small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph.", + "prompt_details": { + "word_count": 19, + "complexity": "Very short / retained runtime pair", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 43 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 96.670077, + "comfy_server_seconds": 95.234, + "stages_seconds": { + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 67.14100000000326, + "node_7_observer_receive_interval": 27.82800000003772, + "node_8_observer_receive_interval": 0.06299999996554106 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "10", + "11", + "12", + "2", + "3", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": true, + "context": { + "workload": "Same brief, next seed", + "native_inter_run_gap_seconds": 34.322, + "only_seed_changed_from_predecessor": true, + "text_changed_from_predecessor": false, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 48, + "inflight_sample_count": 47, + "minimum_ram_available_bytes": 8770334720, + "minimum_commit_available_bytes": 11553648640, + "minimum_system_drive_free_bytes": 18671108096, + "maximum_sampled_device_gpu_used_mib": 4315 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": true, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "A coherent front-facing seated fox figurine in the same basic studio setup, with a visibly different pose and face.", + "Orange fox, light wood and cream background are present; composition follows the brief.", + "Matte textured surface with fur-like ridges; not strongly ceramic.", + "Stylized anatomy is mostly coherent. Eye appearance is uneven, with a pale/cyan mark around the left visible eye.", + "No text requested or visibly generated.", + "Ceramic appearance is weak.", + "Small asymmetric eye detail may be a visual artifact." + ], + "media": [ + { + "path": "media/qwen-turbo-02.png", + "mime_type": "image/png", + "size_bytes": 283217, + "sha256": "238d33136a36561317196ed2a9c0096cbafe4109059ab2d0d8df5ec3079d74bd", + "source_sha256": "f8bd3fcfba3462d120142e83fb0d7ec637a9eceebd25f6db11f61bda76254eb1", + "width": 512, + "height": 512, + "pixel_sha256": "ac9f30aded9dc1c28b70d2b87cebd61a4df3d93e3dc62206d1ed31874410fe48", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-02.json", + "workflow_sha256": "a48c407f70a6f7ffb6722d35f80c1d304116d197f2307961e7f88bbd31456fa4", + "source_workflow_canonical_sha256": "84fabe49d8035de7dc9124d16ea9729ca181a0fdbca0f9e681520453dbb87ef2" +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-03.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-03.json new file mode 100644 index 0000000..91c768d --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-03.json @@ -0,0 +1,145 @@ +{ + "id": "qwen-turbo-03", + "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 44", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 44, + "status": "succeeded", + "accepted_result": true, + "prompt": "A turquoise alpine lake beneath snowy mountains at sunrise, a yellow canoe near the shore, crisp natural landscape photograph.", + "prompt_details": { + "word_count": 19, + "complexity": "Short landscape", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA turquoise alpine lake beneath snowy mountains at sunrise, a yellow canoe near the shore, crisp natural landscape photograph." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA turquoise alpine lake beneath snowy mountains at sunrise, a yellow canoe near the shore, crisp natural landscape photograph." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 44 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 195.952306, + "comfy_server_seconds": 194.865, + "stages_seconds": { + "node_3_observer_receive_interval": 100.20300000003772, + "node_10_observer_receive_interval": 0.014999999955762178, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 66.93800000002375, + "node_7_observer_receive_interval": 27.64100000000326, + "node_8_observer_receive_interval": 0.045999999972991645 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "11", + "12", + "2", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "New brief, retained model runtime", + "native_inter_run_gap_seconds": 22.977, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": true, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 96, + "inflight_sample_count": 95, + "minimum_ram_available_bytes": 4403757056, + "minimum_commit_available_bytes": 11454955520, + "minimum_system_drive_free_bytes": 18670780416, + "maximum_sampled_device_gpu_used_mib": 4315 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "Recognizable and visually coherent mountain-lake landscape with a yellow canoe-like object at the near shoreline.", + "Turquoise water, snowy mountains and warm sunlight on peaks are present. The yellow canoe is close to shore.", + "Water reflections and rock/snow separation are convincing at this size; fine granular texture is visible.", + "The canoe is heavily cropped by the lower edge, showing mostly its bow/top; full canoe geometry cannot be assessed.", + "No text requested or visibly generated.", + "Canoe is strongly cropped and less clearly legible than a full near-shore canoe.", + "Fine speckled texture is visible on the mountain surfaces." + ], + "media": [ + { + "path": "media/qwen-turbo-03.png", + "mime_type": "image/png", + "size_bytes": 474011, + "sha256": "57418cb5f47b5647b8773c0e012572e528e150b0f27f422e78f4f05d5a683a11", + "source_sha256": "c5e2e3901c3129df9943c2f995d5b42c7d646c708eb6f9053295778ed77012e7", + "width": 512, + "height": 512, + "pixel_sha256": "ff22ec81f55d0251c62e225baca505db986d0d734728cb05f57b4d2087cd12d1", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-03.json", + "workflow_sha256": "d26655caaebfa0bd74939c8f5f1977d3759141277d816b6eb1790a1c18f853c1", + "source_workflow_canonical_sha256": "96b30904e6a2fa858c6bc001bf5a0b3e4168fbcbc36fcf53b796c7ec9f07cfd2" +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-04.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-04.json new file mode 100644 index 0000000..6d17ca3 --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-04.json @@ -0,0 +1,144 @@ +{ + "id": "qwen-turbo-04", + "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 45", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 45, + "status": "succeeded", + "accepted_result": true, + "prompt": "Product photograph of a matte black ceramic teapot with a warm copper handle, resting on dark slate beside two pale linen napkins. Three-quarter view, softbox lighting, crisp material detail, uncluttered background.", + "prompt_details": { + "word_count": 31, + "complexity": "Product materials", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nProduct photograph of a matte black ceramic teapot with a warm copper handle, resting on dark slate beside two pale linen napkins. Three-quarter view, softbox lighting, crisp material detail, uncluttered background." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nProduct photograph of a matte black ceramic teapot with a warm copper handle, resting on dark slate beside two pale linen napkins. Three-quarter view, softbox lighting, crisp material detail, uncluttered background." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 45 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 198.952217, + "comfy_server_seconds": 196.417, + "stages_seconds": { + "node_3_observer_receive_interval": 100.10899999999674, + "node_10_observer_receive_interval": 0.01600000000325963, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 67.98399999999674, + "node_7_observer_receive_interval": 28.21899999998277, + "node_8_observer_receive_interval": 0.061999999976251274 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "11", + "12", + "2", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "New brief, retained model runtime", + "native_inter_run_gap_seconds": 16.828, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": true, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 98, + "inflight_sample_count": 97, + "minimum_ram_available_bytes": 4163956736, + "minimum_commit_available_bytes": 11323981824, + "minimum_system_drive_free_bytes": 18670002176, + "maximum_sampled_device_gpu_used_mib": 4351 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "A coherent, usable-looking product composition of a black teapot, copper-colored handle, dark slate and pale linen.", + "Main object, materials, slate and three-quarter product lighting are present. Only one clearly distinguishable rolled/folded napkin appears, rather than two separate napkins.", + "Black matte body, metallic handle and woven linen are well differentiated.", + "Teapot body, lid, spout and handle joins are broadly plausible; no obvious broken geometry at 512px.", + "No text requested or visibly generated.", + "Requested count of two napkins is not clearly met." + ], + "media": [ + { + "path": "media/qwen-turbo-04.png", + "mime_type": "image/png", + "size_bytes": 280344, + "sha256": "e5b3093fdb6bb39fcec4b803b1345733cf212bb274806338d27e140d3b882c3d", + "source_sha256": "1c2b9ddeca53a1fd944073a596b050e43f8f9b1b4a7e2fee57608095395b97d3", + "width": 512, + "height": 512, + "pixel_sha256": "2db9b0cf39ba6fdb3f0febf48a4f32d9fbd8012ed303d8d3ec711a4a074f8b81", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-04.json", + "workflow_sha256": "434243d6ae778adde215a2487cc10e1ee4123a6826f0f50c21e1b41238930d75", + "source_workflow_canonical_sha256": "7592e027ed1ef1f6903c6ca2bdae9faad31a02bf99cd7f67c71689cc3427c0ce" +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-05.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-05.json new file mode 100644 index 0000000..9e3c9bb --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-05.json @@ -0,0 +1,144 @@ +{ + "id": "qwen-turbo-05", + "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 46", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 46, + "status": "succeeded", + "accepted_result": true, + "prompt": "A cozy neighborhood bakery storefront on a rainy evening. The main sign clearly reads \"MORNING BREAD\" in cream block letters. Warm light glows through the windows, a bicycle leans beside the door, wet paving reflects the orange interior, realistic street photograph.", + "prompt_details": { + "word_count": 41, + "complexity": "Readable sign", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA cozy neighborhood bakery storefront on a rainy evening. The main sign clearly reads \"MORNING BREAD\" in cream block letters. Warm light glows through the windows, a bicycle leans beside the door, wet paving reflects the orange interior, realistic street photograph." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA cozy neighborhood bakery storefront on a rainy evening. The main sign clearly reads \"MORNING BREAD\" in cream block letters. Warm light glows through the windows, a bicycle leans beside the door, wet paving reflects the orange interior, realistic street photograph." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 46 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 201.997778, + "comfy_server_seconds": 199.64, + "stages_seconds": { + "node_3_observer_receive_interval": 102.10899999999674, + "node_10_observer_receive_interval": 0.01600000000325963, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 68.81200000003446, + "node_7_observer_receive_interval": 28.625, + "node_8_observer_receive_interval": 0.06299999996554106 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "11", + "12", + "2", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "New brief, retained model runtime", + "native_inter_run_gap_seconds": 16.283, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": true, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 99, + "inflight_sample_count": 98, + "minimum_ram_available_bytes": 4295344128, + "minimum_commit_available_bytes": 11412852736, + "minimum_system_drive_free_bytes": 18669809664, + "maximum_sampled_device_gpu_used_mib": 4351 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "A coherent rainy-evening bakery storefront with a legible sign, bicycle, warm windows and wet paving.", + "The main sign visibly reads exactly MORNING BREAD in large cream block letters. Bicycle, orange interior and reflected light are present.", + "Brick, glazing and wet paving read distinctly; the dark storefront feels photograph-like.", + "Storefront perspective and bicycle placement are broadly plausible at this scale; fine bicycle detail is somewhat soft.", + "Exact requested wording MORNING BREAD is visibly correct. This is a subjective visual transcription, not OCR or an automated text benchmark.", + "Fine bicycle/spoke details are soft at the supplied resolution." + ], + "media": [ + { + "path": "media/qwen-turbo-05.png", + "mime_type": "image/png", + "size_bytes": 434268, + "sha256": "94e64b09278ffff4feb04c3197545dd8d6371b281a514d0e3d1cf8550648d919", + "source_sha256": "5a9008f775161ef48b16915597d158f4a0cd80883a7c4d452cbf240bba049ce2", + "width": 512, + "height": 512, + "pixel_sha256": "18129fd4fc85277093ff6d330be029055bfde66f2a18af657ce735b6e928f894", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-05.json", + "workflow_sha256": "8768f247abc378db4fd3057cb67987743253489f11b42e75bef673f1b6c39658", + "source_workflow_canonical_sha256": "874d22b62ef466732130e9e79fa45e12bf32d748a3b1124d21b16aedac7abf59" +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-06.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-06.json new file mode 100644 index 0000000..238217f --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-06.json @@ -0,0 +1,146 @@ +{ + "id": "qwen-turbo-06", + "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 47", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 47, + "status": "succeeded", + "accepted_result": true, + "prompt": "A bright Victorian glass greenhouse filled with lush ferns, hanging orchids and terracotta pots. A narrow brick path curves from the bottom left toward an open green door at the far end. Morning sunlight makes geometric shadows through the iron window frames. Include a single brass watering can beside the path and misty trees beyond the glass. Architectural photograph, balanced natural colors, no people or text.", + "prompt_details": { + "word_count": 66, + "complexity": "Botanical architecture", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA bright Victorian glass greenhouse filled with lush ferns, hanging orchids and terracotta pots. A narrow brick path curves from the bottom left toward an open green door at the far end. Morning sunlight makes geometric shadows through the iron window frames. Include a single brass watering can beside the path and misty trees beyond the glass. Architectural photograph, balanced natural colors, no people or text." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA bright Victorian glass greenhouse filled with lush ferns, hanging orchids and terracotta pots. A narrow brick path curves from the bottom left toward an open green door at the far end. Morning sunlight makes geometric shadows through the iron window frames. Include a single brass watering can beside the path and misty trees beyond the glass. Architectural photograph, balanced natural colors, no people or text." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 47 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 201.931, + "comfy_server_seconds": 199.412, + "stages_seconds": { + "node_3_observer_receive_interval": 102.79699999996228, + "node_10_observer_receive_interval": 0.015000000013969839, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 68.43800000002375, + "node_7_observer_receive_interval": 28.04699999996228, + "node_8_observer_receive_interval": 0.062000000034458935 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "11", + "12", + "2", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "New brief, retained model runtime", + "native_inter_run_gap_seconds": 21.517, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": true, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 99, + "inflight_sample_count": 98, + "minimum_ram_available_bytes": 4470677504, + "minimum_commit_available_bytes": 11370139648, + "minimum_system_drive_free_bytes": 18651901952, + "maximum_sampled_device_gpu_used_mib": 4351 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "A coherent glass greenhouse with dense foliage, terracotta pots, a brick walkway, an open green door and a brass-colored watering can.", + "Most named scene elements are present. The path leads toward an open far door; ferns and hanging plants dominate. Distinct hanging orchid flowers are not obvious, and geometric sunlight shadows are subtle.", + "Glass/metal frames, foliage, pots and brick are distinguishable, but foliage and frame textures have noticeable fine speckling.", + "Overall greenhouse perspective is convincing; dense plant/frame detail is less clean on close inspection. Watering can is partly occluded/near the left edge.", + "No people or text requested; none visibly apparent.", + "Requested orchids are not clearly identifiable.", + "Geometric cast shadows are weaker than the brief suggests.", + "Dense foliage and framing show grainy/speckled texture." + ], + "media": [ + { + "path": "media/qwen-turbo-06.png", + "mime_type": "image/png", + "size_bytes": 649441, + "sha256": "f1ae1c3294bcffe1c0d283e957bcc81c5d48df83fe6e0f316eb4182d57db2af1", + "source_sha256": "bd1ac8888861a03d6b9a2436cede0ad7b3009f1535c11b360f794ad47a9c0321", + "width": 512, + "height": 512, + "pixel_sha256": "f74bd726756ed8f29d230b4c506533b708c21ecd53275092c3e54672dbb5a1cd", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-06.json", + "workflow_sha256": "c48538fbde2bdf10bb5849c8f29aaa79f9f05da8b1e53518af7b56789aa396a0", + "source_workflow_canonical_sha256": "6cc4ea59bd3d96d0811844809d534eda117290b8b705f13f0069b45cb5d2da6b" +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-07.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-07.json new file mode 100644 index 0000000..beec8b9 --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-07.json @@ -0,0 +1,145 @@ +{ + "id": "qwen-turbo-07", + "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 48", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 48, + "status": "succeeded", + "accepted_result": true, + "prompt": "An editorial portrait of an adult female ceramic artist in her forties working in a sunlit pottery studio. She wears a blue linen apron over a cream shirt and calmly holds a small unfinished bowl in both hands, fingers naturally curled around the rim. Her face is clear and expressive, with a few silver strands in dark curly hair. Shelves of glazed pottery remain softly out of focus behind her. Frame from the waist upward, eye-level camera, gentle window light from the left, documentary realism.", + "prompt_details": { + "word_count": 85, + "complexity": "Portrait / hands", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn editorial portrait of an adult female ceramic artist in her forties working in a sunlit pottery studio. She wears a blue linen apron over a cream shirt and calmly holds a small unfinished bowl in both hands, fingers naturally curled around the rim. Her face is clear and expressive, with a few silver strands in dark curly hair. Shelves of glazed pottery remain softly out of focus behind her. Frame from the waist upward, eye-level camera, gentle window light from the left, documentary realism." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn editorial portrait of an adult female ceramic artist in her forties working in a sunlit pottery studio. She wears a blue linen apron over a cream shirt and calmly holds a small unfinished bowl in both hands, fingers naturally curled around the rim. Her face is clear and expressive, with a few silver strands in dark curly hair. Shelves of glazed pottery remain softly out of focus behind her. Frame from the waist upward, eye-level camera, gentle window light from the left, documentary realism." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 48 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 201.952675, + "comfy_server_seconds": 199.784, + "stages_seconds": { + "node_3_observer_receive_interval": 103.42200000002049, + "node_10_observer_receive_interval": 0.015000000013969839, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 68.59399999998277, + "node_7_observer_receive_interval": 27.65600000001723, + "node_8_observer_receive_interval": 0.0779999999795109 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "11", + "12", + "2", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "New brief, retained model runtime", + "native_inter_run_gap_seconds": 13.667, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": true, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 99, + "inflight_sample_count": 98, + "minimum_ram_available_bytes": 4176404480, + "minimum_commit_available_bytes": 11358388224, + "minimum_system_drive_free_bytes": 18671104000, + "maximum_sampled_device_gpu_used_mib": 4351 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "A coherent and convincing waist-up studio portrait, with a clear adult face and a bowl held in two hands.", + "Blue apron, cream shirt, dark curly hair, pottery shelves, eye-level framing and soft window light from the left all appear. Silver hair strands are not clearly visible. Hands support the bowl from the side and underneath rather than both curling around its rim.", + "Cloth texture, skin and unfinished pottery are visibly differentiated; the bowl has a plausible unglazed look.", + "Hand shapes, wrist connections and bowl contact look broadly plausible at 512px. Some fingertips overlap or are occluded, so an exact finger-count claim is unwarranted. No obvious extra limb.", + "No text requested or visibly generated.", + "Specific silver hair detail is not clear.", + "The precise requested fingers-around-rim pose is only partly followed." + ], + "media": [ + { + "path": "media/qwen-turbo-07.png", + "mime_type": "image/png", + "size_bytes": 418100, + "sha256": "7b4feea8f758374a5544ad0e9078c8455f32f5a48e8305256ab0839f2afd9472", + "source_sha256": "ed1e6c70637edfc74eb3f051c51632087649ce6e88c705b2c4f08f1ee81d756a", + "width": 512, + "height": 512, + "pixel_sha256": "23fe51347e19cff881406c1152d713a2773a46cd9032f54cabc9dc4b5129c108", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-07.json", + "workflow_sha256": "23cb0fb14fb4cf31e9a4a67434ac21ab3e74b9858411f7e5724aaf8abbbbb755", + "source_workflow_canonical_sha256": "fc32cc9a0f7f8172edfda1400d2b8083ecbf5c86b636b5db6db024c9d20d4142" +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-08.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-08.json new file mode 100644 index 0000000..19ae05a --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-08.json @@ -0,0 +1,145 @@ +{ + "id": "qwen-turbo-08", + "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 49", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 49, + "status": "succeeded", + "accepted_result": true, + "prompt": "A charming small brass robot with a round glass head stands on a wooden step stool inside an enormous old library, carefully returning a red book to a high shelf. The robot has two articulated arms and three visible fingers on each hand. A tiny blue mechanical bird perches on its shoulder. Bookshelves curve toward a domed ceiling painted with constellations; the far aisle is illuminated by warm amber lamps. Render as a detailed hand-painted storybook illustration with teal shadows, golden highlights and a clear focal point. Keep the robot, stool and book visibly separate, with no extra limbs, no captions and no written lettering.", + "prompt_details": { + "word_count": 105, + "complexity": "Stylized story scene", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA charming small brass robot with a round glass head stands on a wooden step stool inside an enormous old library, carefully returning a red book to a high shelf. The robot has two articulated arms and three visible fingers on each hand. A tiny blue mechanical bird perches on its shoulder. Bookshelves curve toward a domed ceiling painted with constellations; the far aisle is illuminated by warm amber lamps. Render as a detailed hand-painted storybook illustration with teal shadows, golden highlights and a clear focal point. Keep the robot, stool and book visibly separate, with no extra limbs, no captions and no written lettering." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA charming small brass robot with a round glass head stands on a wooden step stool inside an enormous old library, carefully returning a red book to a high shelf. The robot has two articulated arms and three visible fingers on each hand. A tiny blue mechanical bird perches on its shoulder. Bookshelves curve toward a domed ceiling painted with constellations; the far aisle is illuminated by warm amber lamps. Render as a detailed hand-painted storybook illustration with teal shadows, golden highlights and a clear focal point. Keep the robot, stool and book visibly separate, with no extra limbs, no captions and no written lettering." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 49 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 210.875124, + "comfy_server_seconds": 210.023, + "stages_seconds": { + "node_3_observer_receive_interval": 107.75, + "node_10_observer_receive_interval": 0.01600000000325963, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 72.96900000004098, + "node_7_observer_receive_interval": 29.20299999997951, + "node_8_observer_receive_interval": 0.061999999976251274 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "11", + "12", + "2", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "New brief, retained model runtime", + "native_inter_run_gap_seconds": 14.441, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": true, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 104, + "inflight_sample_count": 103, + "minimum_ram_available_bytes": 4373700608, + "minimum_commit_available_bytes": 11362136064, + "minimum_system_drive_free_bytes": 18670817280, + "maximum_sampled_device_gpu_used_mib": 4351 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "A coherent and attractive storybook-style library scene: brass robot, glass-like head, wooden stool, red book, blue shoulder bird and long illuminated aisle are all recognizable.", + "The specific main objects, articulated arms, teal/gold palette, warm far lamps and ceiling decoration are strongly represented. The exact three-visible-fingers-per-hand requirement is hard to certify because several small digits overlap or turn away.", + "Brass, a reflective glass dome, wood and book covers read distinctly within the painted style.", + "Robot, stool, book and shelves remain visibly separate. Arm joints and stool perspective are broadly coherent. Finger-count correctness cannot be established confidently at the supplied resolution.", + "No legible caption, title or written lettering is apparent.", + "Exact requested finger count remains uncertain; small jointed digits overlap.", + "Glass head is mainly reflective/opaque rather than clearly transparent, but still reads as a glass dome." + ], + "media": [ + { + "path": "media/qwen-turbo-08.png", + "mime_type": "image/png", + "size_bytes": 518714, + "sha256": "bd1bcc5079a48be9f59a4d72511eae46abb6e2aa788fe22f72bbac1cc4ba8b7a", + "source_sha256": "63abbaf5c49b2baca103bd65761fd4fed09f7e5f173406e91eaaf47d136694f2", + "width": 512, + "height": 512, + "pixel_sha256": "224125f9ca31c3c15dfa975dd61d42fa8f1703d9c012374876d2242054759658", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-08.json", + "workflow_sha256": "000bdfef44a50c9cd6275031d4ab57a84f36f216784d1f8c78d637eee60c820a", + "source_workflow_canonical_sha256": "3a9b68b7e06cf4b2a375d516a8f5d38b555d7ba5e593354edce22542adf12865" +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-09.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-09.json new file mode 100644 index 0000000..316bd41 --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-09.json @@ -0,0 +1,146 @@ +{ + "id": "qwen-turbo-09", + "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 50", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 50, + "status": "succeeded", + "accepted_result": true, + "prompt": "A detailed travel photograph of a lively Mediterranean open-air produce market in early morning. In the foreground, three wooden crates hold red tomatoes, striped eggplants and yellow lemons, each type kept in its own crate. An adult vendor wearing a plain white shirt stands behind the stall, handing a small paper bag to an adult customer in a green jacket. A striped blue-and-white canopy casts cool shade over the stall. Beyond them, a narrow pedestrian street slopes toward a distant harbor with two white sailboats. Warm sunlight touches the upper walls but the foreground remains softly shaded. Compose the scene in layers with natural human proportions, believable hands, clear separation between people, baskets and produce, and rich but realistic color. Avoid artificial lens flares, watermarks and unreadable signs.", + "prompt_details": { + "word_count": 128, + "complexity": "Crowded composition", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA detailed travel photograph of a lively Mediterranean open-air produce market in early morning. In the foreground, three wooden crates hold red tomatoes, striped eggplants and yellow lemons, each type kept in its own crate. An adult vendor wearing a plain white shirt stands behind the stall, handing a small paper bag to an adult customer in a green jacket. A striped blue-and-white canopy casts cool shade over the stall. Beyond them, a narrow pedestrian street slopes toward a distant harbor with two white sailboats. Warm sunlight touches the upper walls but the foreground remains softly shaded. Compose the scene in layers with natural human proportions, believable hands, clear separation between people, baskets and produce, and rich but realistic color. Avoid artificial lens flares, watermarks and unreadable signs." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA detailed travel photograph of a lively Mediterranean open-air produce market in early morning. In the foreground, three wooden crates hold red tomatoes, striped eggplants and yellow lemons, each type kept in its own crate. An adult vendor wearing a plain white shirt stands behind the stall, handing a small paper bag to an adult customer in a green jacket. A striped blue-and-white canopy casts cool shade over the stall. Beyond them, a narrow pedestrian street slopes toward a distant harbor with two white sailboats. Warm sunlight touches the upper walls but the foreground remains softly shaded. Compose the scene in layers with natural human proportions, believable hands, clear separation between people, baskets and produce, and rich but realistic color. Avoid artificial lens flares, watermarks and unreadable signs." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 50 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 210.965741, + "comfy_server_seconds": 208.566, + "stages_seconds": { + "node_3_observer_receive_interval": 107.82800000003772, + "node_10_observer_receive_interval": 0, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 72.20299999997951, + "node_7_observer_receive_interval": 28.43800000002375, + "node_8_observer_receive_interval": 0.061999999976251274 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "11", + "12", + "2", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "New brief, retained model runtime", + "native_inter_run_gap_seconds": 13.101, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": true, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 103, + "inflight_sample_count": 102, + "minimum_ram_available_bytes": 4364185600, + "minimum_commit_available_bytes": 11356860416, + "minimum_system_drive_free_bytes": 18670198784, + "maximum_sampled_device_gpu_used_mib": 4351 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "Recognizable market transaction under a striped canopy, with three separate produce crates and a harbor beyond. Main scene layers are coherent.", + "White-shirt vendor, green-jacket customer, blue-and-white canopy, separate tomatoes/eggplants/lemons and sunlit upper wall over shaded foreground are all present. The paper bag and precise handoff are visually ambiguous. Harbor sailboats are visible but the exact requested count of two is not certifiable at this size.", + "Vegetables, wood, clothing and canopy fabric are distinguishable. Fine background detail is smeared/brushy despite a photographic brief.", + "Main adult body proportions look plausible. The central touching hands are soft and partially merged/occluded; finger anatomy and grasp cannot be verified. Background masts and architecture become muddled.", + "No prominent legible sign, watermark or caption appears.", + "Paper-bag handoff is less clear than requested.", + "Hands and distant harbor details are imprecise.", + "Exact two-sailboat count remains uncertain." + ], + "media": [ + { + "path": "media/qwen-turbo-09.png", + "mime_type": "image/png", + "size_bytes": 437788, + "sha256": "847ec6dd9be4503a94967f96ca4b91d223ffb7638749c4fd08fb318723f544e0", + "source_sha256": "5b992fa647b0461c1eb8514854a32b5f44cdd297a129a110f72b1abff6a830ef", + "width": 512, + "height": 512, + "pixel_sha256": "ffaa52b367e79221e07aeed4066dcadc2039859372848f38c9bc028d623b7f50", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-09.json", + "workflow_sha256": "049b53d4c92125dbe083563cea3d836330cfd864f19317933f71cb8794bc7db6", + "source_workflow_canonical_sha256": "e1fd339039ed8d1dde53d4a0668665f4b78470cce872ff0c3f831156539dd328" +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-10.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-10.json new file mode 100644 index 0000000..cb82167 --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/records/qwen-turbo-10.json @@ -0,0 +1,146 @@ +{ + "id": "qwen-turbo-10", + "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 51", + "group": "qwen21-turbo-q4km-ten", + "kind": "image", + "model_id": "qwen-turbo", + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "seed": 51, + "status": "succeeded", + "accepted_result": true, + "prompt": "Create a highly detailed cinematic architectural illustration of a remote mountain observatory at blue hour after a light snowfall. The building occupies the right two-thirds of the square image: a circular pale-stone base supports one weathered copper dome, whose open viewing slit points toward the upper left. A narrow exterior staircase with dark iron handrails climbs from a small lower entrance to a balcony; every step has a thin undisturbed layer of snow. One adult astronomer in a dark red winter coat stands on the balcony with a small white notebook held closed in one gloved hand, looking toward the sky. Place a warm amber lamp immediately above the entrance, creating a pool of light on the stone without illuminating the whole landscape. In the lower left foreground, a crooked pine tree and two exposed granite rocks frame a winding path that leads toward the building. Behind the observatory, show distant mountain ridges fading from slate blue into violet mist. The sky contains scattered sharp stars and a faint natural Milky Way band, with no oversized planets or fantasy moons. Balance warm interior light against cold ambient snow light, use physically plausible perspective, careful separation of the staircase rails and the figure, and visible differences between copper, stone, glass, fabric and snow. The final scene should feel quiet, inhabited and believable. Do not add titles, labels, logos, watermarks, extra people, extra domes or duplicate staircases.", + "prompt_details": { + "word_count": 235, + "complexity": "Elaborate constraints", + "conditioning_texts": { + "3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nCreate a highly detailed cinematic architectural illustration of a remote mountain observatory at blue hour after a light snowfall. The building occupies the right two-thirds of the square image: a circular pale-stone base supports one weathered copper dome, whose open viewing slit points toward the upper left. A narrow exterior staircase with dark iron handrails climbs from a small lower entrance to a balcony; every step has a thin undisturbed layer of snow. One adult astronomer in a dark red winter coat stands on the balcony with a small white notebook held closed in one gloved hand, looking toward the sky. Place a warm amber lamp immediately above the entrance, creating a pool of light on the stone without illuminating the whole landscape. In the lower left foreground, a crooked pine tree and two exposed granite rocks frame a winding path that leads toward the building. Behind the observatory, show distant mountain ridges fading from slate blue into violet mist. The sky contains scattered sharp stars and a faint natural Milky Way band, with no oversized planets or fantasy moons. Balance warm interior light against cold ambient snow light, use physically plausible perspective, careful separation of the staircase rails and the figure, and visible differences between copper, stone, glass, fabric and snow. The final scene should feel quiet, inhabited and believable. Do not add titles, labels, logos, watermarks, extra people, extra domes or duplicate staircases." + }, + "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nCreate a highly detailed cinematic architectural illustration of a remote mountain observatory at blue hour after a light snowfall. The building occupies the right two-thirds of the square image: a circular pale-stone base supports one weathered copper dome, whose open viewing slit points toward the upper left. A narrow exterior staircase with dark iron handrails climbs from a small lower entrance to a balcony; every step has a thin undisturbed layer of snow. One adult astronomer in a dark red winter coat stands on the balcony with a small white notebook held closed in one gloved hand, looking toward the sky. Place a warm amber lamp immediately above the entrance, creating a pool of light on the stone without illuminating the whole landscape. In the lower left foreground, a crooked pine tree and two exposed granite rocks frame a winding path that leads toward the building. Behind the observatory, show distant mountain ridges fading from slate blue into violet mist. The sky contains scattered sharp stars and a faint natural Milky Way band, with no oversized planets or fantasy moons. Balance warm interior light against cold ambient snow light, use physically plausible perspective, careful separation of the staircase rails and the figure, and visible differences between copper, stone, glass, fabric and snow. The final scene should feel quiet, inhabited and believable. Do not add titles, labels, logos, watermarks, extra people, extra domes or duplicate staircases." + }, + "parameters": { + "1 UnetLoaderGGUFAdvanced": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + }, + "2 CLIPLoader": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + }, + "3 TextEncodeQwenImage21": { + "negative_prompt": "", + "resolution": 512 + }, + "4 VAELoader": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + }, + "5 QwenImage21Cache": { + "device": "off", + "dtype": "default" + }, + "6 SamplerCustomAdvanced": {}, + "7 VAEDecode": {}, + "8 SaveImage": {}, + "9 RandomNoise": { + "noise_seed": 51 + }, + "10 CFGGuider": { + "cfg": 1.0 + }, + "11 KSamplerSelect": { + "sampler_name": "euler" + }, + "12 ManualSigmas": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + }, + "timing": { + "factory_trace_seconds": 226.020651, + "comfy_server_seconds": 223.135, + "stages_seconds": { + "node_3_observer_receive_interval": 115.98399999999674, + "node_10_observer_receive_interval": 0.01600000000325963, + "node_9_observer_receive_interval": 0, + "node_6_observer_receive_interval": 78.40600000001723, + "node_7_observer_receive_interval": 28.639999999955762, + "node_8_observer_receive_interval": 0.07900000002700835 + }, + "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times." + }, + "cache": { + "cached_nodes": [ + "1", + "11", + "12", + "2", + "4", + "5" + ], + "same_runtime_retained": true, + "immediate_warm_claim": false, + "encoder_cached": false, + "context": { + "workload": "New brief, retained model runtime", + "native_inter_run_gap_seconds": 10.771, + "only_seed_changed_from_predecessor": false, + "text_changed_from_predecessor": true, + "cold_start_scope": null + } + }, + "resources": { + "sampled_summary": { + "sample_count": 111, + "inflight_sample_count": 110, + "minimum_ram_available_bytes": 4460896256, + "minimum_commit_available_bytes": 11257696256, + "minimum_system_drive_free_bytes": 18669465600, + "maximum_sampled_device_gpu_used_mib": 4351 + }, + "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging." + }, + "encoding": { + "encoder_cached": false, + "token_truncation_verified": false, + "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained." + }, + "verification": { + "full_decode": true, + "checksum": true, + "acknowledged": true, + "trace_attachment": true, + "native_experiment": true, + "independent_public_copy_decode": true, + "visual_review": true, + "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule." + }, + "findings": [ + "A coherent quiet blue-hour observatory scene with one large dome, one red-coated figure, one snow-covered staircase, foreground pine/rocks and distant layered mountains.", + "Building placement, open slit toward the upper left, red coat, small light-colored notebook, amber entrance lamp, snow on steps and cold/warm lighting contrast are represented. The figure appears to attend to the notebook rather than clearly look at the sky. Requested stars and faint Milky Way are not visibly present. The exact two-rock count is not cleanly met or certifiable among overlapping foreground boulders.", + "Dome reads as dark weathered metal, base as smooth pale masonry/stucco, and snow, coat and rails are distinct. Fine stone material detail is less explicit than the brief.", + "Overall building, balcony and staircase perspective are plausible; thin rail details and figure-to-railing separation are imperfect but readable. No obvious extra person, dome or duplicate staircase.", + "No title, label, logo or watermark is apparent.", + "Stars and Milky Way are absent from the visibly plain sky.", + "Figure does not clearly follow the requested skyward gaze.", + "Exact rock-count and some fine architectural/material details are weaker than the broad composition." + ], + "media": [ + { + "path": "media/qwen-turbo-10.png", + "mime_type": "image/png", + "size_bytes": 402891, + "sha256": "7bc1033039962507ee2aa153ebfacaa4cdd1f77e93e152bd7241ebae686b85bc", + "source_sha256": "f051ec1ab055f23d3fd6513e3428d807734a846f5e51881403cf419ccb33c161", + "width": 512, + "height": 512, + "pixel_sha256": "9672ee9777924c1b99ccff8952362c0119f99a87bb1e22c7bc82a24b579ccb6f", + "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", + "public_copy_full_decode_passed": true, + "source_content_preserved": "Decoded pixels byte-identical" + } + ], + "workflow_path": "workflows/qwen-turbo-10.json", + "workflow_sha256": "11fe5261e6abff03fd6a82be45d93b3827c421e3cc3e684f03b72df24694e0b0", + "source_workflow_canonical_sha256": "7120b879556ee04ac859a6126f8d1ab24771a28bf208d58fff10b120bf50e68a" +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/results.csv b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/results.csv index 6906684..8eab942 100644 --- a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/results.csv +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/results.csv @@ -522,6 +522,16 @@ Arrange the tallest violet plant on the left, the compact white flowering plant Treat the image as a carefully composed study of invented but botanically plausible alpine plants, without claiming an exact real species identification. The violet specimen has slender upright stems, narrow alternating leaves, and a few hanging bell shaped flowers at different stages of opening. The central white specimen grows in a low rosette with broader slightly hairy leaves and short stalks supporting small five pointed blossoms. The yellow specimen branches modestly near the top, carrying tight clusters of tiny round flowers above toothed leaves. Each plant has a distinct root architecture: fine branching fibers on the violet plant, a compact fibrous crown on the white plant, and one thicker central root with smaller side branches on the yellow plant. Roots should taper naturally and remain attached to their crowns. Use small graphite accents to clarify leaf veins and flower structures without turning every edge into a heavy outline. The detached leaves show their undersides, revealing slightly paler colors and raised veins, while the seed heads provide a contrasting dry brown texture. Keep shadows extremely light and close to the paper so the sheet reads as an illustration rather than a floating specimen photograph. The title should be modest, evenly spaced, and legible, with no invented extra text elsewhere. Preserve a quiet hierarchy of fine detail, open paper, and three distinct growth habits, suitable for close visual comparison across the botanical forms." cog-03,cogvideox2b-q8-reference-ddim-text-v1,zai-org/CogVideoX-2b:Q8_0,video,42,succeeded,True,2439.989339,2433.832,media/cog-03.mp4,"A realistic red fox walks steadily across a snowy clearing from the left side of the frame to the right. Show the fox in full body from the side. Its four legs visibly alternate with each walking step, its body advances across the ground, and its bushy tail swings gently. Near the end the fox stops walking and turns its head toward the camera. The camera stays completely fixed in one wide shot. Continuous visible animal movement throughout the walk. Natural daylight, realistic anatomy, one continuous six-second shot, no cuts, no text." cog-04,cogvideox2b-q8-reference-ddim-text-v1,zai-org/CogVideoX-2b:Q8_0,video,43,succeeded,True,2393.862082,2390.252,media/cog-04.mp4,"A realistic red fox walks steadily across a snowy clearing from the left side of the frame to the right. Show the fox in full body from the side. Its four legs visibly alternate with each walking step, its body advances across the ground, and its bushy tail swings gently. Near the end the fox stops walking and turns its head toward the camera. The camera stays completely fixed in one wide shot. Continuous visible animal movement throughout the walk. Natural daylight, realistic anatomy, one continuous six-second shot, no cuts, no text." +qwen-turbo-01,qwen21-turbo-q4km-ten,unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M,image,42,succeeded,True,334.589337,320.246,media/qwen-turbo-01.png,"A small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph." +qwen-turbo-02,qwen21-turbo-q4km-ten,unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M,image,43,succeeded,True,96.670077,95.234,media/qwen-turbo-02.png,"A small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph." +qwen-turbo-03,qwen21-turbo-q4km-ten,unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M,image,44,succeeded,True,195.952306,194.865,media/qwen-turbo-03.png,"A turquoise alpine lake beneath snowy mountains at sunrise, a yellow canoe near the shore, crisp natural landscape photograph." +qwen-turbo-04,qwen21-turbo-q4km-ten,unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M,image,45,succeeded,True,198.952217,196.417,media/qwen-turbo-04.png,"Product photograph of a matte black ceramic teapot with a warm copper handle, resting on dark slate beside two pale linen napkins. Three-quarter view, softbox lighting, crisp material detail, uncluttered background." +qwen-turbo-05,qwen21-turbo-q4km-ten,unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M,image,46,succeeded,True,201.997778,199.64,media/qwen-turbo-05.png,"A cozy neighborhood bakery storefront on a rainy evening. The main sign clearly reads ""MORNING BREAD"" in cream block letters. Warm light glows through the windows, a bicycle leans beside the door, wet paving reflects the orange interior, realistic street photograph." +qwen-turbo-06,qwen21-turbo-q4km-ten,unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M,image,47,succeeded,True,201.931,199.412,media/qwen-turbo-06.png,"A bright Victorian glass greenhouse filled with lush ferns, hanging orchids and terracotta pots. A narrow brick path curves from the bottom left toward an open green door at the far end. Morning sunlight makes geometric shadows through the iron window frames. Include a single brass watering can beside the path and misty trees beyond the glass. Architectural photograph, balanced natural colors, no people or text." +qwen-turbo-07,qwen21-turbo-q4km-ten,unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M,image,48,succeeded,True,201.952675,199.784,media/qwen-turbo-07.png,"An editorial portrait of an adult female ceramic artist in her forties working in a sunlit pottery studio. She wears a blue linen apron over a cream shirt and calmly holds a small unfinished bowl in both hands, fingers naturally curled around the rim. Her face is clear and expressive, with a few silver strands in dark curly hair. Shelves of glazed pottery remain softly out of focus behind her. Frame from the waist upward, eye-level camera, gentle window light from the left, documentary realism." +qwen-turbo-08,qwen21-turbo-q4km-ten,unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M,image,49,succeeded,True,210.875124,210.023,media/qwen-turbo-08.png,"A charming small brass robot with a round glass head stands on a wooden step stool inside an enormous old library, carefully returning a red book to a high shelf. The robot has two articulated arms and three visible fingers on each hand. A tiny blue mechanical bird perches on its shoulder. Bookshelves curve toward a domed ceiling painted with constellations; the far aisle is illuminated by warm amber lamps. Render as a detailed hand-painted storybook illustration with teal shadows, golden highlights and a clear focal point. Keep the robot, stool and book visibly separate, with no extra limbs, no captions and no written lettering." +qwen-turbo-09,qwen21-turbo-q4km-ten,unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M,image,50,succeeded,True,210.965741,208.566,media/qwen-turbo-09.png,"A detailed travel photograph of a lively Mediterranean open-air produce market in early morning. In the foreground, three wooden crates hold red tomatoes, striped eggplants and yellow lemons, each type kept in its own crate. An adult vendor wearing a plain white shirt stands behind the stall, handing a small paper bag to an adult customer in a green jacket. A striped blue-and-white canopy casts cool shade over the stall. Beyond them, a narrow pedestrian street slopes toward a distant harbor with two white sailboats. Warm sunlight touches the upper walls but the foreground remains softly shaded. Compose the scene in layers with natural human proportions, believable hands, clear separation between people, baskets and produce, and rich but realistic color. Avoid artificial lens flares, watermarks and unreadable signs." +qwen-turbo-10,qwen21-turbo-q4km-ten,unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M,image,51,succeeded,True,226.020651,223.135,media/qwen-turbo-10.png,"Create a highly detailed cinematic architectural illustration of a remote mountain observatory at blue hour after a light snowfall. The building occupies the right two-thirds of the square image: a circular pale-stone base supports one weathered copper dome, whose open viewing slit points toward the upper left. A narrow exterior staircase with dark iron handrails climbs from a small lower entrance to a balcony; every step has a thin undisturbed layer of snow. One adult astronomer in a dark red winter coat stands on the balcony with a small white notebook held closed in one gloved hand, looking toward the sky. Place a warm amber lamp immediately above the entrance, creating a pool of light on the stone without illuminating the whole landscape. In the lower left foreground, a crooked pine tree and two exposed granite rocks frame a winding path that leads toward the building. Behind the observatory, show distant mountain ridges fading from slate blue into violet mist. The sky contains scattered sharp stars and a faint natural Milky Way band, with no oversized planets or fantasy moons. Balance warm interior light against cold ambient snow light, use physically plausible perspective, careful separation of the staircase rails and the figure, and visible differences between copper, stone, glass, fabric and snow. The final scene should feel quiet, inhabited and believable. Do not add titles, labels, logos, watermarks, extra people, extra domes or duplicate staircases." wan-failure-01,wan21-small-v1,Wan-AI/Wan2.1-T2V-1.3B,video,43,cancelled,False,,,, ltx-failure-01,ltxv2b-q8-text-v1,Lightricks/LTX-Video:2B-0.9.8-distilled-Q8_0,video,42,cancelled,False,,,,"A realistic red fox walks steadily across a snowy clearing from the left side of the frame to the right. Show the fox in full body from the side. Its four legs visibly alternate with each walking step, its body advances across the ground, and its bushy tail swings gently. Near the end the fox stops walking and turns its head toward the camera. The camera stays completely fixed in one wide shot. Continuous visible animal movement throughout the walk. Natural daylight, realistic anatomy, one continuous five-second shot, no cuts, no text." ltx-failure-02,ltxv2b-q8-text-v2,Lightricks/LTX-Video:2B-0.9.8-distilled-Q8_0,video,42,cancelled,False,,,,"A realistic red fox walks steadily across a snowy clearing from the left side of the frame to the right. Show the fox in full body from the side. Its four legs visibly alternate with each walking step, its body advances across the ground, and its bushy tail swings gently. Near the end the fox stops walking and turns its head toward the camera. The camera stays completely fixed in one wide shot. Continuous visible animal movement throughout the walk. Natural daylight, realistic anatomy, one continuous five-second shot, no cuts, no text." diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/run.jsonl b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/run.jsonl index 608e475..b33badf 100644 --- a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/run.jsonl +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/run.jsonl @@ -121,6 +121,16 @@ {"id": "pf-img100-20-t5-s43", "label": "An alpine botanical study", "group": "image100", "kind": "image", "model_id": "flux3", "model": "Abiray/flux-3-image-klein-sd-GGUF:Q4_K_M", "seed": 43, "status": "succeeded", "accepted_result": true, "prompt": "A botanical study arranges alpine wildflowers, roots, leaves, and seed heads beneath the small title ALPINE STUDY.\n\nUse delicate watercolor and fine graphite on warm white paper. Show three different mountain plants side by side, each complete from roots to blossoms, with ample space between the specimens.\n\nThe exact title ALPINE STUDY sits centered above the plants in small dark uppercase letters. One specimen has violet bell flowers, one has white star shaped flowers, and one carries yellow clustered blooms. Paint translucent petals, veined leaves, fine roots, and a few separate seed heads below. Keep the presentation observational and elegant, with no additional labels, measurements, or decorative border.\n\nArrange the tallest violet plant on the left, the compact white flowering plant in the center, and the yellow branching plant on the right. Their stems should emerge naturally from the corresponding root crowns, with leaves attached at believable nodes. Include one detached leaf and one seed head beneath each specimen, aligned loosely with that plant rather than scattered randomly. Subtle graphite construction marks may remain around a few petal edges. Use a consistent implied scale across the sheet, and keep the root systems from tangling together. The title must remain separate from every stem and blossom. The paper stays largely clean, with only gentle variation in the watercolor washes.\n\nTreat the image as a carefully composed study of invented but botanically plausible alpine plants, without claiming an exact real species identification. The violet specimen has slender upright stems, narrow alternating leaves, and a few hanging bell shaped flowers at different stages of opening. The central white specimen grows in a low rosette with broader slightly hairy leaves and short stalks supporting small five pointed blossoms. The yellow specimen branches modestly near the top, carrying tight clusters of tiny round flowers above toothed leaves. Each plant has a distinct root architecture: fine branching fibers on the violet plant, a compact fibrous crown on the white plant, and one thicker central root with smaller side branches on the yellow plant. Roots should taper naturally and remain attached to their crowns. Use small graphite accents to clarify leaf veins and flower structures without turning every edge into a heavy outline. The detached leaves show their undersides, revealing slightly paler colors and raised veins, while the seed heads provide a contrasting dry brown texture. Keep shadows extremely light and close to the paper so the sheet reads as an illustration rather than a floating specimen photograph. The title should be modest, evenly spaced, and legible, with no invented extra text elsewhere. Preserve a quiet hierarchy of fine detail, open paper, and three distinct growth habits, suitable for close visual comparison across the botanical forms.", "prompt_details": {"conditioning_texts": {"4.text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA botanical study arranges alpine wildflowers, roots, leaves, and seed heads beneath the small title ALPINE STUDY.\n\nUse delicate watercolor and fine graphite on warm white paper. Show three different mountain plants side by side, each complete from roots to blossoms, with ample space between the specimens.\n\nThe exact title ALPINE STUDY sits centered above the plants in small dark uppercase letters. One specimen has violet bell flowers, one has white star shaped flowers, and one carries yellow clustered blooms. Paint translucent petals, veined leaves, fine roots, and a few separate seed heads below. Keep the presentation observational and elegant, with no additional labels, measurements, or decorative border.\n\nArrange the tallest violet plant on the left, the compact white flowering plant in the center, and the yellow branching plant on the right. Their stems should emerge naturally from the corresponding root crowns, with leaves attached at believable nodes. Include one detached leaf and one seed head beneath each specimen, aligned loosely with that plant rather than scattered randomly. Subtle graphite construction marks may remain around a few petal edges. Use a consistent implied scale across the sheet, and keep the root systems from tangling together. The title must remain separate from every stem and blossom. The paper stays largely clean, with only gentle variation in the watercolor washes.\n\nTreat the image as a carefully composed study of invented but botanically plausible alpine plants, without claiming an exact real species identification. The violet specimen has slender upright stems, narrow alternating leaves, and a few hanging bell shaped flowers at different stages of opening. The central white specimen grows in a low rosette with broader slightly hairy leaves and short stalks supporting small five pointed blossoms. The yellow specimen branches modestly near the top, carrying tight clusters of tiny round flowers above toothed leaves. Each plant has a distinct root architecture: fine branching fibers on the violet plant, a compact fibrous crown on the white plant, and one thicker central root with smaller side branches on the yellow plant. Roots should taper naturally and remain attached to their crowns. Use small graphite accents to clarify leaf veins and flower structures without turning every edge into a heavy outline. The detached leaves show their undersides, revealing slightly paler colors and raised veins, while the seed heads provide a contrasting dry brown texture. Keep shadows extremely light and close to the paper so the sheet reads as an illustration rather than a floating specimen photograph. The title should be modest, evenly spaced, and legible, with no invented extra text elsewhere. Preserve a quiet hierarchy of fine detail, open paper, and three distinct growth habits, suitable for close visual comparison across the botanical forms.", "5.text": ""}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA botanical study arranges alpine wildflowers, roots, leaves, and seed heads beneath the small title ALPINE STUDY.\n\nUse delicate watercolor and fine graphite on warm white paper. Show three different mountain plants side by side, each complete from roots to blossoms, with ample space between the specimens.\n\nThe exact title ALPINE STUDY sits centered above the plants in small dark uppercase letters. One specimen has violet bell flowers, one has white star shaped flowers, and one carries yellow clustered blooms. Paint translucent petals, veined leaves, fine roots, and a few separate seed heads below. Keep the presentation observational and elegant, with no additional labels, measurements, or decorative border.\n\nArrange the tallest violet plant on the left, the compact white flowering plant in the center, and the yellow branching plant on the right. Their stems should emerge naturally from the corresponding root crowns, with leaves attached at believable nodes. Include one detached leaf and one seed head beneath each specimen, aligned loosely with that plant rather than scattered randomly. Subtle graphite construction marks may remain around a few petal edges. Use a consistent implied scale across the sheet, and keep the root systems from tangling together. The title must remain separate from every stem and blossom. The paper stays largely clean, with only gentle variation in the watercolor washes.\n\nTreat the image as a carefully composed study of invented but botanically plausible alpine plants, without claiming an exact real species identification. The violet specimen has slender upright stems, narrow alternating leaves, and a few hanging bell shaped flowers at different stages of opening. The central white specimen grows in a low rosette with broader slightly hairy leaves and short stalks supporting small five pointed blossoms. The yellow specimen branches modestly near the top, carrying tight clusters of tiny round flowers above toothed leaves. Each plant has a distinct root architecture: fine branching fibers on the violet plant, a compact fibrous crown on the white plant, and one thicker central root with smaller side branches on the yellow plant. Roots should taper naturally and remain attached to their crowns. Use small graphite accents to clarify leaf veins and flower structures without turning every edge into a heavy outline. The detached leaves show their undersides, revealing slightly paler colors and raised veins, while the seed heads provide a contrasting dry brown texture. Keep shadows extremely light and close to the paper so the sheet reads as an illustration rather than a floating specimen photograph. The title should be modest, evenly spaced, and legible, with no invented extra text elsewhere. Preserve a quiet hierarchy of fine detail, open paper, and three distinct growth habits, suitable for close visual comparison across the botanical forms.", "scene_id": "alpine-study", "scene_title": "An alpine botanical study", "tier": 5, "tier_name": "extended", "word_count": 449, "requested_traits": ["Three botanical specimens", "Title ALPINE STUDY", "Watercolor and graphite", "Whole plants including roots", "Violet bells, white stars and yellow clusters", "No extra labels", "Attached leaves and root crowns", "Detached parts grouped by plant", "Distinct plausible root systems", "Clean legible title", "No claim of exact real species"], "prompt_case_id": "pf-img100-20-t5", "arm": "second"}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "flux3_image_klein_sd_Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_4b_int8_convrot.safetensors", "type": "flux3", "device": "cpu"}, "3 VAELoader": {"vae_name": "flux2-vae.safetensors"}, "6 EmptyFlux2LatentImage": {"width": 512, "height": 512, "batch_size": 1}, "7 KSampler": {"seed": 43, "steps": 4, "cfg": 1.0, "sampler_name": "euler", "scheduler": "simple", "denoise": 1.0}}, "timing": {"factory_trace_seconds": 45.375145, "comfy_server_seconds": 44.37, "comfy_printed_seconds": null, "stages_seconds": {}, "limits": "Factory trace and server totals have different boundaries. Observer node intervals include preparation/event delivery; not isolated kernel or weight-loading times."}, "cache": {"cached_nodes": ["1", "2", "3", "4", "5", "6"], "same_runtime_retained": true, "immediate_warm_claim": null, "context": {}, "encoder_cached": true, "negative_encoder_cached": true}, "resources": {"sample_count": 24, "inflight_sample_count": 23, "minimum_ram_available_bytes": 8594112512, "minimum_commit_available_bytes": 18824183808, "minimum_system_drive_free_bytes": 20274868224, "maximum_sampled_device_gpu_used_mib": 2435}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "quality_scope": "Limited sampled review; no systematic full-resolution/per-trait/full-rate quality score.", "independent_public_copy_decode": true}, "findings": ["All50thumbnails show recognizable requested scene families.", "No blank or colored-noise output observed.", "Longer briefs visibly affect composition, supporting details and poster layout.", "Fine anatomy, exact lettering, object-part geometry and all requested constraints remain unscored at full resolution."], "media": [{"path": "media/pf-img100-20-t5-s43.png", "mime_type": "image/png", "size_bytes": 380354, "sha256": "c44512027c26fe213a4cfb979380c25ea045f7872233f29575e19a2af76dd7af", "source_sha256": "23e6212f00c73480203502f88adbb0124f21300199c8d835b2208d74afaf8b35", "width": 512, "height": 512, "pixel_sha256": "9f5e9319f1dbcc5d287b35dc3b69dafccef9a3a5c03f68632aa98908df10f728", "metadata_handling": "PNG re-encoded losslessly without ancillarymetadata; decodedpixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "encoding": {"token_probes": [{"tokenizer_settings": {"qwen3vl_4b": {"max_length": 99999999, "min_length": 1, "pad_to_max_length": false, "pad_token": 151643}}, "text_sha256": "0561c982561d5459c23ebbbe4d1ef1ae734485c72868798a79d8f2a74b26afc9", "text_characters": 2997, "text_utf8_bytes": 2997, "text_words": 460, "actual_token_chunks": {"qwen3vl_4b": {"chunk_lengths": [640], "total_padded_token_slots": 640}}, "tokenization_seconds": 0.0, "tokenizer_class": "Flux3Tokenizer", "prompt_shortened_by_this_adapter": false, "padding_is_not_content_token_count": true, "encoding_status": "completed", "conditioning_shapes": [[1, 640, 20480]], "encoding_seconds": 56.32800000003772, "recorded_at": "2026-10-21T01:10:44.906414+00:00"}, {"tokenizer_settings": {"qwen3vl_4b": {"max_length": 99999999, "min_length": 1, "pad_to_max_length": false, "pad_token": 151643}}, "text_sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855", "text_characters": 0, "text_utf8_bytes": 0, "text_words": 0, "actual_token_chunks": {"qwen3vl_4b": {"chunk_lengths": [80], "total_padded_token_slots": 80}}, "tokenization_seconds": 0.125, "tokenizer_class": "Flux3Tokenizer", "prompt_shortened_by_this_adapter": false, "padding_is_not_content_token_count": true, "encoding_status": "completed", "conditioning_shapes": [[1, 80, 20480]], "encoding_seconds": 107.25, "recorded_at": "2026-10-20T23:01:19.412810+00:00"}], "encoder_cached": true, "token_probe_may_be_reused_from_cached_encoding": true, "interpretation": "Tokenprobes are observations fromthe last actualencoding of that text; cachedrequests reuse them. An encoding_seconds value on a reused probe is not new encodingtime for thisrequest. Paddedslots are not content-onlytokens."}, "workflow_path": "workflows/pf-img100-20-t5-s43.json", "workflow_sha256": "1ac73ff33e46c92e9e480622cd213249c197f92baf3911047d9c9056a405e1c4", "source_workflow_canonical_sha256": "7776d92fbc05cdcac37e2e4445f043949cb1390b540d3ebc429847f4c48e0523"} {"id": "cog-03", "label": "CogVideoX 2B local Q8_0 · seed 42", "group": "cogvideox2b-q8-reference-ddim-text-v1", "kind": "video", "model_id": "cog", "model": "zai-org/CogVideoX-2b:Q8_0", "seed": 42, "status": "succeeded", "accepted_result": true, "prompt": "A realistic red fox walks steadily across a snowy clearing from the left side of the frame to the right. Show the fox in full body from the side. Its four legs visibly alternate with each walking step, its body advances across the ground, and its bushy tail swings gently. Near the end the fox stops walking and turns its head toward the camera. The camera stays completely fixed in one wide shot. Continuous visible animal movement throughout the walk. Natural daylight, realistic anatomy, one continuous six-second shot, no cuts, no text.", "prompt_details": {"conditioning_texts": {"4.positive": "Create one video clip from the supplied scene description.\n\nSource generation_brief:\nA realistic red fox walks steadily across a snowy clearing from the left side of the frame to the right. Show the fox in full body from the side. Its four legs visibly alternate with each walking step, its body advances across the ground, and its bushy tail swings gently. Near the end the fox stops walking and turns its head toward the camera. The camera stays completely fixed in one wide shot. Continuous visible animal movement throughout the walk. Natural daylight, realistic anatomy, one continuous six-second shot, no cuts, no text.", "4.negative": "motionless animal, frozen pose, static image, slideshow, distorted anatomy, blurry, flickering, subtitles, watermark"}, "actual_positive_text": "Create one video clip from the supplied scene description.\n\nSource generation_brief:\nA realistic red fox walks steadily across a snowy clearing from the left side of the frame to the right. Show the fox in full body from the side. Its four legs visibly alternate with each walking step, its body advances across the ground, and its bushy tail swings gently. Near the end the fox stops walking and turns its head toward the camera. The camera stays completely fixed in one wide shot. Continuous visible animal movement throughout the walk. Natural daylight, realistic anatomy, one continuous six-second shot, no cuts, no text."}, "parameters": {"2 CLIPLoaderGGUF": {"clip_name": "t5-v1_1-xxl-encoder-Q4_K_M.gguf", "type": "cogvideox"}, "3 VAELoader": {"vae_name": "cogvideox-2b-vae.safetensors"}, "7 EmptyHunyuanLatentVideo": {"width": 512, "height": 320, "length": 49, "batch_size": 1}, "8 PFCogWholeStageReferenceDDIMSampler": {"seed": 42, "steps": 30, "cfg": 6.0}, "90 PFWholeStageDecode": {"tile_size": 256, "overlap": 64, "temporal_size": 32, "temporal_overlap": 8}, "91 CreateVideo": {"fps": 8.0}, "92 SaveVideo": {"format": "mp4", "format.codec": "h264"}}, "timing": {"factory_trace_seconds": 2439.989339, "comfy_server_seconds": 2433.832, "comfy_printed_seconds": null, "stages_seconds": {"text_encoder": 13.53100000001723, "generator": 2277.1089999999967, "video_decoder": 125.875}, "limits": "Factory trace and server totals have different boundaries. Observer node intervals include preparation/event delivery; not isolated kernel or weight-loading times."}, "cache": {"cached_nodes": [], "same_runtime_retained": null, "immediate_warm_claim": null, "context": {}, "idle_gap_after_previous_server_seconds": null}, "resources": {"sample_count": 1187, "scope": "Sampled values; GPU use is device-wide, not continuous per-process physical residency.", "minimum_ram_available_bytes": 10108645376, "minimum_commit_available_bytes": 15845658624, "minimum_system_drive_free_bytes": 20265025536, "maximum_gpu_total_used_mib": 5949}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "quality_scope": "Limited sampled review; no systematic full-resolution/per-trait/full-rate quality score.", "independent_public_copy_decode": true}, "findings": ["Recognizable red fox, snow, distant trees and grass.", "Fox exits at left, leaving the scene empty in the quarter sample.", "Fox reappears at right and travels left, opposite the requested direction.", "A detached or duplicated tail-like region appears at the far right in the last sample.", "Requested end head-turn is not established in the inspected frames."], "media": [{"path": "media/cog-03.mp4", "mime_type": "video/mp4", "size_bytes": 214270, "sha256": "4ab00c3b745f3893bee84552cd7dd261ec113747e2544ecac917be75bd920a4c", "source_sha256": "8bc0fdbf8b06d263cccf83ce36f8ccaf3b232778d88d160ffc7f05c6a30f5741", "metadata_handling": "Container metadata stripped with stream copy; encoded audio/video streams not re-encoded.", "width": 512, "height": 320, "duration_seconds": 6.125, "frames": 49, "fps": 8.0, "public_copy_full_decode_passed": true, "source_content_preserved": "Encodedstream hashes identical aftermetadata-only remux"}], "workflow_path": "workflows/cog-03.json", "workflow_sha256": "fcb4bc417fd364d877497a3677db5ac0dc2cdadfa687465cad91cd5552af07c2", "contact_sheet": "media/cog-03-contact.jpg", "stage_observations": [{"stage": "text_encoder", "event": "complete", "seconds": 13.53100000001723, "torch_peak_allocated_bytes": 3705248256, "torch_peak_reserved_bytes": 4559208448, "cpu_execution_fallback": false}, {"stage": "generator", "event": "complete", "seconds": 2277.1089999999967, "torch_peak_allocated_bytes": 2925896192, "torch_peak_reserved_bytes": 3265265664, "cpu_execution_fallback": false}, {"stage": "video_decoder", "event": "complete", "seconds": 125.875, "torch_peak_allocated_bytes": 5270681088, "torch_peak_reserved_bytes": 8076132352, "cpu_execution_fallback": false}], "source_workflow_canonical_sha256": "c522a48838c0dea4ceb87de9a844223ad3db9652939ac68bf70dc05d5e472c22"} {"id": "cog-04", "label": "CogVideoX 2B local Q8_0 · seed 43", "group": "cogvideox2b-q8-reference-ddim-text-v1", "kind": "video", "model_id": "cog", "model": "zai-org/CogVideoX-2b:Q8_0", "seed": 43, "status": "succeeded", "accepted_result": true, "prompt": "A realistic red fox walks steadily across a snowy clearing from the left side of the frame to the right. Show the fox in full body from the side. Its four legs visibly alternate with each walking step, its body advances across the ground, and its bushy tail swings gently. Near the end the fox stops walking and turns its head toward the camera. The camera stays completely fixed in one wide shot. Continuous visible animal movement throughout the walk. Natural daylight, realistic anatomy, one continuous six-second shot, no cuts, no text.", "prompt_details": {"conditioning_texts": {"4.positive": "Create one video clip from the supplied scene description.\n\nSource generation_brief:\nA realistic red fox walks steadily across a snowy clearing from the left side of the frame to the right. Show the fox in full body from the side. Its four legs visibly alternate with each walking step, its body advances across the ground, and its bushy tail swings gently. Near the end the fox stops walking and turns its head toward the camera. The camera stays completely fixed in one wide shot. Continuous visible animal movement throughout the walk. Natural daylight, realistic anatomy, one continuous six-second shot, no cuts, no text.", "4.negative": "motionless animal, frozen pose, static image, slideshow, distorted anatomy, blurry, flickering, subtitles, watermark"}, "actual_positive_text": "Create one video clip from the supplied scene description.\n\nSource generation_brief:\nA realistic red fox walks steadily across a snowy clearing from the left side of the frame to the right. Show the fox in full body from the side. Its four legs visibly alternate with each walking step, its body advances across the ground, and its bushy tail swings gently. Near the end the fox stops walking and turns its head toward the camera. The camera stays completely fixed in one wide shot. Continuous visible animal movement throughout the walk. Natural daylight, realistic anatomy, one continuous six-second shot, no cuts, no text."}, "parameters": {"2 CLIPLoaderGGUF": {"clip_name": "t5-v1_1-xxl-encoder-Q4_K_M.gguf", "type": "cogvideox"}, "3 VAELoader": {"vae_name": "cogvideox-2b-vae.safetensors"}, "7 EmptyHunyuanLatentVideo": {"width": 512, "height": 320, "length": 49, "batch_size": 1}, "8 PFCogWholeStageReferenceDDIMSampler": {"seed": 43, "steps": 30, "cfg": 6.0}, "90 PFWholeStageDecode": {"tile_size": 256, "overlap": 64, "temporal_size": 32, "temporal_overlap": 8}, "91 CreateVideo": {"fps": 8.0}, "92 SaveVideo": {"format": "mp4", "format.codec": "h264"}}, "timing": {"factory_trace_seconds": 2393.862082, "comfy_server_seconds": 2390.252, "comfy_printed_seconds": null, "stages_seconds": {"generator": 2274.4220000000205, "video_decoder": 109.92099999997299}, "limits": "Factory trace and server totals have different boundaries. Observer node intervals include preparation/event delivery; not isolated kernel or weight-loading times."}, "cache": {"cached_nodes": ["2", "4", "1", "3", "7"], "same_runtime_retained": true, "immediate_warm_claim": null, "context": {}, "idle_gap_after_previous_server_seconds": 46.902}, "resources": {"sample_count": 1165, "scope": "Sampled values; GPU use is device-wide, not continuous per-process physical residency.", "minimum_ram_available_bytes": 9643405312, "minimum_commit_available_bytes": 14991953920, "minimum_system_drive_free_bytes": 20270510080, "maximum_gpu_total_used_mib": 5935}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "quality_scope": "Limited sampled review; no systematic full-resolution/per-trait/full-rate quality score.", "independent_public_copy_decode": true}, "findings": ["Coherent snow and background with a recognizable fox in the three-quarter sample.", "Four of five inspected samples contain only empty snow.", "The fox is strongly cropped at the bottom, without a full-body view or visible four-leg walk.", "Requested continuous full-body walk and end head-turn are not established."], "media": [{"path": "media/cog-04.mp4", "mime_type": "video/mp4", "size_bytes": 186525, "sha256": "190985d4238c7e0e5c51490fe700b18c67e7cfca80bce4cbd7e50ddac7dd7648", "source_sha256": "117e009968ec1f69ad15f789a64b68999749c14c707e2d270c52d5608945c897", "metadata_handling": "Container metadata stripped with stream copy; encoded audio/video streams not re-encoded.", "width": 512, "height": 320, "duration_seconds": 6.125, "frames": 49, "fps": 8.0, "public_copy_full_decode_passed": true, "source_content_preserved": "Encodedstream hashes identical aftermetadata-only remux"}], "workflow_path": "workflows/cog-04.json", "workflow_sha256": "4dc8d577d94aafb5f5ea7f0aff9d9f256281509f31dec8bc8c3167742bd34042", "contact_sheet": "media/cog-04-contact.jpg", "stage_observations": [{"stage": "generator", "event": "complete", "seconds": 2274.4220000000205, "torch_peak_allocated_bytes": 2925896192, "torch_peak_reserved_bytes": 3265265664, "cpu_execution_fallback": false}, {"stage": "video_decoder", "event": "complete", "seconds": 109.92099999997299, "torch_peak_allocated_bytes": 5270681088, "torch_peak_reserved_bytes": 8076132352, "cpu_execution_fallback": false}], "source_workflow_canonical_sha256": "ab6c1546ca57009496c946973c7c6bdf9cf3359119cd420cfe00cb42ad4a606e"} +{"id": "qwen-turbo-01", "label": "Qwen Image 2.1 Turbo Q4_K_M · Initial request including model loading · seed 42", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 42, "status": "succeeded", "accepted_result": true, "prompt": "A small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph.", "prompt_details": {"word_count": 19, "complexity": "Very short / retained runtime pair", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 42}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 334.589337, "comfy_server_seconds": 320.246, "stages_seconds": {"node_4_observer_receive_interval": 4.436999999976251, "node_2_observer_receive_interval": 6.48399999999674, "node_3_observer_receive_interval": 104.32900000002701, "node_12_observer_receive_interval": 0.031000000017229468, "node_11_observer_receive_interval": 0.014999999955762178, "node_1_observer_receive_interval": 1.6570000000065193, "node_5_observer_receive_interval": 0.10899999999674037, "node_10_observer_receive_interval": 0, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 159.85899999999674, "node_7_observer_receive_interval": 40.75, "node_8_observer_receive_interval": 2.047000000020489}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": [], "same_runtime_retained": null, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "Initial request including model loading", "native_inter_run_gap_seconds": null, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": false, "cold_start_scope": "The first request includes initial/model work; OS disk-cache state was not isolated."}}, "resources": {"sampled_summary": {"sample_count": 162, "inflight_sample_count": 161, "minimum_ram_available_bytes": 1912422400, "minimum_commit_available_bytes": 11296174080, "minimum_system_drive_free_bytes": 18670813184, "maximum_sampled_device_gpu_used_mib": 3983}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent fox figurine standing on a pale wooden desk against a cream background; recognizably matches the simple scene.", "Orange fox, desk, cream background and soft studio light are present.", "Fine fur-like ridges dominate the body; ceramic material is not convincingly conveyed.", "Legs, ears and tail are largely plausible for a stylized figurine. Face is slightly toy-like.", "No text requested or visibly generated.", "Ceramic material appears more like textured wood, fur or a molded toy."], "media": [{"path": "media/qwen-turbo-01.png", "mime_type": "image/png", "size_bytes": 261046, "sha256": "56129ce43e04bb2aca5849e23b9d6a0c9adfdee3047a9329f1fcc360c2d9eb2e", "source_sha256": "c4a363cb2d3fbcb5d51b02226bdb57facc1963d98cac306cefb3ad252700b3d0", "width": 512, "height": 512, "pixel_sha256": "5df320c6bdbdad6dc8dbc07af73f2dc513a55deecad9b2f49b659e85ba5ba2f4", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-01.json", "workflow_sha256": "4369239d25533a14c3b5bb3a15d663b4049d3285e9804cc52483e7ffccaf4852", "source_workflow_canonical_sha256": "1ff73d5851a428f07381767c01cfdf4675c6c0b6f37e9df58ff08a9c74de77bb"} +{"id": "qwen-turbo-02", "label": "Qwen Image 2.1 Turbo Q4_K_M · Same brief, next seed · seed 43", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 43, "status": "succeeded", "accepted_result": true, "prompt": "A small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph.", "prompt_details": {"word_count": 19, "complexity": "Very short / retained runtime pair", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 43}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 96.670077, "comfy_server_seconds": 95.234, "stages_seconds": {"node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 67.14100000000326, "node_7_observer_receive_interval": 27.82800000003772, "node_8_observer_receive_interval": 0.06299999996554106}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "10", "11", "12", "2", "3", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": true, "context": {"workload": "Same brief, next seed", "native_inter_run_gap_seconds": 34.322, "only_seed_changed_from_predecessor": true, "text_changed_from_predecessor": false, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 48, "inflight_sample_count": 47, "minimum_ram_available_bytes": 8770334720, "minimum_commit_available_bytes": 11553648640, "minimum_system_drive_free_bytes": 18671108096, "maximum_sampled_device_gpu_used_mib": 4315}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": true, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent front-facing seated fox figurine in the same basic studio setup, with a visibly different pose and face.", "Orange fox, light wood and cream background are present; composition follows the brief.", "Matte textured surface with fur-like ridges; not strongly ceramic.", "Stylized anatomy is mostly coherent. Eye appearance is uneven, with a pale/cyan mark around the left visible eye.", "No text requested or visibly generated.", "Ceramic appearance is weak.", "Small asymmetric eye detail may be a visual artifact."], "media": [{"path": "media/qwen-turbo-02.png", "mime_type": "image/png", "size_bytes": 283217, "sha256": "238d33136a36561317196ed2a9c0096cbafe4109059ab2d0d8df5ec3079d74bd", "source_sha256": "f8bd3fcfba3462d120142e83fb0d7ec637a9eceebd25f6db11f61bda76254eb1", "width": 512, "height": 512, "pixel_sha256": "ac9f30aded9dc1c28b70d2b87cebd61a4df3d93e3dc62206d1ed31874410fe48", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-02.json", "workflow_sha256": "a48c407f70a6f7ffb6722d35f80c1d304116d197f2307961e7f88bbd31456fa4", "source_workflow_canonical_sha256": "84fabe49d8035de7dc9124d16ea9729ca181a0fdbca0f9e681520453dbb87ef2"} +{"id": "qwen-turbo-03", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 44", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 44, "status": "succeeded", "accepted_result": true, "prompt": "A turquoise alpine lake beneath snowy mountains at sunrise, a yellow canoe near the shore, crisp natural landscape photograph.", "prompt_details": {"word_count": 19, "complexity": "Short landscape", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA turquoise alpine lake beneath snowy mountains at sunrise, a yellow canoe near the shore, crisp natural landscape photograph."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA turquoise alpine lake beneath snowy mountains at sunrise, a yellow canoe near the shore, crisp natural landscape photograph."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 44}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 195.952306, "comfy_server_seconds": 194.865, "stages_seconds": {"node_3_observer_receive_interval": 100.20300000003772, "node_10_observer_receive_interval": 0.014999999955762178, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 66.93800000002375, "node_7_observer_receive_interval": 27.64100000000326, "node_8_observer_receive_interval": 0.045999999972991645}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 22.977, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 96, "inflight_sample_count": 95, "minimum_ram_available_bytes": 4403757056, "minimum_commit_available_bytes": 11454955520, "minimum_system_drive_free_bytes": 18670780416, "maximum_sampled_device_gpu_used_mib": 4315}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["Recognizable and visually coherent mountain-lake landscape with a yellow canoe-like object at the near shoreline.", "Turquoise water, snowy mountains and warm sunlight on peaks are present. The yellow canoe is close to shore.", "Water reflections and rock/snow separation are convincing at this size; fine granular texture is visible.", "The canoe is heavily cropped by the lower edge, showing mostly its bow/top; full canoe geometry cannot be assessed.", "No text requested or visibly generated.", "Canoe is strongly cropped and less clearly legible than a full near-shore canoe.", "Fine speckled texture is visible on the mountain surfaces."], "media": [{"path": "media/qwen-turbo-03.png", "mime_type": "image/png", "size_bytes": 474011, "sha256": "57418cb5f47b5647b8773c0e012572e528e150b0f27f422e78f4f05d5a683a11", "source_sha256": "c5e2e3901c3129df9943c2f995d5b42c7d646c708eb6f9053295778ed77012e7", "width": 512, "height": 512, "pixel_sha256": "ff22ec81f55d0251c62e225baca505db986d0d734728cb05f57b4d2087cd12d1", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-03.json", "workflow_sha256": "d26655caaebfa0bd74939c8f5f1977d3759141277d816b6eb1790a1c18f853c1", "source_workflow_canonical_sha256": "96b30904e6a2fa858c6bc001bf5a0b3e4168fbcbc36fcf53b796c7ec9f07cfd2"} +{"id": "qwen-turbo-04", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 45", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 45, "status": "succeeded", "accepted_result": true, "prompt": "Product photograph of a matte black ceramic teapot with a warm copper handle, resting on dark slate beside two pale linen napkins. Three-quarter view, softbox lighting, crisp material detail, uncluttered background.", "prompt_details": {"word_count": 31, "complexity": "Product materials", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nProduct photograph of a matte black ceramic teapot with a warm copper handle, resting on dark slate beside two pale linen napkins. Three-quarter view, softbox lighting, crisp material detail, uncluttered background."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nProduct photograph of a matte black ceramic teapot with a warm copper handle, resting on dark slate beside two pale linen napkins. Three-quarter view, softbox lighting, crisp material detail, uncluttered background."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 45}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 198.952217, "comfy_server_seconds": 196.417, "stages_seconds": {"node_3_observer_receive_interval": 100.10899999999674, "node_10_observer_receive_interval": 0.01600000000325963, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 67.98399999999674, "node_7_observer_receive_interval": 28.21899999998277, "node_8_observer_receive_interval": 0.061999999976251274}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 16.828, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 98, "inflight_sample_count": 97, "minimum_ram_available_bytes": 4163956736, "minimum_commit_available_bytes": 11323981824, "minimum_system_drive_free_bytes": 18670002176, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent, usable-looking product composition of a black teapot, copper-colored handle, dark slate and pale linen.", "Main object, materials, slate and three-quarter product lighting are present. Only one clearly distinguishable rolled/folded napkin appears, rather than two separate napkins.", "Black matte body, metallic handle and woven linen are well differentiated.", "Teapot body, lid, spout and handle joins are broadly plausible; no obvious broken geometry at 512px.", "No text requested or visibly generated.", "Requested count of two napkins is not clearly met."], "media": [{"path": "media/qwen-turbo-04.png", "mime_type": "image/png", "size_bytes": 280344, "sha256": "e5b3093fdb6bb39fcec4b803b1345733cf212bb274806338d27e140d3b882c3d", "source_sha256": "1c2b9ddeca53a1fd944073a596b050e43f8f9b1b4a7e2fee57608095395b97d3", "width": 512, "height": 512, "pixel_sha256": "2db9b0cf39ba6fdb3f0febf48a4f32d9fbd8012ed303d8d3ec711a4a074f8b81", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-04.json", "workflow_sha256": "434243d6ae778adde215a2487cc10e1ee4123a6826f0f50c21e1b41238930d75", "source_workflow_canonical_sha256": "7592e027ed1ef1f6903c6ca2bdae9faad31a02bf99cd7f67c71689cc3427c0ce"} +{"id": "qwen-turbo-05", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 46", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 46, "status": "succeeded", "accepted_result": true, "prompt": "A cozy neighborhood bakery storefront on a rainy evening. The main sign clearly reads \"MORNING BREAD\" in cream block letters. Warm light glows through the windows, a bicycle leans beside the door, wet paving reflects the orange interior, realistic street photograph.", "prompt_details": {"word_count": 41, "complexity": "Readable sign", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA cozy neighborhood bakery storefront on a rainy evening. The main sign clearly reads \"MORNING BREAD\" in cream block letters. Warm light glows through the windows, a bicycle leans beside the door, wet paving reflects the orange interior, realistic street photograph."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA cozy neighborhood bakery storefront on a rainy evening. The main sign clearly reads \"MORNING BREAD\" in cream block letters. Warm light glows through the windows, a bicycle leans beside the door, wet paving reflects the orange interior, realistic street photograph."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 46}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 201.997778, "comfy_server_seconds": 199.64, "stages_seconds": {"node_3_observer_receive_interval": 102.10899999999674, "node_10_observer_receive_interval": 0.01600000000325963, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 68.81200000003446, "node_7_observer_receive_interval": 28.625, "node_8_observer_receive_interval": 0.06299999996554106}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 16.283, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 99, "inflight_sample_count": 98, "minimum_ram_available_bytes": 4295344128, "minimum_commit_available_bytes": 11412852736, "minimum_system_drive_free_bytes": 18669809664, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent rainy-evening bakery storefront with a legible sign, bicycle, warm windows and wet paving.", "The main sign visibly reads exactly MORNING BREAD in large cream block letters. Bicycle, orange interior and reflected light are present.", "Brick, glazing and wet paving read distinctly; the dark storefront feels photograph-like.", "Storefront perspective and bicycle placement are broadly plausible at this scale; fine bicycle detail is somewhat soft.", "Exact requested wording MORNING BREAD is visibly correct. This is a subjective visual transcription, not OCR or an automated text benchmark.", "Fine bicycle/spoke details are soft at the supplied resolution."], "media": [{"path": "media/qwen-turbo-05.png", "mime_type": "image/png", "size_bytes": 434268, "sha256": "94e64b09278ffff4feb04c3197545dd8d6371b281a514d0e3d1cf8550648d919", "source_sha256": "5a9008f775161ef48b16915597d158f4a0cd80883a7c4d452cbf240bba049ce2", "width": 512, "height": 512, "pixel_sha256": "18129fd4fc85277093ff6d330be029055bfde66f2a18af657ce735b6e928f894", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-05.json", "workflow_sha256": "8768f247abc378db4fd3057cb67987743253489f11b42e75bef673f1b6c39658", "source_workflow_canonical_sha256": "874d22b62ef466732130e9e79fa45e12bf32d748a3b1124d21b16aedac7abf59"} +{"id": "qwen-turbo-06", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 47", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 47, "status": "succeeded", "accepted_result": true, "prompt": "A bright Victorian glass greenhouse filled with lush ferns, hanging orchids and terracotta pots. A narrow brick path curves from the bottom left toward an open green door at the far end. Morning sunlight makes geometric shadows through the iron window frames. Include a single brass watering can beside the path and misty trees beyond the glass. Architectural photograph, balanced natural colors, no people or text.", "prompt_details": {"word_count": 66, "complexity": "Botanical architecture", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA bright Victorian glass greenhouse filled with lush ferns, hanging orchids and terracotta pots. A narrow brick path curves from the bottom left toward an open green door at the far end. Morning sunlight makes geometric shadows through the iron window frames. Include a single brass watering can beside the path and misty trees beyond the glass. Architectural photograph, balanced natural colors, no people or text."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA bright Victorian glass greenhouse filled with lush ferns, hanging orchids and terracotta pots. A narrow brick path curves from the bottom left toward an open green door at the far end. Morning sunlight makes geometric shadows through the iron window frames. Include a single brass watering can beside the path and misty trees beyond the glass. Architectural photograph, balanced natural colors, no people or text."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 47}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 201.931, "comfy_server_seconds": 199.412, "stages_seconds": {"node_3_observer_receive_interval": 102.79699999996228, "node_10_observer_receive_interval": 0.015000000013969839, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 68.43800000002375, "node_7_observer_receive_interval": 28.04699999996228, "node_8_observer_receive_interval": 0.062000000034458935}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 21.517, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 99, "inflight_sample_count": 98, "minimum_ram_available_bytes": 4470677504, "minimum_commit_available_bytes": 11370139648, "minimum_system_drive_free_bytes": 18651901952, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent glass greenhouse with dense foliage, terracotta pots, a brick walkway, an open green door and a brass-colored watering can.", "Most named scene elements are present. The path leads toward an open far door; ferns and hanging plants dominate. Distinct hanging orchid flowers are not obvious, and geometric sunlight shadows are subtle.", "Glass/metal frames, foliage, pots and brick are distinguishable, but foliage and frame textures have noticeable fine speckling.", "Overall greenhouse perspective is convincing; dense plant/frame detail is less clean on close inspection. Watering can is partly occluded/near the left edge.", "No people or text requested; none visibly apparent.", "Requested orchids are not clearly identifiable.", "Geometric cast shadows are weaker than the brief suggests.", "Dense foliage and framing show grainy/speckled texture."], "media": [{"path": "media/qwen-turbo-06.png", "mime_type": "image/png", "size_bytes": 649441, "sha256": "f1ae1c3294bcffe1c0d283e957bcc81c5d48df83fe6e0f316eb4182d57db2af1", "source_sha256": "bd1ac8888861a03d6b9a2436cede0ad7b3009f1535c11b360f794ad47a9c0321", "width": 512, "height": 512, "pixel_sha256": "f74bd726756ed8f29d230b4c506533b708c21ecd53275092c3e54672dbb5a1cd", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-06.json", "workflow_sha256": "c48538fbde2bdf10bb5849c8f29aaa79f9f05da8b1e53518af7b56789aa396a0", "source_workflow_canonical_sha256": "6cc4ea59bd3d96d0811844809d534eda117290b8b705f13f0069b45cb5d2da6b"} +{"id": "qwen-turbo-07", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 48", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 48, "status": "succeeded", "accepted_result": true, "prompt": "An editorial portrait of an adult female ceramic artist in her forties working in a sunlit pottery studio. She wears a blue linen apron over a cream shirt and calmly holds a small unfinished bowl in both hands, fingers naturally curled around the rim. Her face is clear and expressive, with a few silver strands in dark curly hair. Shelves of glazed pottery remain softly out of focus behind her. Frame from the waist upward, eye-level camera, gentle window light from the left, documentary realism.", "prompt_details": {"word_count": 85, "complexity": "Portrait / hands", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn editorial portrait of an adult female ceramic artist in her forties working in a sunlit pottery studio. She wears a blue linen apron over a cream shirt and calmly holds a small unfinished bowl in both hands, fingers naturally curled around the rim. Her face is clear and expressive, with a few silver strands in dark curly hair. Shelves of glazed pottery remain softly out of focus behind her. Frame from the waist upward, eye-level camera, gentle window light from the left, documentary realism."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn editorial portrait of an adult female ceramic artist in her forties working in a sunlit pottery studio. She wears a blue linen apron over a cream shirt and calmly holds a small unfinished bowl in both hands, fingers naturally curled around the rim. Her face is clear and expressive, with a few silver strands in dark curly hair. Shelves of glazed pottery remain softly out of focus behind her. Frame from the waist upward, eye-level camera, gentle window light from the left, documentary realism."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 48}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 201.952675, "comfy_server_seconds": 199.784, "stages_seconds": {"node_3_observer_receive_interval": 103.42200000002049, "node_10_observer_receive_interval": 0.015000000013969839, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 68.59399999998277, "node_7_observer_receive_interval": 27.65600000001723, "node_8_observer_receive_interval": 0.0779999999795109}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 13.667, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 99, "inflight_sample_count": 98, "minimum_ram_available_bytes": 4176404480, "minimum_commit_available_bytes": 11358388224, "minimum_system_drive_free_bytes": 18671104000, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent and convincing waist-up studio portrait, with a clear adult face and a bowl held in two hands.", "Blue apron, cream shirt, dark curly hair, pottery shelves, eye-level framing and soft window light from the left all appear. Silver hair strands are not clearly visible. Hands support the bowl from the side and underneath rather than both curling around its rim.", "Cloth texture, skin and unfinished pottery are visibly differentiated; the bowl has a plausible unglazed look.", "Hand shapes, wrist connections and bowl contact look broadly plausible at 512px. Some fingertips overlap or are occluded, so an exact finger-count claim is unwarranted. No obvious extra limb.", "No text requested or visibly generated.", "Specific silver hair detail is not clear.", "The precise requested fingers-around-rim pose is only partly followed."], "media": [{"path": "media/qwen-turbo-07.png", "mime_type": "image/png", "size_bytes": 418100, "sha256": "7b4feea8f758374a5544ad0e9078c8455f32f5a48e8305256ab0839f2afd9472", "source_sha256": "ed1e6c70637edfc74eb3f051c51632087649ce6e88c705b2c4f08f1ee81d756a", "width": 512, "height": 512, "pixel_sha256": "23fe51347e19cff881406c1152d713a2773a46cd9032f54cabc9dc4b5129c108", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-07.json", "workflow_sha256": "23cb0fb14fb4cf31e9a4a67434ac21ab3e74b9858411f7e5724aaf8abbbbb755", "source_workflow_canonical_sha256": "fc32cc9a0f7f8172edfda1400d2b8083ecbf5c86b636b5db6db024c9d20d4142"} +{"id": "qwen-turbo-08", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 49", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 49, "status": "succeeded", "accepted_result": true, "prompt": "A charming small brass robot with a round glass head stands on a wooden step stool inside an enormous old library, carefully returning a red book to a high shelf. The robot has two articulated arms and three visible fingers on each hand. A tiny blue mechanical bird perches on its shoulder. Bookshelves curve toward a domed ceiling painted with constellations; the far aisle is illuminated by warm amber lamps. Render as a detailed hand-painted storybook illustration with teal shadows, golden highlights and a clear focal point. Keep the robot, stool and book visibly separate, with no extra limbs, no captions and no written lettering.", "prompt_details": {"word_count": 105, "complexity": "Stylized story scene", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA charming small brass robot with a round glass head stands on a wooden step stool inside an enormous old library, carefully returning a red book to a high shelf. The robot has two articulated arms and three visible fingers on each hand. A tiny blue mechanical bird perches on its shoulder. Bookshelves curve toward a domed ceiling painted with constellations; the far aisle is illuminated by warm amber lamps. Render as a detailed hand-painted storybook illustration with teal shadows, golden highlights and a clear focal point. Keep the robot, stool and book visibly separate, with no extra limbs, no captions and no written lettering."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA charming small brass robot with a round glass head stands on a wooden step stool inside an enormous old library, carefully returning a red book to a high shelf. The robot has two articulated arms and three visible fingers on each hand. A tiny blue mechanical bird perches on its shoulder. Bookshelves curve toward a domed ceiling painted with constellations; the far aisle is illuminated by warm amber lamps. Render as a detailed hand-painted storybook illustration with teal shadows, golden highlights and a clear focal point. Keep the robot, stool and book visibly separate, with no extra limbs, no captions and no written lettering."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 49}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 210.875124, "comfy_server_seconds": 210.023, "stages_seconds": {"node_3_observer_receive_interval": 107.75, "node_10_observer_receive_interval": 0.01600000000325963, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 72.96900000004098, "node_7_observer_receive_interval": 29.20299999997951, "node_8_observer_receive_interval": 0.061999999976251274}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 14.441, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 104, "inflight_sample_count": 103, "minimum_ram_available_bytes": 4373700608, "minimum_commit_available_bytes": 11362136064, "minimum_system_drive_free_bytes": 18670817280, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent and attractive storybook-style library scene: brass robot, glass-like head, wooden stool, red book, blue shoulder bird and long illuminated aisle are all recognizable.", "The specific main objects, articulated arms, teal/gold palette, warm far lamps and ceiling decoration are strongly represented. The exact three-visible-fingers-per-hand requirement is hard to certify because several small digits overlap or turn away.", "Brass, a reflective glass dome, wood and book covers read distinctly within the painted style.", "Robot, stool, book and shelves remain visibly separate. Arm joints and stool perspective are broadly coherent. Finger-count correctness cannot be established confidently at the supplied resolution.", "No legible caption, title or written lettering is apparent.", "Exact requested finger count remains uncertain; small jointed digits overlap.", "Glass head is mainly reflective/opaque rather than clearly transparent, but still reads as a glass dome."], "media": [{"path": "media/qwen-turbo-08.png", "mime_type": "image/png", "size_bytes": 518714, "sha256": "bd1bcc5079a48be9f59a4d72511eae46abb6e2aa788fe22f72bbac1cc4ba8b7a", "source_sha256": "63abbaf5c49b2baca103bd65761fd4fed09f7e5f173406e91eaaf47d136694f2", "width": 512, "height": 512, "pixel_sha256": "224125f9ca31c3c15dfa975dd61d42fa8f1703d9c012374876d2242054759658", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-08.json", "workflow_sha256": "000bdfef44a50c9cd6275031d4ab57a84f36f216784d1f8c78d637eee60c820a", "source_workflow_canonical_sha256": "3a9b68b7e06cf4b2a375d516a8f5d38b555d7ba5e593354edce22542adf12865"} +{"id": "qwen-turbo-09", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 50", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 50, "status": "succeeded", "accepted_result": true, "prompt": "A detailed travel photograph of a lively Mediterranean open-air produce market in early morning. In the foreground, three wooden crates hold red tomatoes, striped eggplants and yellow lemons, each type kept in its own crate. An adult vendor wearing a plain white shirt stands behind the stall, handing a small paper bag to an adult customer in a green jacket. A striped blue-and-white canopy casts cool shade over the stall. Beyond them, a narrow pedestrian street slopes toward a distant harbor with two white sailboats. Warm sunlight touches the upper walls but the foreground remains softly shaded. Compose the scene in layers with natural human proportions, believable hands, clear separation between people, baskets and produce, and rich but realistic color. Avoid artificial lens flares, watermarks and unreadable signs.", "prompt_details": {"word_count": 128, "complexity": "Crowded composition", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA detailed travel photograph of a lively Mediterranean open-air produce market in early morning. In the foreground, three wooden crates hold red tomatoes, striped eggplants and yellow lemons, each type kept in its own crate. An adult vendor wearing a plain white shirt stands behind the stall, handing a small paper bag to an adult customer in a green jacket. A striped blue-and-white canopy casts cool shade over the stall. Beyond them, a narrow pedestrian street slopes toward a distant harbor with two white sailboats. Warm sunlight touches the upper walls but the foreground remains softly shaded. Compose the scene in layers with natural human proportions, believable hands, clear separation between people, baskets and produce, and rich but realistic color. Avoid artificial lens flares, watermarks and unreadable signs."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA detailed travel photograph of a lively Mediterranean open-air produce market in early morning. In the foreground, three wooden crates hold red tomatoes, striped eggplants and yellow lemons, each type kept in its own crate. An adult vendor wearing a plain white shirt stands behind the stall, handing a small paper bag to an adult customer in a green jacket. A striped blue-and-white canopy casts cool shade over the stall. Beyond them, a narrow pedestrian street slopes toward a distant harbor with two white sailboats. Warm sunlight touches the upper walls but the foreground remains softly shaded. Compose the scene in layers with natural human proportions, believable hands, clear separation between people, baskets and produce, and rich but realistic color. Avoid artificial lens flares, watermarks and unreadable signs."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 50}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 210.965741, "comfy_server_seconds": 208.566, "stages_seconds": {"node_3_observer_receive_interval": 107.82800000003772, "node_10_observer_receive_interval": 0, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 72.20299999997951, "node_7_observer_receive_interval": 28.43800000002375, "node_8_observer_receive_interval": 0.061999999976251274}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 13.101, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 103, "inflight_sample_count": 102, "minimum_ram_available_bytes": 4364185600, "minimum_commit_available_bytes": 11356860416, "minimum_system_drive_free_bytes": 18670198784, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["Recognizable market transaction under a striped canopy, with three separate produce crates and a harbor beyond. Main scene layers are coherent.", "White-shirt vendor, green-jacket customer, blue-and-white canopy, separate tomatoes/eggplants/lemons and sunlit upper wall over shaded foreground are all present. The paper bag and precise handoff are visually ambiguous. Harbor sailboats are visible but the exact requested count of two is not certifiable at this size.", "Vegetables, wood, clothing and canopy fabric are distinguishable. Fine background detail is smeared/brushy despite a photographic brief.", "Main adult body proportions look plausible. The central touching hands are soft and partially merged/occluded; finger anatomy and grasp cannot be verified. Background masts and architecture become muddled.", "No prominent legible sign, watermark or caption appears.", "Paper-bag handoff is less clear than requested.", "Hands and distant harbor details are imprecise.", "Exact two-sailboat count remains uncertain."], "media": [{"path": "media/qwen-turbo-09.png", "mime_type": "image/png", "size_bytes": 437788, "sha256": "847ec6dd9be4503a94967f96ca4b91d223ffb7638749c4fd08fb318723f544e0", "source_sha256": "5b992fa647b0461c1eb8514854a32b5f44cdd297a129a110f72b1abff6a830ef", "width": 512, "height": 512, "pixel_sha256": "ffaa52b367e79221e07aeed4066dcadc2039859372848f38c9bc028d623b7f50", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-09.json", "workflow_sha256": "049b53d4c92125dbe083563cea3d836330cfd864f19317933f71cb8794bc7db6", "source_workflow_canonical_sha256": "e1fd339039ed8d1dde53d4a0668665f4b78470cce872ff0c3f831156539dd328"} +{"id": "qwen-turbo-10", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 51", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 51, "status": "succeeded", "accepted_result": true, "prompt": "Create a highly detailed cinematic architectural illustration of a remote mountain observatory at blue hour after a light snowfall. The building occupies the right two-thirds of the square image: a circular pale-stone base supports one weathered copper dome, whose open viewing slit points toward the upper left. A narrow exterior staircase with dark iron handrails climbs from a small lower entrance to a balcony; every step has a thin undisturbed layer of snow. One adult astronomer in a dark red winter coat stands on the balcony with a small white notebook held closed in one gloved hand, looking toward the sky. Place a warm amber lamp immediately above the entrance, creating a pool of light on the stone without illuminating the whole landscape. In the lower left foreground, a crooked pine tree and two exposed granite rocks frame a winding path that leads toward the building. Behind the observatory, show distant mountain ridges fading from slate blue into violet mist. The sky contains scattered sharp stars and a faint natural Milky Way band, with no oversized planets or fantasy moons. Balance warm interior light against cold ambient snow light, use physically plausible perspective, careful separation of the staircase rails and the figure, and visible differences between copper, stone, glass, fabric and snow. The final scene should feel quiet, inhabited and believable. Do not add titles, labels, logos, watermarks, extra people, extra domes or duplicate staircases.", "prompt_details": {"word_count": 235, "complexity": "Elaborate constraints", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nCreate a highly detailed cinematic architectural illustration of a remote mountain observatory at blue hour after a light snowfall. The building occupies the right two-thirds of the square image: a circular pale-stone base supports one weathered copper dome, whose open viewing slit points toward the upper left. A narrow exterior staircase with dark iron handrails climbs from a small lower entrance to a balcony; every step has a thin undisturbed layer of snow. One adult astronomer in a dark red winter coat stands on the balcony with a small white notebook held closed in one gloved hand, looking toward the sky. Place a warm amber lamp immediately above the entrance, creating a pool of light on the stone without illuminating the whole landscape. In the lower left foreground, a crooked pine tree and two exposed granite rocks frame a winding path that leads toward the building. Behind the observatory, show distant mountain ridges fading from slate blue into violet mist. The sky contains scattered sharp stars and a faint natural Milky Way band, with no oversized planets or fantasy moons. Balance warm interior light against cold ambient snow light, use physically plausible perspective, careful separation of the staircase rails and the figure, and visible differences between copper, stone, glass, fabric and snow. The final scene should feel quiet, inhabited and believable. Do not add titles, labels, logos, watermarks, extra people, extra domes or duplicate staircases."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nCreate a highly detailed cinematic architectural illustration of a remote mountain observatory at blue hour after a light snowfall. The building occupies the right two-thirds of the square image: a circular pale-stone base supports one weathered copper dome, whose open viewing slit points toward the upper left. A narrow exterior staircase with dark iron handrails climbs from a small lower entrance to a balcony; every step has a thin undisturbed layer of snow. One adult astronomer in a dark red winter coat stands on the balcony with a small white notebook held closed in one gloved hand, looking toward the sky. Place a warm amber lamp immediately above the entrance, creating a pool of light on the stone without illuminating the whole landscape. In the lower left foreground, a crooked pine tree and two exposed granite rocks frame a winding path that leads toward the building. Behind the observatory, show distant mountain ridges fading from slate blue into violet mist. The sky contains scattered sharp stars and a faint natural Milky Way band, with no oversized planets or fantasy moons. Balance warm interior light against cold ambient snow light, use physically plausible perspective, careful separation of the staircase rails and the figure, and visible differences between copper, stone, glass, fabric and snow. The final scene should feel quiet, inhabited and believable. Do not add titles, labels, logos, watermarks, extra people, extra domes or duplicate staircases."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 51}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 226.020651, "comfy_server_seconds": 223.135, "stages_seconds": {"node_3_observer_receive_interval": 115.98399999999674, "node_10_observer_receive_interval": 0.01600000000325963, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 78.40600000001723, "node_7_observer_receive_interval": 28.639999999955762, "node_8_observer_receive_interval": 0.07900000002700835}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 10.771, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 111, "inflight_sample_count": 110, "minimum_ram_available_bytes": 4460896256, "minimum_commit_available_bytes": 11257696256, "minimum_system_drive_free_bytes": 18669465600, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent quiet blue-hour observatory scene with one large dome, one red-coated figure, one snow-covered staircase, foreground pine/rocks and distant layered mountains.", "Building placement, open slit toward the upper left, red coat, small light-colored notebook, amber entrance lamp, snow on steps and cold/warm lighting contrast are represented. The figure appears to attend to the notebook rather than clearly look at the sky. Requested stars and faint Milky Way are not visibly present. The exact two-rock count is not cleanly met or certifiable among overlapping foreground boulders.", "Dome reads as dark weathered metal, base as smooth pale masonry/stucco, and snow, coat and rails are distinct. Fine stone material detail is less explicit than the brief.", "Overall building, balcony and staircase perspective are plausible; thin rail details and figure-to-railing separation are imperfect but readable. No obvious extra person, dome or duplicate staircase.", "No title, label, logo or watermark is apparent.", "Stars and Milky Way are absent from the visibly plain sky.", "Figure does not clearly follow the requested skyward gaze.", "Exact rock-count and some fine architectural/material details are weaker than the broad composition."], "media": [{"path": "media/qwen-turbo-10.png", "mime_type": "image/png", "size_bytes": 402891, "sha256": "7bc1033039962507ee2aa153ebfacaa4cdd1f77e93e152bd7241ebae686b85bc", "source_sha256": "f051ec1ab055f23d3fd6513e3428d807734a846f5e51881403cf419ccb33c161", "width": 512, "height": 512, "pixel_sha256": "9672ee9777924c1b99ccff8952362c0119f99a87bb1e22c7bc82a24b579ccb6f", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-10.json", "workflow_sha256": "11fe5261e6abff03fd6a82be45d93b3827c421e3cc3e684f03b72df24694e0b0", "source_workflow_canonical_sha256": "7120b879556ee04ac859a6126f8d1ab24771a28bf208d58fff10b120bf50e68a"} {"id": "wan-failure-01", "label": "Wan2.1 T2V 1.3B · cancelled", "group": "wan21-small-v1", "kind": "video", "model_id": "wan", "model": "Wan-AI/Wan2.1-T2V-1.3B", "seed": 43, "status": "cancelled", "accepted_result": false, "prompt": null, "prompt_details": {"last_known_brief_description": "A blue ceramic fox figurine on a desk beside a plant, with requested camera movement.", "exact_request_text_verified": false, "source": "Retained baseline discussion/approved template; exact submitted manifest not available in the local evidence projection."}, "parameters": {"width": 512, "height": 320, "frames": 81, "fps": 16, "template": "wan21-13b-text-v1", "steps": 20, "cfg": 5, "sampler": "uni_pc", "scheduler": "simple", "math": "FP32", "generator_storage": "BF16 source checkpoint", "placement": "low-VRAM generator, CPU encoder and decoder"}, "timing": {"factory_trace_seconds": null, "stages_seconds": {}, "limits": "No completed output or completed duration; completed stage observations do not establish delivered video."}, "cache": {"immediate_warm_claim": false}, "resources": {"measurement_gap": "No complete per-request RAM/VRAM extrema summary retained in this public projection; failure-specific allocation/guard limits appear in findings."}, "verification": {"full_decode": false, "checksum": false, "acknowledged": false, "trace_attachment": false, "native_experiment": false, "quality_scope": "No accepted media."}, "findings": ["User cancelled the slow FP32 baseline repeat and requested Q8 next."], "media": []} {"id": "ltx-failure-01", "label": "LTX-Video 2B 0.9.8 distilled Q8_0 · cancelled", "group": "ltxv2b-q8-text-v1", "kind": "video", "model_id": "ltx", "model": "Lightricks/LTX-Video:2B-0.9.8-distilled-Q8_0", "seed": 42, "status": "cancelled", "accepted_result": false, "prompt": "A realistic red fox walks steadily across a snowy clearing from the left side of the frame to the right. Show the fox in full body from the side. Its four legs visibly alternate with each walking step, its body advances across the ground, and its bushy tail swings gently. Near the end the fox stops walking and turns its head toward the camera. The camera stays completely fixed in one wide shot. Continuous visible animal movement throughout the walk. Natural daylight, realistic anatomy, one continuous five-second shot, no cuts, no text.", "prompt_details": {"source": "Saved submitted-run ledger."}, "parameters": {"width": 512, "height": 320, "frames": 121, "fps": 24, "template": "ltxv2b-q8-text-v1", "steps": 8, "cfg": 1, "sampler": "euler", "sigmas": "1.0,0.9937,0.9875,0.9812,0.975,0.9094,0.725,0.4219,0.0", "math": "FP32"}, "timing": {"factory_trace_seconds": null, "stages_seconds": {"generation_stage_seconds": 79.98499999998603, "decoder_stage_seconds": 46.48399999999674, "text_encoder": 44.578}, "limits": "No completed output or completed duration; completed stage observations do not establish delivered video."}, "cache": {"immediate_warm_claim": null}, "resources": {"measurement_gap": "No complete per-request RAM/VRAM extrema summary retained in this public projection; failure-specific allocation/guard limits appear in findings."}, "verification": {"full_decode": false, "checksum": false, "acknowledged": false, "trace_attachment": false, "native_experiment": false, "quality_scope": "No accepted media."}, "findings": ["system drive free below6GiB; overlappingCPUCogloaderpreflight also active; avoid overlapping setup in retry"], "media": []} {"id": "ltx-failure-02", "label": "LTX-Video 2B 0.9.8 distilled Q8_0 · cancelled", "group": "ltxv2b-q8-text-v2", "kind": "video", "model_id": "ltx", "model": "Lightricks/LTX-Video:2B-0.9.8-distilled-Q8_0", "seed": 42, "status": "cancelled", "accepted_result": false, "prompt": "A realistic red fox walks steadily across a snowy clearing from the left side of the frame to the right. Show the fox in full body from the side. Its four legs visibly alternate with each walking step, its body advances across the ground, and its bushy tail swings gently. Near the end the fox stops walking and turns its head toward the camera. The camera stays completely fixed in one wide shot. Continuous visible animal movement throughout the walk. Natural daylight, realistic anatomy, one continuous five-second shot, no cuts, no text.", "prompt_details": {"source": "Saved submitted-run ledger."}, "parameters": {"width": 512, "height": 320, "frames": 121, "fps": 24, "template": "ltxv2b-q8-text-v1", "steps": 8, "cfg": 1, "sampler": "euler", "sigmas": "1.0,0.9937,0.9875,0.9812,0.975,0.9094,0.725,0.4219,0.0", "math": "FP32"}, "timing": {"factory_trace_seconds": null, "stages_seconds": {"generation_stage_seconds": 79.71900000004098, "decoder_stage_seconds": 44.46899999998277, "text_encoder": 34.719}, "limits": "No completed output or completed duration; completed stage observations do not establish delivered video."}, "cache": {"immediate_warm_claim": null}, "resources": {"measurement_gap": "No complete per-request RAM/VRAM extrema summary retained in this public projection; failure-specific allocation/guard limits appear in findings."}, "verification": {"full_decode": false, "checksum": false, "acknowledged": false, "trace_attachment": false, "native_experiment": false, "quality_scope": "No accepted media."}, "findings": ["C drive free below6GiB again, without CPU preflight overlap; D scratch and Windows paging capacity configured afterward"], "media": []} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/run.md b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/run.md index 74e0b25..d18ebcc 100644 --- a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/run.md +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/run.md @@ -1,6 +1,6 @@ # Images, music, and the limits of video on a 6 GB GTX 1060 -An older gaming GPU can still produce a useful library of images and short music clips. Video is a much harder proposition. Our October 2026 Packet Foundry tests on a Windows machine with a GTX 1060 6 GB produced 100 images in a planned prompt study, two music clips the user judged usable, and a collection of videos ranging from recognizable but flawed motion to complete colored noise. Every bad clip stays in the archive alongside the better ones. +An older gaming GPU can still produce a useful library of images and short music clips. Our October 2026 Packet Foundry campaign on a Windows GTX 1060 with 6 GB VRAM now retains **133 generated files: 120 images, 11 videos and two music clips**, across **140 recorded attempts**. The images include the 100-request FLUX study, ten earlier checks and ten Qwen Image 2.1 Turbo follow-up images. The user judged both music samples usable; the videos range from recognizable but flawed motion to colored noise. Every poor output and all seven no-output failures remain in the archive. The most useful result is the range: roughly a minute for a 512-pixel image in the large FLUX.3 study, around eleven minutes for fifteen seconds of music, and anything from under two minutes to more than an hour for a short video. Those numbers describe these exact configurations and timing boundaries. They are not a general ranking of the models. @@ -26,7 +26,7 @@ Briefs ranged from 16 to 463 words. Scenes included a fox crossing a creek, tea | 95th percentile | 99.69 s | | First request after the fresh process | 242.34 s | | Median of first seed variants | 74.03 s | -| Median of immediately following variants | 39.40 s | +| Median of next seed variants | 39.40 s | The paired speed difference has concrete cache evidence. Positive conditioning executed fifty times and was reused fifty times; negative conditioning executed once and was reused ninety-nine times. Sampling still ran for each changed seed. Fresh positive-conditioning intervals had an approximate 31.69-second median from recorded event boundaries. Calling every first image “cold” would obscure that most first variants reused an already running model while encoding a new brief. @@ -46,6 +46,29 @@ Earlier fresh-prompt image pairs covered three generators. Their trace times sho These pairs retained the same Comfy process and loader state, with fresh conditioning and sampling on the second request. They used different architectures, step counts and settings, so this table is a record of tested configurations rather than an equal-quality speed contest. Earlier functionality checks, a delivery interruption and an explicit factory-restart recovery test are archived separately instead of being folded into clean pair statistics. +## Generating several variations of one image brief + +A useful production workflow is to ask for five seed variations of one brief and choose the preferred image. Our actual evidence covers **two seed variants per brief**, not five, but it shows why keeping those requests together matters. + +The existing FLUX.3 study gives the stronger repeated evidence: all **50 second requests** reused both positive and negative text conditioning in the same runtime. Every paired graph differed only in sampler seed 42→43, the sampler ran again, and all 50 pairs produced different retained images. Excluding the initial-load pair, the other 49 pairs had a first/new-brief native median of **70.05s** and a following same-brief median of **38.49s**. The median within-pair saving was **32.12s**, with a median per-pair speed ratio of **1.85×**. Those are native server boundaries, separate from the Factory trace table above. + +| Measured FLUX.3 population | First/new-brief native median | Same-brief second native median | +|---|---:|---:| +| All 50 pairs | 71.67s | 38.40s | +| 49 pairs excluding initial loading | 70.05s | 38.49s | + +The actual gap from first completion to the second start ranged from **10.422–31.712s**, with a **22.380s** median. These were consecutive requests with observed cache reuse, not zero-gap batch execution. The measured savings closely track the skipped positive encoding, but we did not randomize seed order or run a matched cache-disabled control. A copied encoder-probe duration on a cached request records its earlier encoding; it is not extra encoder work in the second request. + +### Qwen Image 2.1 Turbo: ten additional images + +We also tested the distinct `unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M` checkpoint at 512×512, eight Euler steps and CFG1, using the exact official ManualSigmas schedule. All ten retained outputs passed the Foundry relay, independent PNG/checksum checks, actual LangSmith attachment retrieval and the native ten-case experiment check. Every original was also inspected visually; successful delivery does not remove material, geometry or prompt-adherence defects, which remain with each image in the gallery. + +Its first image took **320.246s native**, including loading and encoding. The next seed of the same brief took **95.234s** after an actual **34.322s** gap. Text encoding was cached; sampling and decoding executed again. This first-to-second difference combines initial/model-loading and encoding reuse; it does not isolate text-cache benefit alone. The eight following **new briefs**, still using the retained model runtime, had a native median of **199.712s**, range **194.865–223.135s**. Cached seed-variation time therefore must not stand in for fresh-brief latency. + +The practical arrangement is to keep a brief's complete composed prompt, input materials and relevant settings identical and send its seed variations consecutively to the same factory, profile and Comfy runtime. The cache belongs to that runtime; it does not travel inside a Foundry packet. Changing the prompt envelope, switching models or factories, restarting, or evicting cached state may remove the benefit. **No five-variation group or five-image completion time has been measured.** These FLUX and Qwen configurations also differ in architecture and step count; this is not an equal-quality speed comparison. + +[Browse the ten Turbo images](https://benchmarks.weeyuga.com/benchmarks/bd801c85.html) · [Read the sanitized pair evidence and cache findings](https://benchmarks.weeyuga.com/data/predator-media-20261021/seed-variant-cache-findings.json). + ## Music: usable to the listener, expensive per second YuE2 INT8 ConvRot produced two fully decoded fifteen-second stereo FLAC files at 48 kHz. Factory traces took **700.26 and 634.17 seconds**, approximately 11 minutes 40 seconds and 10 minutes 34 seconds. The native graph durations were 696.921 and 630.994 seconds. Loader state remained available between requests; the second request changed the style wording and seed, so semantic generation and acoustic sampling both ran again. @@ -93,4 +116,4 @@ Other retained records include the cancelled YuE2 CPU trial, Cog's initial atten For accepted outputs, the delivery checks verify checksums, full decoding, durable collection and acknowledgement. Saved LangSmith attachment bytes were independently decoded, and native experiment exports were checked where recorded. LangSmith's media-player UI was not separately inspected. The original noisy Cog videos passed file and delivery checks while failing visual usefulness: the separation is deliberate. -The practical lesson from this machine is specific. A repeatable local image workload worked well enough to complete all one hundred planned requests. Two short music samples were usable to their listener, at a substantial time cost. Short video could complete, sometimes quickly, but our tested clips did not establish reliable requested motion. The published prompts, settings, poor outputs and failed attempts give the next builder something more actionable than a success headline. +The practical lesson from this machine is specific. The FLUX study completed all 100 planned image requests, and the Qwen Turbo follow-up completed ten more; those subsets remain separate in the measurements. Two short music samples were usable to their listener, at a substantial time cost. Short video could complete, sometimes quickly, but our tested clips did not establish reliable requested motion. The published prompts, settings, poor outputs and failed attempts give the next builder something more actionable than a success headline. diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/runs/bd801c85-11e3-5033-b076-acd84846b820/metadata.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/runs/bd801c85-11e3-5033-b076-acd84846b820/metadata.json new file mode 100644 index 0000000..fe26c35 --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/runs/bd801c85-11e3-5033-b076-acd84846b820/metadata.json @@ -0,0 +1,51 @@ +{ + "id": "bd801c85-11e3-5033-b076-acd84846b820", + "kind": "media-generation", + "title": "Predator media: Qwen Image 2.1 Turbo Q4 · ten-image seed/cache trial", + "date": "2026-10-21", + "started_at": "2026-10-21T08:41:18.067000+00:00", + "git_sha": "63be2f9b82ad0dac88a7ceaa28a9cf3154125a18", + "hardware": "predator", + "hardware_label": "Predator · GTX 1060 6 GB · 28 GB RAM · Windows", + "engine": "comfyui", + "harness": "packet-foundry-media-v1", + "model_family": "generation", + "model_sizes": [ + "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M" + ], + "task_kind": "media-generation", + "tags": [ + "image", + "video", + "music", + "retained-failures" + ], + "headline": "10 retained outputs · 10 accepted results · 10 recorded attempts · median Factory trace 202.0s", + "janie_blurb_md": "[Browse the complete campaign and media](/benchmarks/predator-media-20261021.html). [Read the findings](/findings/images-video-music-on-6gb.html).", + "janie_blurb_status": "published", + "caveat": "Technical delivery and media decoding do not establish visual or musical quality. Timings include adapter overhead; cache behavior differs between runs. Failed and cancelled attempts remain in the evidence.", + "caveat_severity": "warning", + "methodology_ref": "packet-foundry-media-v1", + "methodology_deviations_md": "Source commit identifies ComfyUI 0.42.1. Private bridge revisions and workflow hashes, where captured, are recorded separately in the campaign evidence. No clean matched-quality comparison or isolated quantization speedup is claimed.", + "results_table": [], + "cells": [], + "n_calls_total": 10, + "n_errors_total": 0, + "chart_spec": null, + "site_grade": "standard", + "site_grade_reason": "Retained media with explicit verification and quality limitations", + "raw_data_urls": { + "jsonl": "/data/predator-media-20261021/runs/bd801c85-11e3-5033-b076-acd84846b820/run.jsonl", + "md": "/data/predator-media-20261021/runs/bd801c85-11e3-5033-b076-acd84846b820/run.md", + "log": "/data/predator-media-20261021/failures.json", + "gitea_dir": "/data/predator-media-20261021/runs/bd801c85-11e3-5033-b076-acd84846b820/metadata.json" + }, + "synthesis_doc": "/findings/images-video-music-on-6gb.html", + "synthesis_docs_all": [ + "/findings/images-video-music-on-6gb.html" + ], + "related_ids": [], + "status": "complete", + "visibility": "public", + "media_group": "qwen21-turbo-q4km-ten" +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/runs/bd801c85-11e3-5033-b076-acd84846b820/run.jsonl b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/runs/bd801c85-11e3-5033-b076-acd84846b820/run.jsonl new file mode 100644 index 0000000..068b7dc --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/runs/bd801c85-11e3-5033-b076-acd84846b820/run.jsonl @@ -0,0 +1,10 @@ +{"id": "qwen-turbo-01", "label": "Qwen Image 2.1 Turbo Q4_K_M · Initial request including model loading · seed 42", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 42, "status": "succeeded", "accepted_result": true, "prompt": "A small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph.", "prompt_details": {"word_count": 19, "complexity": "Very short / retained runtime pair", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 42}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 334.589337, "comfy_server_seconds": 320.246, "stages_seconds": {"node_4_observer_receive_interval": 4.436999999976251, "node_2_observer_receive_interval": 6.48399999999674, "node_3_observer_receive_interval": 104.32900000002701, "node_12_observer_receive_interval": 0.031000000017229468, "node_11_observer_receive_interval": 0.014999999955762178, "node_1_observer_receive_interval": 1.6570000000065193, "node_5_observer_receive_interval": 0.10899999999674037, "node_10_observer_receive_interval": 0, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 159.85899999999674, "node_7_observer_receive_interval": 40.75, "node_8_observer_receive_interval": 2.047000000020489}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": [], "same_runtime_retained": null, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "Initial request including model loading", "native_inter_run_gap_seconds": null, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": false, "cold_start_scope": "The first request includes initial/model work; OS disk-cache state was not isolated."}}, "resources": {"sampled_summary": {"sample_count": 162, "inflight_sample_count": 161, "minimum_ram_available_bytes": 1912422400, "minimum_commit_available_bytes": 11296174080, "minimum_system_drive_free_bytes": 18670813184, "maximum_sampled_device_gpu_used_mib": 3983}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent fox figurine standing on a pale wooden desk against a cream background; recognizably matches the simple scene.", "Orange fox, desk, cream background and soft studio light are present.", "Fine fur-like ridges dominate the body; ceramic material is not convincingly conveyed.", "Legs, ears and tail are largely plausible for a stylized figurine. Face is slightly toy-like.", "No text requested or visibly generated.", "Ceramic material appears more like textured wood, fur or a molded toy."], "media": [{"path": "media/qwen-turbo-01.png", "mime_type": "image/png", "size_bytes": 261046, "sha256": "56129ce43e04bb2aca5849e23b9d6a0c9adfdee3047a9329f1fcc360c2d9eb2e", "source_sha256": "c4a363cb2d3fbcb5d51b02226bdb57facc1963d98cac306cefb3ad252700b3d0", "width": 512, "height": 512, "pixel_sha256": "5df320c6bdbdad6dc8dbc07af73f2dc513a55deecad9b2f49b659e85ba5ba2f4", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-01.json", "workflow_sha256": "4369239d25533a14c3b5bb3a15d663b4049d3285e9804cc52483e7ffccaf4852", "source_workflow_canonical_sha256": "1ff73d5851a428f07381767c01cfdf4675c6c0b6f37e9df58ff08a9c74de77bb"} +{"id": "qwen-turbo-02", "label": "Qwen Image 2.1 Turbo Q4_K_M · Same brief, next seed · seed 43", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 43, "status": "succeeded", "accepted_result": true, "prompt": "A small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph.", "prompt_details": {"word_count": 19, "complexity": "Very short / retained runtime pair", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 43}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 96.670077, "comfy_server_seconds": 95.234, "stages_seconds": {"node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 67.14100000000326, "node_7_observer_receive_interval": 27.82800000003772, "node_8_observer_receive_interval": 0.06299999996554106}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "10", "11", "12", "2", "3", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": true, "context": {"workload": "Same brief, next seed", "native_inter_run_gap_seconds": 34.322, "only_seed_changed_from_predecessor": true, "text_changed_from_predecessor": false, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 48, "inflight_sample_count": 47, "minimum_ram_available_bytes": 8770334720, "minimum_commit_available_bytes": 11553648640, "minimum_system_drive_free_bytes": 18671108096, "maximum_sampled_device_gpu_used_mib": 4315}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": true, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent front-facing seated fox figurine in the same basic studio setup, with a visibly different pose and face.", "Orange fox, light wood and cream background are present; composition follows the brief.", "Matte textured surface with fur-like ridges; not strongly ceramic.", "Stylized anatomy is mostly coherent. Eye appearance is uneven, with a pale/cyan mark around the left visible eye.", "No text requested or visibly generated.", "Ceramic appearance is weak.", "Small asymmetric eye detail may be a visual artifact."], "media": [{"path": "media/qwen-turbo-02.png", "mime_type": "image/png", "size_bytes": 283217, "sha256": "238d33136a36561317196ed2a9c0096cbafe4109059ab2d0d8df5ec3079d74bd", "source_sha256": "f8bd3fcfba3462d120142e83fb0d7ec637a9eceebd25f6db11f61bda76254eb1", "width": 512, "height": 512, "pixel_sha256": "ac9f30aded9dc1c28b70d2b87cebd61a4df3d93e3dc62206d1ed31874410fe48", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-02.json", "workflow_sha256": "a48c407f70a6f7ffb6722d35f80c1d304116d197f2307961e7f88bbd31456fa4", "source_workflow_canonical_sha256": "84fabe49d8035de7dc9124d16ea9729ca181a0fdbca0f9e681520453dbb87ef2"} +{"id": "qwen-turbo-03", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 44", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 44, "status": "succeeded", "accepted_result": true, "prompt": "A turquoise alpine lake beneath snowy mountains at sunrise, a yellow canoe near the shore, crisp natural landscape photograph.", "prompt_details": {"word_count": 19, "complexity": "Short landscape", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA turquoise alpine lake beneath snowy mountains at sunrise, a yellow canoe near the shore, crisp natural landscape photograph."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA turquoise alpine lake beneath snowy mountains at sunrise, a yellow canoe near the shore, crisp natural landscape photograph."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 44}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 195.952306, "comfy_server_seconds": 194.865, "stages_seconds": {"node_3_observer_receive_interval": 100.20300000003772, "node_10_observer_receive_interval": 0.014999999955762178, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 66.93800000002375, "node_7_observer_receive_interval": 27.64100000000326, "node_8_observer_receive_interval": 0.045999999972991645}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 22.977, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 96, "inflight_sample_count": 95, "minimum_ram_available_bytes": 4403757056, "minimum_commit_available_bytes": 11454955520, "minimum_system_drive_free_bytes": 18670780416, "maximum_sampled_device_gpu_used_mib": 4315}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["Recognizable and visually coherent mountain-lake landscape with a yellow canoe-like object at the near shoreline.", "Turquoise water, snowy mountains and warm sunlight on peaks are present. The yellow canoe is close to shore.", "Water reflections and rock/snow separation are convincing at this size; fine granular texture is visible.", "The canoe is heavily cropped by the lower edge, showing mostly its bow/top; full canoe geometry cannot be assessed.", "No text requested or visibly generated.", "Canoe is strongly cropped and less clearly legible than a full near-shore canoe.", "Fine speckled texture is visible on the mountain surfaces."], "media": [{"path": "media/qwen-turbo-03.png", "mime_type": "image/png", "size_bytes": 474011, "sha256": "57418cb5f47b5647b8773c0e012572e528e150b0f27f422e78f4f05d5a683a11", "source_sha256": "c5e2e3901c3129df9943c2f995d5b42c7d646c708eb6f9053295778ed77012e7", "width": 512, "height": 512, "pixel_sha256": "ff22ec81f55d0251c62e225baca505db986d0d734728cb05f57b4d2087cd12d1", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-03.json", "workflow_sha256": "d26655caaebfa0bd74939c8f5f1977d3759141277d816b6eb1790a1c18f853c1", "source_workflow_canonical_sha256": "96b30904e6a2fa858c6bc001bf5a0b3e4168fbcbc36fcf53b796c7ec9f07cfd2"} +{"id": "qwen-turbo-04", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 45", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 45, "status": "succeeded", "accepted_result": true, "prompt": "Product photograph of a matte black ceramic teapot with a warm copper handle, resting on dark slate beside two pale linen napkins. Three-quarter view, softbox lighting, crisp material detail, uncluttered background.", "prompt_details": {"word_count": 31, "complexity": "Product materials", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nProduct photograph of a matte black ceramic teapot with a warm copper handle, resting on dark slate beside two pale linen napkins. Three-quarter view, softbox lighting, crisp material detail, uncluttered background."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nProduct photograph of a matte black ceramic teapot with a warm copper handle, resting on dark slate beside two pale linen napkins. Three-quarter view, softbox lighting, crisp material detail, uncluttered background."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 45}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 198.952217, "comfy_server_seconds": 196.417, "stages_seconds": {"node_3_observer_receive_interval": 100.10899999999674, "node_10_observer_receive_interval": 0.01600000000325963, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 67.98399999999674, "node_7_observer_receive_interval": 28.21899999998277, "node_8_observer_receive_interval": 0.061999999976251274}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 16.828, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 98, "inflight_sample_count": 97, "minimum_ram_available_bytes": 4163956736, "minimum_commit_available_bytes": 11323981824, "minimum_system_drive_free_bytes": 18670002176, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent, usable-looking product composition of a black teapot, copper-colored handle, dark slate and pale linen.", "Main object, materials, slate and three-quarter product lighting are present. Only one clearly distinguishable rolled/folded napkin appears, rather than two separate napkins.", "Black matte body, metallic handle and woven linen are well differentiated.", "Teapot body, lid, spout and handle joins are broadly plausible; no obvious broken geometry at 512px.", "No text requested or visibly generated.", "Requested count of two napkins is not clearly met."], "media": [{"path": "media/qwen-turbo-04.png", "mime_type": "image/png", "size_bytes": 280344, "sha256": "e5b3093fdb6bb39fcec4b803b1345733cf212bb274806338d27e140d3b882c3d", "source_sha256": "1c2b9ddeca53a1fd944073a596b050e43f8f9b1b4a7e2fee57608095395b97d3", "width": 512, "height": 512, "pixel_sha256": "2db9b0cf39ba6fdb3f0febf48a4f32d9fbd8012ed303d8d3ec711a4a074f8b81", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-04.json", "workflow_sha256": "434243d6ae778adde215a2487cc10e1ee4123a6826f0f50c21e1b41238930d75", "source_workflow_canonical_sha256": "7592e027ed1ef1f6903c6ca2bdae9faad31a02bf99cd7f67c71689cc3427c0ce"} +{"id": "qwen-turbo-05", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 46", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 46, "status": "succeeded", "accepted_result": true, "prompt": "A cozy neighborhood bakery storefront on a rainy evening. The main sign clearly reads \"MORNING BREAD\" in cream block letters. Warm light glows through the windows, a bicycle leans beside the door, wet paving reflects the orange interior, realistic street photograph.", "prompt_details": {"word_count": 41, "complexity": "Readable sign", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA cozy neighborhood bakery storefront on a rainy evening. The main sign clearly reads \"MORNING BREAD\" in cream block letters. Warm light glows through the windows, a bicycle leans beside the door, wet paving reflects the orange interior, realistic street photograph."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA cozy neighborhood bakery storefront on a rainy evening. The main sign clearly reads \"MORNING BREAD\" in cream block letters. Warm light glows through the windows, a bicycle leans beside the door, wet paving reflects the orange interior, realistic street photograph."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 46}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 201.997778, "comfy_server_seconds": 199.64, "stages_seconds": {"node_3_observer_receive_interval": 102.10899999999674, "node_10_observer_receive_interval": 0.01600000000325963, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 68.81200000003446, "node_7_observer_receive_interval": 28.625, "node_8_observer_receive_interval": 0.06299999996554106}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 16.283, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 99, "inflight_sample_count": 98, "minimum_ram_available_bytes": 4295344128, "minimum_commit_available_bytes": 11412852736, "minimum_system_drive_free_bytes": 18669809664, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent rainy-evening bakery storefront with a legible sign, bicycle, warm windows and wet paving.", "The main sign visibly reads exactly MORNING BREAD in large cream block letters. Bicycle, orange interior and reflected light are present.", "Brick, glazing and wet paving read distinctly; the dark storefront feels photograph-like.", "Storefront perspective and bicycle placement are broadly plausible at this scale; fine bicycle detail is somewhat soft.", "Exact requested wording MORNING BREAD is visibly correct. This is a subjective visual transcription, not OCR or an automated text benchmark.", "Fine bicycle/spoke details are soft at the supplied resolution."], "media": [{"path": "media/qwen-turbo-05.png", "mime_type": "image/png", "size_bytes": 434268, "sha256": "94e64b09278ffff4feb04c3197545dd8d6371b281a514d0e3d1cf8550648d919", "source_sha256": "5a9008f775161ef48b16915597d158f4a0cd80883a7c4d452cbf240bba049ce2", "width": 512, "height": 512, "pixel_sha256": "18129fd4fc85277093ff6d330be029055bfde66f2a18af657ce735b6e928f894", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-05.json", "workflow_sha256": "8768f247abc378db4fd3057cb67987743253489f11b42e75bef673f1b6c39658", "source_workflow_canonical_sha256": "874d22b62ef466732130e9e79fa45e12bf32d748a3b1124d21b16aedac7abf59"} +{"id": "qwen-turbo-06", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 47", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 47, "status": "succeeded", "accepted_result": true, "prompt": "A bright Victorian glass greenhouse filled with lush ferns, hanging orchids and terracotta pots. A narrow brick path curves from the bottom left toward an open green door at the far end. Morning sunlight makes geometric shadows through the iron window frames. Include a single brass watering can beside the path and misty trees beyond the glass. Architectural photograph, balanced natural colors, no people or text.", "prompt_details": {"word_count": 66, "complexity": "Botanical architecture", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA bright Victorian glass greenhouse filled with lush ferns, hanging orchids and terracotta pots. A narrow brick path curves from the bottom left toward an open green door at the far end. Morning sunlight makes geometric shadows through the iron window frames. Include a single brass watering can beside the path and misty trees beyond the glass. Architectural photograph, balanced natural colors, no people or text."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA bright Victorian glass greenhouse filled with lush ferns, hanging orchids and terracotta pots. A narrow brick path curves from the bottom left toward an open green door at the far end. Morning sunlight makes geometric shadows through the iron window frames. Include a single brass watering can beside the path and misty trees beyond the glass. Architectural photograph, balanced natural colors, no people or text."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 47}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 201.931, "comfy_server_seconds": 199.412, "stages_seconds": {"node_3_observer_receive_interval": 102.79699999996228, "node_10_observer_receive_interval": 0.015000000013969839, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 68.43800000002375, "node_7_observer_receive_interval": 28.04699999996228, "node_8_observer_receive_interval": 0.062000000034458935}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 21.517, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 99, "inflight_sample_count": 98, "minimum_ram_available_bytes": 4470677504, "minimum_commit_available_bytes": 11370139648, "minimum_system_drive_free_bytes": 18651901952, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent glass greenhouse with dense foliage, terracotta pots, a brick walkway, an open green door and a brass-colored watering can.", "Most named scene elements are present. The path leads toward an open far door; ferns and hanging plants dominate. Distinct hanging orchid flowers are not obvious, and geometric sunlight shadows are subtle.", "Glass/metal frames, foliage, pots and brick are distinguishable, but foliage and frame textures have noticeable fine speckling.", "Overall greenhouse perspective is convincing; dense plant/frame detail is less clean on close inspection. Watering can is partly occluded/near the left edge.", "No people or text requested; none visibly apparent.", "Requested orchids are not clearly identifiable.", "Geometric cast shadows are weaker than the brief suggests.", "Dense foliage and framing show grainy/speckled texture."], "media": [{"path": "media/qwen-turbo-06.png", "mime_type": "image/png", "size_bytes": 649441, "sha256": "f1ae1c3294bcffe1c0d283e957bcc81c5d48df83fe6e0f316eb4182d57db2af1", "source_sha256": "bd1ac8888861a03d6b9a2436cede0ad7b3009f1535c11b360f794ad47a9c0321", "width": 512, "height": 512, "pixel_sha256": "f74bd726756ed8f29d230b4c506533b708c21ecd53275092c3e54672dbb5a1cd", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-06.json", "workflow_sha256": "c48538fbde2bdf10bb5849c8f29aaa79f9f05da8b1e53518af7b56789aa396a0", "source_workflow_canonical_sha256": "6cc4ea59bd3d96d0811844809d534eda117290b8b705f13f0069b45cb5d2da6b"} +{"id": "qwen-turbo-07", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 48", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 48, "status": "succeeded", "accepted_result": true, "prompt": "An editorial portrait of an adult female ceramic artist in her forties working in a sunlit pottery studio. She wears a blue linen apron over a cream shirt and calmly holds a small unfinished bowl in both hands, fingers naturally curled around the rim. Her face is clear and expressive, with a few silver strands in dark curly hair. Shelves of glazed pottery remain softly out of focus behind her. Frame from the waist upward, eye-level camera, gentle window light from the left, documentary realism.", "prompt_details": {"word_count": 85, "complexity": "Portrait / hands", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn editorial portrait of an adult female ceramic artist in her forties working in a sunlit pottery studio. She wears a blue linen apron over a cream shirt and calmly holds a small unfinished bowl in both hands, fingers naturally curled around the rim. Her face is clear and expressive, with a few silver strands in dark curly hair. Shelves of glazed pottery remain softly out of focus behind her. Frame from the waist upward, eye-level camera, gentle window light from the left, documentary realism."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn editorial portrait of an adult female ceramic artist in her forties working in a sunlit pottery studio. She wears a blue linen apron over a cream shirt and calmly holds a small unfinished bowl in both hands, fingers naturally curled around the rim. Her face is clear and expressive, with a few silver strands in dark curly hair. Shelves of glazed pottery remain softly out of focus behind her. Frame from the waist upward, eye-level camera, gentle window light from the left, documentary realism."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 48}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 201.952675, "comfy_server_seconds": 199.784, "stages_seconds": {"node_3_observer_receive_interval": 103.42200000002049, "node_10_observer_receive_interval": 0.015000000013969839, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 68.59399999998277, "node_7_observer_receive_interval": 27.65600000001723, "node_8_observer_receive_interval": 0.0779999999795109}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 13.667, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 99, "inflight_sample_count": 98, "minimum_ram_available_bytes": 4176404480, "minimum_commit_available_bytes": 11358388224, "minimum_system_drive_free_bytes": 18671104000, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent and convincing waist-up studio portrait, with a clear adult face and a bowl held in two hands.", "Blue apron, cream shirt, dark curly hair, pottery shelves, eye-level framing and soft window light from the left all appear. Silver hair strands are not clearly visible. Hands support the bowl from the side and underneath rather than both curling around its rim.", "Cloth texture, skin and unfinished pottery are visibly differentiated; the bowl has a plausible unglazed look.", "Hand shapes, wrist connections and bowl contact look broadly plausible at 512px. Some fingertips overlap or are occluded, so an exact finger-count claim is unwarranted. No obvious extra limb.", "No text requested or visibly generated.", "Specific silver hair detail is not clear.", "The precise requested fingers-around-rim pose is only partly followed."], "media": [{"path": "media/qwen-turbo-07.png", "mime_type": "image/png", "size_bytes": 418100, "sha256": "7b4feea8f758374a5544ad0e9078c8455f32f5a48e8305256ab0839f2afd9472", "source_sha256": "ed1e6c70637edfc74eb3f051c51632087649ce6e88c705b2c4f08f1ee81d756a", "width": 512, "height": 512, "pixel_sha256": "23fe51347e19cff881406c1152d713a2773a46cd9032f54cabc9dc4b5129c108", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-07.json", "workflow_sha256": "23cb0fb14fb4cf31e9a4a67434ac21ab3e74b9858411f7e5724aaf8abbbbb755", "source_workflow_canonical_sha256": "fc32cc9a0f7f8172edfda1400d2b8083ecbf5c86b636b5db6db024c9d20d4142"} +{"id": "qwen-turbo-08", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 49", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 49, "status": "succeeded", "accepted_result": true, "prompt": "A charming small brass robot with a round glass head stands on a wooden step stool inside an enormous old library, carefully returning a red book to a high shelf. The robot has two articulated arms and three visible fingers on each hand. A tiny blue mechanical bird perches on its shoulder. Bookshelves curve toward a domed ceiling painted with constellations; the far aisle is illuminated by warm amber lamps. Render as a detailed hand-painted storybook illustration with teal shadows, golden highlights and a clear focal point. Keep the robot, stool and book visibly separate, with no extra limbs, no captions and no written lettering.", "prompt_details": {"word_count": 105, "complexity": "Stylized story scene", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA charming small brass robot with a round glass head stands on a wooden step stool inside an enormous old library, carefully returning a red book to a high shelf. The robot has two articulated arms and three visible fingers on each hand. A tiny blue mechanical bird perches on its shoulder. Bookshelves curve toward a domed ceiling painted with constellations; the far aisle is illuminated by warm amber lamps. Render as a detailed hand-painted storybook illustration with teal shadows, golden highlights and a clear focal point. Keep the robot, stool and book visibly separate, with no extra limbs, no captions and no written lettering."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA charming small brass robot with a round glass head stands on a wooden step stool inside an enormous old library, carefully returning a red book to a high shelf. The robot has two articulated arms and three visible fingers on each hand. A tiny blue mechanical bird perches on its shoulder. Bookshelves curve toward a domed ceiling painted with constellations; the far aisle is illuminated by warm amber lamps. Render as a detailed hand-painted storybook illustration with teal shadows, golden highlights and a clear focal point. Keep the robot, stool and book visibly separate, with no extra limbs, no captions and no written lettering."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 49}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 210.875124, "comfy_server_seconds": 210.023, "stages_seconds": {"node_3_observer_receive_interval": 107.75, "node_10_observer_receive_interval": 0.01600000000325963, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 72.96900000004098, "node_7_observer_receive_interval": 29.20299999997951, "node_8_observer_receive_interval": 0.061999999976251274}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 14.441, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 104, "inflight_sample_count": 103, "minimum_ram_available_bytes": 4373700608, "minimum_commit_available_bytes": 11362136064, "minimum_system_drive_free_bytes": 18670817280, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent and attractive storybook-style library scene: brass robot, glass-like head, wooden stool, red book, blue shoulder bird and long illuminated aisle are all recognizable.", "The specific main objects, articulated arms, teal/gold palette, warm far lamps and ceiling decoration are strongly represented. The exact three-visible-fingers-per-hand requirement is hard to certify because several small digits overlap or turn away.", "Brass, a reflective glass dome, wood and book covers read distinctly within the painted style.", "Robot, stool, book and shelves remain visibly separate. Arm joints and stool perspective are broadly coherent. Finger-count correctness cannot be established confidently at the supplied resolution.", "No legible caption, title or written lettering is apparent.", "Exact requested finger count remains uncertain; small jointed digits overlap.", "Glass head is mainly reflective/opaque rather than clearly transparent, but still reads as a glass dome."], "media": [{"path": "media/qwen-turbo-08.png", "mime_type": "image/png", "size_bytes": 518714, "sha256": "bd1bcc5079a48be9f59a4d72511eae46abb6e2aa788fe22f72bbac1cc4ba8b7a", "source_sha256": "63abbaf5c49b2baca103bd65761fd4fed09f7e5f173406e91eaaf47d136694f2", "width": 512, "height": 512, "pixel_sha256": "224125f9ca31c3c15dfa975dd61d42fa8f1703d9c012374876d2242054759658", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-08.json", "workflow_sha256": "000bdfef44a50c9cd6275031d4ab57a84f36f216784d1f8c78d637eee60c820a", "source_workflow_canonical_sha256": "3a9b68b7e06cf4b2a375d516a8f5d38b555d7ba5e593354edce22542adf12865"} +{"id": "qwen-turbo-09", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 50", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 50, "status": "succeeded", "accepted_result": true, "prompt": "A detailed travel photograph of a lively Mediterranean open-air produce market in early morning. In the foreground, three wooden crates hold red tomatoes, striped eggplants and yellow lemons, each type kept in its own crate. An adult vendor wearing a plain white shirt stands behind the stall, handing a small paper bag to an adult customer in a green jacket. A striped blue-and-white canopy casts cool shade over the stall. Beyond them, a narrow pedestrian street slopes toward a distant harbor with two white sailboats. Warm sunlight touches the upper walls but the foreground remains softly shaded. Compose the scene in layers with natural human proportions, believable hands, clear separation between people, baskets and produce, and rich but realistic color. Avoid artificial lens flares, watermarks and unreadable signs.", "prompt_details": {"word_count": 128, "complexity": "Crowded composition", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA detailed travel photograph of a lively Mediterranean open-air produce market in early morning. In the foreground, three wooden crates hold red tomatoes, striped eggplants and yellow lemons, each type kept in its own crate. An adult vendor wearing a plain white shirt stands behind the stall, handing a small paper bag to an adult customer in a green jacket. A striped blue-and-white canopy casts cool shade over the stall. Beyond them, a narrow pedestrian street slopes toward a distant harbor with two white sailboats. Warm sunlight touches the upper walls but the foreground remains softly shaded. Compose the scene in layers with natural human proportions, believable hands, clear separation between people, baskets and produce, and rich but realistic color. Avoid artificial lens flares, watermarks and unreadable signs."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA detailed travel photograph of a lively Mediterranean open-air produce market in early morning. In the foreground, three wooden crates hold red tomatoes, striped eggplants and yellow lemons, each type kept in its own crate. An adult vendor wearing a plain white shirt stands behind the stall, handing a small paper bag to an adult customer in a green jacket. A striped blue-and-white canopy casts cool shade over the stall. Beyond them, a narrow pedestrian street slopes toward a distant harbor with two white sailboats. Warm sunlight touches the upper walls but the foreground remains softly shaded. Compose the scene in layers with natural human proportions, believable hands, clear separation between people, baskets and produce, and rich but realistic color. Avoid artificial lens flares, watermarks and unreadable signs."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 50}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 210.965741, "comfy_server_seconds": 208.566, "stages_seconds": {"node_3_observer_receive_interval": 107.82800000003772, "node_10_observer_receive_interval": 0, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 72.20299999997951, "node_7_observer_receive_interval": 28.43800000002375, "node_8_observer_receive_interval": 0.061999999976251274}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 13.101, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 103, "inflight_sample_count": 102, "minimum_ram_available_bytes": 4364185600, "minimum_commit_available_bytes": 11356860416, "minimum_system_drive_free_bytes": 18670198784, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["Recognizable market transaction under a striped canopy, with three separate produce crates and a harbor beyond. Main scene layers are coherent.", "White-shirt vendor, green-jacket customer, blue-and-white canopy, separate tomatoes/eggplants/lemons and sunlit upper wall over shaded foreground are all present. The paper bag and precise handoff are visually ambiguous. Harbor sailboats are visible but the exact requested count of two is not certifiable at this size.", "Vegetables, wood, clothing and canopy fabric are distinguishable. Fine background detail is smeared/brushy despite a photographic brief.", "Main adult body proportions look plausible. The central touching hands are soft and partially merged/occluded; finger anatomy and grasp cannot be verified. Background masts and architecture become muddled.", "No prominent legible sign, watermark or caption appears.", "Paper-bag handoff is less clear than requested.", "Hands and distant harbor details are imprecise.", "Exact two-sailboat count remains uncertain."], "media": [{"path": "media/qwen-turbo-09.png", "mime_type": "image/png", "size_bytes": 437788, "sha256": "847ec6dd9be4503a94967f96ca4b91d223ffb7638749c4fd08fb318723f544e0", "source_sha256": "5b992fa647b0461c1eb8514854a32b5f44cdd297a129a110f72b1abff6a830ef", "width": 512, "height": 512, "pixel_sha256": "ffaa52b367e79221e07aeed4066dcadc2039859372848f38c9bc028d623b7f50", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-09.json", "workflow_sha256": "049b53d4c92125dbe083563cea3d836330cfd864f19317933f71cb8794bc7db6", "source_workflow_canonical_sha256": "e1fd339039ed8d1dde53d4a0668665f4b78470cce872ff0c3f831156539dd328"} +{"id": "qwen-turbo-10", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 51", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 51, "status": "succeeded", "accepted_result": true, "prompt": "Create a highly detailed cinematic architectural illustration of a remote mountain observatory at blue hour after a light snowfall. The building occupies the right two-thirds of the square image: a circular pale-stone base supports one weathered copper dome, whose open viewing slit points toward the upper left. A narrow exterior staircase with dark iron handrails climbs from a small lower entrance to a balcony; every step has a thin undisturbed layer of snow. One adult astronomer in a dark red winter coat stands on the balcony with a small white notebook held closed in one gloved hand, looking toward the sky. Place a warm amber lamp immediately above the entrance, creating a pool of light on the stone without illuminating the whole landscape. In the lower left foreground, a crooked pine tree and two exposed granite rocks frame a winding path that leads toward the building. Behind the observatory, show distant mountain ridges fading from slate blue into violet mist. The sky contains scattered sharp stars and a faint natural Milky Way band, with no oversized planets or fantasy moons. Balance warm interior light against cold ambient snow light, use physically plausible perspective, careful separation of the staircase rails and the figure, and visible differences between copper, stone, glass, fabric and snow. The final scene should feel quiet, inhabited and believable. Do not add titles, labels, logos, watermarks, extra people, extra domes or duplicate staircases.", "prompt_details": {"word_count": 235, "complexity": "Elaborate constraints", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nCreate a highly detailed cinematic architectural illustration of a remote mountain observatory at blue hour after a light snowfall. The building occupies the right two-thirds of the square image: a circular pale-stone base supports one weathered copper dome, whose open viewing slit points toward the upper left. A narrow exterior staircase with dark iron handrails climbs from a small lower entrance to a balcony; every step has a thin undisturbed layer of snow. One adult astronomer in a dark red winter coat stands on the balcony with a small white notebook held closed in one gloved hand, looking toward the sky. Place a warm amber lamp immediately above the entrance, creating a pool of light on the stone without illuminating the whole landscape. In the lower left foreground, a crooked pine tree and two exposed granite rocks frame a winding path that leads toward the building. Behind the observatory, show distant mountain ridges fading from slate blue into violet mist. The sky contains scattered sharp stars and a faint natural Milky Way band, with no oversized planets or fantasy moons. Balance warm interior light against cold ambient snow light, use physically plausible perspective, careful separation of the staircase rails and the figure, and visible differences between copper, stone, glass, fabric and snow. The final scene should feel quiet, inhabited and believable. Do not add titles, labels, logos, watermarks, extra people, extra domes or duplicate staircases."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nCreate a highly detailed cinematic architectural illustration of a remote mountain observatory at blue hour after a light snowfall. The building occupies the right two-thirds of the square image: a circular pale-stone base supports one weathered copper dome, whose open viewing slit points toward the upper left. A narrow exterior staircase with dark iron handrails climbs from a small lower entrance to a balcony; every step has a thin undisturbed layer of snow. One adult astronomer in a dark red winter coat stands on the balcony with a small white notebook held closed in one gloved hand, looking toward the sky. Place a warm amber lamp immediately above the entrance, creating a pool of light on the stone without illuminating the whole landscape. In the lower left foreground, a crooked pine tree and two exposed granite rocks frame a winding path that leads toward the building. Behind the observatory, show distant mountain ridges fading from slate blue into violet mist. The sky contains scattered sharp stars and a faint natural Milky Way band, with no oversized planets or fantasy moons. Balance warm interior light against cold ambient snow light, use physically plausible perspective, careful separation of the staircase rails and the figure, and visible differences between copper, stone, glass, fabric and snow. The final scene should feel quiet, inhabited and believable. Do not add titles, labels, logos, watermarks, extra people, extra domes or duplicate staircases."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 51}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 226.020651, "comfy_server_seconds": 223.135, "stages_seconds": {"node_3_observer_receive_interval": 115.98399999999674, "node_10_observer_receive_interval": 0.01600000000325963, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 78.40600000001723, "node_7_observer_receive_interval": 28.639999999955762, "node_8_observer_receive_interval": 0.07900000002700835}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 10.771, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 111, "inflight_sample_count": 110, "minimum_ram_available_bytes": 4460896256, "minimum_commit_available_bytes": 11257696256, "minimum_system_drive_free_bytes": 18669465600, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent quiet blue-hour observatory scene with one large dome, one red-coated figure, one snow-covered staircase, foreground pine/rocks and distant layered mountains.", "Building placement, open slit toward the upper left, red coat, small light-colored notebook, amber entrance lamp, snow on steps and cold/warm lighting contrast are represented. The figure appears to attend to the notebook rather than clearly look at the sky. Requested stars and faint Milky Way are not visibly present. The exact two-rock count is not cleanly met or certifiable among overlapping foreground boulders.", "Dome reads as dark weathered metal, base as smooth pale masonry/stucco, and snow, coat and rails are distinct. Fine stone material detail is less explicit than the brief.", "Overall building, balcony and staircase perspective are plausible; thin rail details and figure-to-railing separation are imperfect but readable. No obvious extra person, dome or duplicate staircase.", "No title, label, logo or watermark is apparent.", "Stars and Milky Way are absent from the visibly plain sky.", "Figure does not clearly follow the requested skyward gaze.", "Exact rock-count and some fine architectural/material details are weaker than the broad composition."], "media": [{"path": "media/qwen-turbo-10.png", "mime_type": "image/png", "size_bytes": 402891, "sha256": "7bc1033039962507ee2aa153ebfacaa4cdd1f77e93e152bd7241ebae686b85bc", "source_sha256": "f051ec1ab055f23d3fd6513e3428d807734a846f5e51881403cf419ccb33c161", "width": 512, "height": 512, "pixel_sha256": "9672ee9777924c1b99ccff8952362c0119f99a87bb1e22c7bc82a24b579ccb6f", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-10.json", "workflow_sha256": "11fe5261e6abff03fd6a82be45d93b3827c421e3cc3e684f03b72df24694e0b0", "source_workflow_canonical_sha256": "7120b879556ee04ac859a6126f8d1ab24771a28bf208d58fff10b120bf50e68a"} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/runs/bd801c85-11e3-5033-b076-acd84846b820/run.md b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/runs/bd801c85-11e3-5033-b076-acd84846b820/run.md new file mode 100644 index 0000000..33fd5d4 --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/runs/bd801c85-11e3-5033-b076-acd84846b820/run.md @@ -0,0 +1,9 @@ +# Predator media: Qwen Image 2.1 Turbo Q4 · ten-image seed/cache trial + +10 retained outputs · 10 accepted results · 10 recorded attempts · median Factory trace 202.0s + +Technical delivery and media decoding do not establish visual or musical quality. Timings include adapter overhead; cache behavior differs between runs. Failed and cancelled attempts remain in the evidence. + +ComfyUI source: `63be2f9b82ad0dac88a7ceaa28a9cf3154125a18`. Per-run prompts, workflow hashes, model versions, timing boundaries, cache evidence, resource observations and verification are in this group’s run.jsonl. Media-generation records use the campaign schema, not the LLM token-throughput schema. Private raw operator logs are excluded; sanitized failed attempts remain. + +[Methodology](https://benchmarks.weeyuga.com/methodology.html#packet-foundry-media-v1) · [All media](https://benchmarks.weeyuga.com/benchmarks/predator-media-20261021.html) · [Findings](https://benchmarks.weeyuga.com/findings/images-video-music-on-6gb.html) diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/seed-variant-cache-findings.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/seed-variant-cache-findings.json new file mode 100644 index 0000000..8023367 --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/seed-variant-cache-findings.json @@ -0,0 +1,1345 @@ +{ + "schema_version": "seed-variant-cache-1", + "scope": "Read-only reanalysis of the existing FLUX 100 images and completed ten-image Qwen Turbo trial. No additional generation for this analysis.", + "flux50_seed_pairs": { + "model": "Abiray/flux-3-image-klein-sd-GGUF:Q4_K_M", + "pairs": 50, + "observed_seed_variants_per_brief": 2, + "five_seed_variants_tested": false, + "configuration": { + "width": 512, + "height": 512, + "steps": 4, + "cfg": 1, + "generator_quant": "Q4_K_M", + "compute": "FP32", + "cpu_vae": true, + "lowvram": true + }, + "cache_counts": { + "second_positive_encoder_cached": 50, + "second_negative_encoder_cached": 50, + "second_sampler_cached": 0, + "same_actual_runtime_pair": 50, + "identical_graph_except_seed": 50, + "second_probe_is_first_probe": 50, + "distinct_output_hashes_per_pair": 50, + "first_positive_encoder_cached": 0 + }, + "all_pairs": { + "first_native_seconds": { + "n": 50, + "min": 64.204, + "median": 71.671, + "mean": 79.30762, + "max": 241.398, + "total": 3965.3810000000003 + }, + "second_native_seconds": { + "n": 50, + "min": 36.576, + "median": 38.397999999999996, + "mean": 39.505, + "max": 44.558, + "total": 1975.2499999999998 + }, + "native_seconds_saved": { + "n": 50, + "min": 27.309, + "median": 33.367000000000004, + "mean": 39.80262, + "max": 204.344, + "total": 1990.1310000000003 + }, + "native_speedup": { + "n": 50, + "min": 1.7400193433990652, + "median": 1.8708729216784072, + "mean": 2.002159338297874, + "max": 6.514762238894586, + "total": 100.10796691489372 + }, + "first_trace_seconds": { + "n": 50, + "min": 66.467404, + "median": 74.03037850000001, + "mean": 81.160127, + "max": 242.338818, + "total": 4058.0063499999997 + }, + "second_trace_seconds": { + "n": 50, + "min": 39.273643, + "median": 39.397068000000004, + "mean": 41.17113458, + "max": 45.413099, + "total": 2058.556729 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 50, + "min": 10.422, + "median": 22.38, + "mean": 21.21556, + "max": 31.712, + "total": 1060.778 + }, + "first_positive_encoding_seconds": { + "n": 50, + "min": 27.51600000000326, + "median": 31.68800000002375, + "mean": 36.278460000006014, + "max": 56.32800000003772, + "total": 1813.9230000003008 + } + }, + "exclude_initial_load_pair": { + "first_native_seconds": { + "n": 49, + "min": 64.204, + "median": 70.046, + "mean": 75.99965306122449, + "max": 101.131, + "total": 3723.983 + }, + "second_native_seconds": { + "n": 49, + "min": 36.576, + "median": 38.491, + "mean": 39.555020408163266, + "max": 44.558, + "total": 1938.196 + }, + "native_seconds_saved": { + "n": 49, + "min": 27.309, + "median": 32.125, + "mean": 36.44463265306123, + "max": 56.761, + "total": 1785.7870000000003 + }, + "native_speedup": { + "n": 49, + "min": 1.7400193433990652, + "median": 1.8471559294322406, + "mean": 1.9100654015510024, + "max": 2.279265269326121, + "total": 93.59320467599913 + }, + "first_trace_seconds": { + "n": 49, + "min": 66.467404, + "median": 72.559234, + "mean": 77.87076595918367, + "max": 102.832423, + "total": 3815.6675319999995 + }, + "second_trace_seconds": { + "n": 49, + "min": 39.273643, + "median": 39.3951, + "mean": 41.20691226530612, + "max": 45.413099, + "total": 2019.138701 + }, + "first_positive_encoding_seconds": { + "n": 49, + "min": 27.51600000000326, + "median": 31.985000000044238, + "mean": 36.452836734700085, + "max": 56.32800000003772, + "total": 1786.189000000304 + } + }, + "by_tier": { + "1": { + "first_native_seconds": { + "n": 10, + "min": 64.57, + "median": 64.88550000000001, + "mean": 82.5835, + "max": 241.398, + "total": 825.835 + }, + "second_native_seconds": { + "n": 10, + "min": 36.576, + "median": 36.931, + "mean": 36.9348, + "max": 37.258, + "total": 369.348 + }, + "native_seconds_saved": { + "n": 10, + "min": 27.545, + "median": 28.106, + "mean": 45.6487, + "max": 204.344, + "total": 456.487 + }, + "first_trace_seconds": { + "n": 10, + "min": 66.467404, + "median": 66.59116850000001, + "mean": 84.1501359, + "max": 242.338818, + "total": 841.5013589999999 + }, + "second_trace_seconds": { + "n": 10, + "min": 39.273643, + "median": 39.350615, + "mean": 39.362302899999996, + "max": 39.476116, + "total": 393.623029 + }, + "first_positive_encoding_seconds": { + "n": 10, + "min": 27.73399999999674, + "median": 27.859499999991385, + "mean": 27.93749999999418, + "max": 28.485000000044238, + "total": 279.3749999999418 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 10, + "min": 10.422, + "median": 26.314999999999998, + "mean": 21.0136, + "max": 27.701, + "total": 210.136 + } + }, + "2": { + "first_native_seconds": { + "n": 10, + "min": 64.204, + "median": 64.979, + "mean": 66.5586, + "max": 70.046, + "total": 665.5860000000001 + }, + "second_native_seconds": { + "n": 10, + "min": 36.682, + "median": 37.0485, + "mean": 37.4005, + "max": 38.292, + "total": 374.005 + }, + "native_seconds_saved": { + "n": 10, + "min": 27.309, + "median": 28.152, + "mean": 29.1581, + "max": 32.125, + "total": 291.581 + }, + "first_trace_seconds": { + "n": 10, + "min": 66.470952, + "median": 66.60241149999999, + "mean": 68.0339363, + "max": 72.466618, + "total": 680.3393630000002 + }, + "second_trace_seconds": { + "n": 10, + "min": 39.288523, + "median": 39.369251000000006, + "mean": 39.3625179, + "max": 39.4423, + "total": 393.625179 + }, + "first_positive_encoding_seconds": { + "n": 10, + "min": 27.51600000000326, + "median": 28.094000000011874, + "mean": 29.17840000001015, + "max": 31.985000000044238, + "total": 291.7840000001015 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 10, + "min": 11.778, + "median": 24.357, + "mean": 23.0208, + "max": 31.712, + "total": 230.20799999999997 + } + }, + "3": { + "first_native_seconds": { + "n": 10, + "min": 69.077, + "median": 69.54599999999999, + "mean": 70.9772, + "max": 73.464, + "total": 709.7719999999999 + }, + "second_native_seconds": { + "n": 10, + "min": 38.008, + "median": 38.397999999999996, + "mean": 38.4107, + "max": 38.705, + "total": 384.107 + }, + "native_seconds_saved": { + "n": 10, + "min": 30.839, + "median": 31.393, + "mean": 32.5665, + "max": 34.852, + "total": 325.66499999999996 + }, + "first_trace_seconds": { + "n": 10, + "min": 69.48063, + "median": 72.549484, + "mean": 72.8756681, + "max": 75.63889, + "total": 728.7566810000001 + }, + "second_trace_seconds": { + "n": 10, + "min": 39.287177, + "median": 39.335657499999996, + "mean": 39.3440559, + "max": 39.400407, + "total": 393.440559 + }, + "first_positive_encoding_seconds": { + "n": 10, + "min": 30.90600000001723, + "median": 31.38300000000163, + "mean": 32.56720000000787, + "max": 34.78099999995902, + "total": 325.6720000000787 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 10, + "min": 18.393, + "median": 21.78, + "mean": 21.6223, + "max": 26.858, + "total": 216.223 + } + }, + "4": { + "first_native_seconds": { + "n": 10, + "min": 77.877, + "median": 78.6135, + "mean": 78.4776, + "max": 78.736, + "total": 784.7760000000001 + }, + "second_native_seconds": { + "n": 10, + "min": 40.423, + "median": 40.4815, + "mean": 40.5054, + "max": 40.68, + "total": 405.05400000000003 + }, + "native_seconds_saved": { + "n": 10, + "min": 37.423, + "median": 37.981, + "mean": 37.9722, + "max": 38.269, + "total": 379.72200000000004 + }, + "first_trace_seconds": { + "n": 10, + "min": 78.600529, + "median": 81.565219, + "mean": 80.7096113, + "max": 81.696329, + "total": 807.0961130000001 + }, + "second_trace_seconds": { + "n": 10, + "min": 42.349654, + "median": 42.403067, + "mean": 42.4041422, + "max": 42.532345, + "total": 424.041422 + }, + "first_positive_encoding_seconds": { + "n": 10, + "min": 37.65600000001723, + "median": 38.10950000002049, + "mean": 38.05140000000247, + "max": 38.32799999997951, + "total": 380.5140000000247 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 10, + "min": 13.12, + "median": 14.9955, + "mean": 15.7422, + "max": 21.868, + "total": 157.42200000000003 + } + }, + "5": { + "first_native_seconds": { + "n": 10, + "min": 96.483, + "median": 97.2595, + "mean": 97.94120000000001, + "max": 101.131, + "total": 979.4119999999999 + }, + "second_native_seconds": { + "n": 10, + "min": 44.121, + "median": 44.253, + "mean": 44.2736, + "max": 44.558, + "total": 442.73600000000005 + }, + "native_seconds_saved": { + "n": 10, + "min": 52.362, + "median": 53.0135, + "mean": 53.6676, + "max": 56.761, + "total": 536.676 + }, + "first_trace_seconds": { + "n": 10, + "min": 96.80878, + "median": 99.68121, + "mean": 100.0312834, + "max": 102.832423, + "total": 1000.3128339999998 + }, + "second_trace_seconds": { + "n": 10, + "min": 45.341639, + "median": 45.3771555, + "mean": 45.382654, + "max": 45.413099, + "total": 453.82654 + }, + "first_positive_encoding_seconds": { + "n": 10, + "min": 52.40600000001723, + "median": 53.0625, + "mean": 53.65780000001541, + "max": 56.32800000003772, + "total": 536.5780000001541 + }, + "second_native_start_after_first_native_finish_seconds": { + "n": 10, + "min": 12.666, + "median": 27.817, + "mean": 24.6789, + "max": 29.406, + "total": 246.78900000000004 + } + } + }, + "pair_evidence": [ + { + "first_record": "pf-img100-01-t1-s42", + "second_record": "pf-img100-01-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 241.398, + "second_native_seconds": 37.054, + "native_inter_run_gap_seconds": 14.139, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-02-t1-s42", + "second_record": "pf-img100-02-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 65.35, + "second_native_seconds": 37.258, + "native_inter_run_gap_seconds": 11.281, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-03-t1-s42", + "second_record": "pf-img100-03-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.767, + "second_native_seconds": 37.222, + "native_inter_run_gap_seconds": 26.604, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-04-t1-s42", + "second_record": "pf-img100-04-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.679, + "second_native_seconds": 37.016, + "native_inter_run_gap_seconds": 26.543, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-05-t1-s42", + "second_record": "pf-img100-05-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.57, + "second_native_seconds": 36.908, + "native_inter_run_gap_seconds": 10.422, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-06-t1-s42", + "second_record": "pf-img100-06-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.943, + "second_native_seconds": 36.896, + "native_inter_run_gap_seconds": 27.239, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-08-t1-s42", + "second_record": "pf-img100-08-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.791, + "second_native_seconds": 36.576, + "native_inter_run_gap_seconds": 13.335, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-09-t1-s42", + "second_record": "pf-img100-09-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.828, + "second_native_seconds": 36.708, + "native_inter_run_gap_seconds": 27.701, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-16-t1-s42", + "second_record": "pf-img100-16-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 65.354, + "second_native_seconds": 36.954, + "native_inter_run_gap_seconds": 26.087, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-20-t1-s42", + "second_record": "pf-img100-20-t1-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 65.155, + "second_native_seconds": 36.756, + "native_inter_run_gap_seconds": 26.785, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-01-t2-s42", + "second_record": "pf-img100-01-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.204, + "second_native_seconds": 36.895, + "native_inter_run_gap_seconds": 11.778, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-02-t2-s42", + "second_record": "pf-img100-02-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.965, + "second_native_seconds": 36.897, + "native_inter_run_gap_seconds": 31.712, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-03-t2-s42", + "second_record": "pf-img100-03-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.918, + "second_native_seconds": 36.682, + "native_inter_run_gap_seconds": 26.206, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-04-t2-s42", + "second_record": "pf-img100-04-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.993, + "second_native_seconds": 37.056, + "native_inter_run_gap_seconds": 27.134, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-05-t2-s42", + "second_record": "pf-img100-05-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.762, + "second_native_seconds": 37.041, + "native_inter_run_gap_seconds": 11.901, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-06-t2-s42", + "second_record": "pf-img100-06-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 70.046, + "second_native_seconds": 37.921, + "native_inter_run_gap_seconds": 22.36, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-08-t2-s42", + "second_record": "pf-img100-08-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.192, + "second_native_seconds": 38.292, + "native_inter_run_gap_seconds": 22.4, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-09-t2-s42", + "second_record": "pf-img100-09-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 64.379, + "second_native_seconds": 36.876, + "native_inter_run_gap_seconds": 28.003, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-16-t2-s42", + "second_record": "pf-img100-16-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 68.971, + "second_native_seconds": 38.169, + "native_inter_run_gap_seconds": 24.357, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-20-t2-s42", + "second_record": "pf-img100-20-t2-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.156, + "second_native_seconds": 38.176, + "native_inter_run_gap_seconds": 24.357, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-01-t3-s42", + "second_record": "pf-img100-01-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.601, + "second_native_seconds": 38.305, + "native_inter_run_gap_seconds": 23.193, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-02-t3-s42", + "second_record": "pf-img100-02-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.491, + "second_native_seconds": 38.008, + "native_inter_run_gap_seconds": 22.005, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-03-t3-s42", + "second_record": "pf-img100-03-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.451, + "second_native_seconds": 38.148, + "native_inter_run_gap_seconds": 22.876, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-04-t3-s42", + "second_record": "pf-img100-04-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.125, + "second_native_seconds": 38.262, + "native_inter_run_gap_seconds": 26.858, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-05-t3-s42", + "second_record": "pf-img100-05-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.388, + "second_native_seconds": 38.491, + "native_inter_run_gap_seconds": 24.166, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-06-t3-s42", + "second_record": "pf-img100-06-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 73.296, + "second_native_seconds": 38.687, + "native_inter_run_gap_seconds": 18.854, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-08-t3-s42", + "second_record": "pf-img100-08-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 73.437, + "second_native_seconds": 38.705, + "native_inter_run_gap_seconds": 18.393, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-09-t3-s42", + "second_record": "pf-img100-09-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 69.077, + "second_native_seconds": 38.238, + "native_inter_run_gap_seconds": 21.555, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-16-t3-s42", + "second_record": "pf-img100-16-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 73.442, + "second_native_seconds": 38.651, + "native_inter_run_gap_seconds": 19.37, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-20-t3-s42", + "second_record": "pf-img100-20-t3-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 73.464, + "second_native_seconds": 38.612, + "native_inter_run_gap_seconds": 18.953, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-01-t4-s42", + "second_record": "pf-img100-01-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.215, + "second_native_seconds": 40.46, + "native_inter_run_gap_seconds": 21.868, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-02-t4-s42", + "second_record": "pf-img100-02-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.617, + "second_native_seconds": 40.423, + "native_inter_run_gap_seconds": 13.502, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-03-t4-s42", + "second_record": "pf-img100-03-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.656, + "second_native_seconds": 40.474, + "native_inter_run_gap_seconds": 14.668, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-04-t4-s42", + "second_record": "pf-img100-04-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.617, + "second_native_seconds": 40.612, + "native_inter_run_gap_seconds": 15.534, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-05-t4-s42", + "second_record": "pf-img100-05-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.467, + "second_native_seconds": 40.51, + "native_inter_run_gap_seconds": 15.323, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-06-t4-s42", + "second_record": "pf-img100-06-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 77.877, + "second_native_seconds": 40.454, + "native_inter_run_gap_seconds": 18.954, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-08-t4-s42", + "second_record": "pf-img100-08-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.727, + "second_native_seconds": 40.458, + "native_inter_run_gap_seconds": 15.354, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-09-t4-s42", + "second_record": "pf-img100-09-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.254, + "second_native_seconds": 40.489, + "native_inter_run_gap_seconds": 14.643, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-16-t4-s42", + "second_record": "pf-img100-16-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.61, + "second_native_seconds": 40.68, + "native_inter_run_gap_seconds": 13.12, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-20-t4-s42", + "second_record": "pf-img100-20-t4-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 78.736, + "second_native_seconds": 40.494, + "native_inter_run_gap_seconds": 14.456, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-01-t5-s42", + "second_record": "pf-img100-01-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.733, + "second_native_seconds": 44.245, + "native_inter_run_gap_seconds": 27.851, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-02-t5-s42", + "second_record": "pf-img100-02-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.058, + "second_native_seconds": 44.319, + "native_inter_run_gap_seconds": 13.702, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-03-t5-s42", + "second_record": "pf-img100-03-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.156, + "second_native_seconds": 44.279, + "native_inter_run_gap_seconds": 28.917, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-04-t5-s42", + "second_record": "pf-img100-04-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 96.93, + "second_native_seconds": 44.162, + "native_inter_run_gap_seconds": 28.792, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-05-t5-s42", + "second_record": "pf-img100-05-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.26, + "second_native_seconds": 44.231, + "native_inter_run_gap_seconds": 28.36, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-06-t5-s42", + "second_record": "pf-img100-06-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 101.049, + "second_native_seconds": 44.558, + "native_inter_run_gap_seconds": 24.123, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-08-t5-s42", + "second_record": "pf-img100-08-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.259, + "second_native_seconds": 44.261, + "native_inter_run_gap_seconds": 29.406, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-09-t5-s42", + "second_record": "pf-img100-09-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 97.353, + "second_native_seconds": 44.19, + "native_inter_run_gap_seconds": 27.783, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-16-t5-s42", + "second_record": "pf-img100-16-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 96.483, + "second_native_seconds": 44.121, + "native_inter_run_gap_seconds": 12.666, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + }, + { + "first_record": "pf-img100-20-t5-s42", + "second_record": "pf-img100-20-t5-s43", + "seeds": [ + 42, + 43 + ], + "first_native_seconds": 101.131, + "second_native_seconds": 44.37, + "native_inter_run_gap_seconds": 25.189, + "same_runtime_retained": true, + "same_graph_except_sampler_seed": true, + "second_positive_encoder_cached": true, + "second_negative_encoder_cached": true, + "sampling_cached_second": false, + "distinct_retained_outputs": true + } + ], + "interpretation": "All 50 second requests reused positive and negative conditioning. Sampling was not cached, and every pair produced distinct outputs. The initial-load pair is excluded from the main steady-runtime comparison. No five-variant group was tested.", + "limits": [ + "Same42/43 seed ordering; no randomized cache-disabled control.", + "Node intervals include preparation/event delivery, not isolated kernels.", + "Observed cache survives the measured gaps but is local to one retained Comfy runtime; no persistence guarantee across model switches, restarts, factory changes or eviction." + ] + }, + "qwen_turbo_ten": { + "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", + "generated_outputs": 10, + "first_native_start_utc": "2026-10-21T08:41:18.067000+00:00", + "distinct_briefs": 9, + "measured_seed_variants_in_first_brief": 2, + "five_seed_variants_tested": false, + "initial_native_seconds": 320.246, + "same_brief_variant_native_seconds": 95.234, + "native_inter_run_gap_seconds": 34.322, + "same_runtime_retained": true, + "only_graph_difference": [ + "/9/inputs/noise_seed" + ], + "second_encoder_cached": true, + "sampling_and_decoding_repeated": true, + "new_briefs_in_retained_runtime": { + "requests": 8, + "native_seconds": [ + 194.865, + 196.417, + 199.64, + 199.412, + 199.784, + 210.023, + 208.566, + 223.135 + ], + "minimum_seconds": 194.865, + "median_seconds": 199.712, + "maximum_seconds": 223.135, + "sum_seconds": 1631.842 + }, + "limits": [ + "The initial request includes loading plus text encoding, so the first/second delta is a combined startup/model/encoding benefit.", + "Only two same-brief seed outputs were measured; do not present five-variant performance as measured.", + "Fresh-prompt latency is a separate workload from cached same-brief variation." + ] + }, + "production_guidance": "Keep the complete composed brief/materials/settings identical and vary only the seed in consecutive requests to the same factory/profile/Comfy runtime. Cache state is local to that runtime and can be lost after restarts, model switches or eviction.", + "five_variant_group_measured": false +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-01.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-01.json new file mode 100644 index 0000000..461ed02 --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-01.json @@ -0,0 +1,132 @@ +{ + "1": { + "class_type": "UnetLoaderGGUFAdvanced", + "inputs": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + } + }, + "2": { + "class_type": "CLIPLoader", + "inputs": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + } + }, + "3": { + "class_type": "TextEncodeQwenImage21", + "inputs": { + "clip": [ + "2", + 0 + ], + "prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph.", + "negative_prompt": "", + "resolution": 512 + } + }, + "4": { + "class_type": "VAELoader", + "inputs": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + } + }, + "5": { + "class_type": "QwenImage21Cache", + "inputs": { + "model": [ + "1", + 0 + ], + "device": "off", + "dtype": "default" + } + }, + "6": { + "class_type": "SamplerCustomAdvanced", + "inputs": { + "noise": [ + "9", + 0 + ], + "guider": [ + "10", + 0 + ], + "sampler": [ + "11", + 0 + ], + "sigmas": [ + "12", + 0 + ], + "latent_image": [ + "3", + 2 + ] + } + }, + "7": { + "class_type": "VAEDecode", + "inputs": { + "samples": [ + "6", + 0 + ], + "vae": [ + "4", + 0 + ] + } + }, + "8": { + "class_type": "SaveImage", + "inputs": { + "images": [ + "7", + 0 + ], + "filename_prefix": "PacketFoundry-Qwen21Turbo" + } + }, + "9": { + "class_type": "RandomNoise", + "inputs": { + "noise_seed": 42 + } + }, + "10": { + "class_type": "CFGGuider", + "inputs": { + "model": [ + "5", + 0 + ], + "positive": [ + "3", + 0 + ], + "negative": [ + "3", + 1 + ], + "cfg": 1.0 + } + }, + "11": { + "class_type": "KSamplerSelect", + "inputs": { + "sampler_name": "euler" + } + }, + "12": { + "class_type": "ManualSigmas", + "inputs": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + } +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-02.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-02.json new file mode 100644 index 0000000..5b6f2df --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-02.json @@ -0,0 +1,132 @@ +{ + "1": { + "class_type": "UnetLoaderGGUFAdvanced", + "inputs": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + } + }, + "2": { + "class_type": "CLIPLoader", + "inputs": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + } + }, + "3": { + "class_type": "TextEncodeQwenImage21", + "inputs": { + "clip": [ + "2", + 0 + ], + "prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph.", + "negative_prompt": "", + "resolution": 512 + } + }, + "4": { + "class_type": "VAELoader", + "inputs": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + } + }, + "5": { + "class_type": "QwenImage21Cache", + "inputs": { + "model": [ + "1", + 0 + ], + "device": "off", + "dtype": "default" + } + }, + "6": { + "class_type": "SamplerCustomAdvanced", + "inputs": { + "noise": [ + "9", + 0 + ], + "guider": [ + "10", + 0 + ], + "sampler": [ + "11", + 0 + ], + "sigmas": [ + "12", + 0 + ], + "latent_image": [ + "3", + 2 + ] + } + }, + "7": { + "class_type": "VAEDecode", + "inputs": { + "samples": [ + "6", + 0 + ], + "vae": [ + "4", + 0 + ] + } + }, + "8": { + "class_type": "SaveImage", + "inputs": { + "images": [ + "7", + 0 + ], + "filename_prefix": "PacketFoundry-Qwen21Turbo" + } + }, + "9": { + "class_type": "RandomNoise", + "inputs": { + "noise_seed": 43 + } + }, + "10": { + "class_type": "CFGGuider", + "inputs": { + "model": [ + "5", + 0 + ], + "positive": [ + "3", + 0 + ], + "negative": [ + "3", + 1 + ], + "cfg": 1.0 + } + }, + "11": { + "class_type": "KSamplerSelect", + "inputs": { + "sampler_name": "euler" + } + }, + "12": { + "class_type": "ManualSigmas", + "inputs": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + } +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-03.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-03.json new file mode 100644 index 0000000..ba21a20 --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-03.json @@ -0,0 +1,132 @@ +{ + "1": { + "class_type": "UnetLoaderGGUFAdvanced", + "inputs": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + } + }, + "2": { + "class_type": "CLIPLoader", + "inputs": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + } + }, + "3": { + "class_type": "TextEncodeQwenImage21", + "inputs": { + "clip": [ + "2", + 0 + ], + "prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA turquoise alpine lake beneath snowy mountains at sunrise, a yellow canoe near the shore, crisp natural landscape photograph.", + "negative_prompt": "", + "resolution": 512 + } + }, + "4": { + "class_type": "VAELoader", + "inputs": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + } + }, + "5": { + "class_type": "QwenImage21Cache", + "inputs": { + "model": [ + "1", + 0 + ], + "device": "off", + "dtype": "default" + } + }, + "6": { + "class_type": "SamplerCustomAdvanced", + "inputs": { + "noise": [ + "9", + 0 + ], + "guider": [ + "10", + 0 + ], + "sampler": [ + "11", + 0 + ], + "sigmas": [ + "12", + 0 + ], + "latent_image": [ + "3", + 2 + ] + } + }, + "7": { + "class_type": "VAEDecode", + "inputs": { + "samples": [ + "6", + 0 + ], + "vae": [ + "4", + 0 + ] + } + }, + "8": { + "class_type": "SaveImage", + "inputs": { + "images": [ + "7", + 0 + ], + "filename_prefix": "PacketFoundry-Qwen21Turbo" + } + }, + "9": { + "class_type": "RandomNoise", + "inputs": { + "noise_seed": 44 + } + }, + "10": { + "class_type": "CFGGuider", + "inputs": { + "model": [ + "5", + 0 + ], + "positive": [ + "3", + 0 + ], + "negative": [ + "3", + 1 + ], + "cfg": 1.0 + } + }, + "11": { + "class_type": "KSamplerSelect", + "inputs": { + "sampler_name": "euler" + } + }, + "12": { + "class_type": "ManualSigmas", + "inputs": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + } +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-04.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-04.json new file mode 100644 index 0000000..d67ef4e --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-04.json @@ -0,0 +1,132 @@ +{ + "1": { + "class_type": "UnetLoaderGGUFAdvanced", + "inputs": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + } + }, + "2": { + "class_type": "CLIPLoader", + "inputs": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + } + }, + "3": { + "class_type": "TextEncodeQwenImage21", + "inputs": { + "clip": [ + "2", + 0 + ], + "prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nProduct photograph of a matte black ceramic teapot with a warm copper handle, resting on dark slate beside two pale linen napkins. Three-quarter view, softbox lighting, crisp material detail, uncluttered background.", + "negative_prompt": "", + "resolution": 512 + } + }, + "4": { + "class_type": "VAELoader", + "inputs": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + } + }, + "5": { + "class_type": "QwenImage21Cache", + "inputs": { + "model": [ + "1", + 0 + ], + "device": "off", + "dtype": "default" + } + }, + "6": { + "class_type": "SamplerCustomAdvanced", + "inputs": { + "noise": [ + "9", + 0 + ], + "guider": [ + "10", + 0 + ], + "sampler": [ + "11", + 0 + ], + "sigmas": [ + "12", + 0 + ], + "latent_image": [ + "3", + 2 + ] + } + }, + "7": { + "class_type": "VAEDecode", + "inputs": { + "samples": [ + "6", + 0 + ], + "vae": [ + "4", + 0 + ] + } + }, + "8": { + "class_type": "SaveImage", + "inputs": { + "images": [ + "7", + 0 + ], + "filename_prefix": "PacketFoundry-Qwen21Turbo" + } + }, + "9": { + "class_type": "RandomNoise", + "inputs": { + "noise_seed": 45 + } + }, + "10": { + "class_type": "CFGGuider", + "inputs": { + "model": [ + "5", + 0 + ], + "positive": [ + "3", + 0 + ], + "negative": [ + "3", + 1 + ], + "cfg": 1.0 + } + }, + "11": { + "class_type": "KSamplerSelect", + "inputs": { + "sampler_name": "euler" + } + }, + "12": { + "class_type": "ManualSigmas", + "inputs": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + } +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-05.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-05.json new file mode 100644 index 0000000..eb3c396 --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-05.json @@ -0,0 +1,132 @@ +{ + "1": { + "class_type": "UnetLoaderGGUFAdvanced", + "inputs": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + } + }, + "2": { + "class_type": "CLIPLoader", + "inputs": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + } + }, + "3": { + "class_type": "TextEncodeQwenImage21", + "inputs": { + "clip": [ + "2", + 0 + ], + "prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA cozy neighborhood bakery storefront on a rainy evening. The main sign clearly reads \"MORNING BREAD\" in cream block letters. Warm light glows through the windows, a bicycle leans beside the door, wet paving reflects the orange interior, realistic street photograph.", + "negative_prompt": "", + "resolution": 512 + } + }, + "4": { + "class_type": "VAELoader", + "inputs": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + } + }, + "5": { + "class_type": "QwenImage21Cache", + "inputs": { + "model": [ + "1", + 0 + ], + "device": "off", + "dtype": "default" + } + }, + "6": { + "class_type": "SamplerCustomAdvanced", + "inputs": { + "noise": [ + "9", + 0 + ], + "guider": [ + "10", + 0 + ], + "sampler": [ + "11", + 0 + ], + "sigmas": [ + "12", + 0 + ], + "latent_image": [ + "3", + 2 + ] + } + }, + "7": { + "class_type": "VAEDecode", + "inputs": { + "samples": [ + "6", + 0 + ], + "vae": [ + "4", + 0 + ] + } + }, + "8": { + "class_type": "SaveImage", + "inputs": { + "images": [ + "7", + 0 + ], + "filename_prefix": "PacketFoundry-Qwen21Turbo" + } + }, + "9": { + "class_type": "RandomNoise", + "inputs": { + "noise_seed": 46 + } + }, + "10": { + "class_type": "CFGGuider", + "inputs": { + "model": [ + "5", + 0 + ], + "positive": [ + "3", + 0 + ], + "negative": [ + "3", + 1 + ], + "cfg": 1.0 + } + }, + "11": { + "class_type": "KSamplerSelect", + "inputs": { + "sampler_name": "euler" + } + }, + "12": { + "class_type": "ManualSigmas", + "inputs": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + } +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-06.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-06.json new file mode 100644 index 0000000..7db9200 --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-06.json @@ -0,0 +1,132 @@ +{ + "1": { + "class_type": "UnetLoaderGGUFAdvanced", + "inputs": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + } + }, + "2": { + "class_type": "CLIPLoader", + "inputs": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + } + }, + "3": { + "class_type": "TextEncodeQwenImage21", + "inputs": { + "clip": [ + "2", + 0 + ], + "prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA bright Victorian glass greenhouse filled with lush ferns, hanging orchids and terracotta pots. A narrow brick path curves from the bottom left toward an open green door at the far end. Morning sunlight makes geometric shadows through the iron window frames. Include a single brass watering can beside the path and misty trees beyond the glass. Architectural photograph, balanced natural colors, no people or text.", + "negative_prompt": "", + "resolution": 512 + } + }, + "4": { + "class_type": "VAELoader", + "inputs": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + } + }, + "5": { + "class_type": "QwenImage21Cache", + "inputs": { + "model": [ + "1", + 0 + ], + "device": "off", + "dtype": "default" + } + }, + "6": { + "class_type": "SamplerCustomAdvanced", + "inputs": { + "noise": [ + "9", + 0 + ], + "guider": [ + "10", + 0 + ], + "sampler": [ + "11", + 0 + ], + "sigmas": [ + "12", + 0 + ], + "latent_image": [ + "3", + 2 + ] + } + }, + "7": { + "class_type": "VAEDecode", + "inputs": { + "samples": [ + "6", + 0 + ], + "vae": [ + "4", + 0 + ] + } + }, + "8": { + "class_type": "SaveImage", + "inputs": { + "images": [ + "7", + 0 + ], + "filename_prefix": "PacketFoundry-Qwen21Turbo" + } + }, + "9": { + "class_type": "RandomNoise", + "inputs": { + "noise_seed": 47 + } + }, + "10": { + "class_type": "CFGGuider", + "inputs": { + "model": [ + "5", + 0 + ], + "positive": [ + "3", + 0 + ], + "negative": [ + "3", + 1 + ], + "cfg": 1.0 + } + }, + "11": { + "class_type": "KSamplerSelect", + "inputs": { + "sampler_name": "euler" + } + }, + "12": { + "class_type": "ManualSigmas", + "inputs": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + } +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-07.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-07.json new file mode 100644 index 0000000..df6d6c9 --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-07.json @@ -0,0 +1,132 @@ +{ + "1": { + "class_type": "UnetLoaderGGUFAdvanced", + "inputs": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + } + }, + "2": { + "class_type": "CLIPLoader", + "inputs": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + } + }, + "3": { + "class_type": "TextEncodeQwenImage21", + "inputs": { + "clip": [ + "2", + 0 + ], + "prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn editorial portrait of an adult female ceramic artist in her forties working in a sunlit pottery studio. She wears a blue linen apron over a cream shirt and calmly holds a small unfinished bowl in both hands, fingers naturally curled around the rim. Her face is clear and expressive, with a few silver strands in dark curly hair. Shelves of glazed pottery remain softly out of focus behind her. Frame from the waist upward, eye-level camera, gentle window light from the left, documentary realism.", + "negative_prompt": "", + "resolution": 512 + } + }, + "4": { + "class_type": "VAELoader", + "inputs": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + } + }, + "5": { + "class_type": "QwenImage21Cache", + "inputs": { + "model": [ + "1", + 0 + ], + "device": "off", + "dtype": "default" + } + }, + "6": { + "class_type": "SamplerCustomAdvanced", + "inputs": { + "noise": [ + "9", + 0 + ], + "guider": [ + "10", + 0 + ], + "sampler": [ + "11", + 0 + ], + "sigmas": [ + "12", + 0 + ], + "latent_image": [ + "3", + 2 + ] + } + }, + "7": { + "class_type": "VAEDecode", + "inputs": { + "samples": [ + "6", + 0 + ], + "vae": [ + "4", + 0 + ] + } + }, + "8": { + "class_type": "SaveImage", + "inputs": { + "images": [ + "7", + 0 + ], + "filename_prefix": "PacketFoundry-Qwen21Turbo" + } + }, + "9": { + "class_type": "RandomNoise", + "inputs": { + "noise_seed": 48 + } + }, + "10": { + "class_type": "CFGGuider", + "inputs": { + "model": [ + "5", + 0 + ], + "positive": [ + "3", + 0 + ], + "negative": [ + "3", + 1 + ], + "cfg": 1.0 + } + }, + "11": { + "class_type": "KSamplerSelect", + "inputs": { + "sampler_name": "euler" + } + }, + "12": { + "class_type": "ManualSigmas", + "inputs": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + } +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-08.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-08.json new file mode 100644 index 0000000..83d9196 --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-08.json @@ -0,0 +1,132 @@ +{ + "1": { + "class_type": "UnetLoaderGGUFAdvanced", + "inputs": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + } + }, + "2": { + "class_type": "CLIPLoader", + "inputs": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + } + }, + "3": { + "class_type": "TextEncodeQwenImage21", + "inputs": { + "clip": [ + "2", + 0 + ], + "prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA charming small brass robot with a round glass head stands on a wooden step stool inside an enormous old library, carefully returning a red book to a high shelf. The robot has two articulated arms and three visible fingers on each hand. A tiny blue mechanical bird perches on its shoulder. Bookshelves curve toward a domed ceiling painted with constellations; the far aisle is illuminated by warm amber lamps. Render as a detailed hand-painted storybook illustration with teal shadows, golden highlights and a clear focal point. Keep the robot, stool and book visibly separate, with no extra limbs, no captions and no written lettering.", + "negative_prompt": "", + "resolution": 512 + } + }, + "4": { + "class_type": "VAELoader", + "inputs": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + } + }, + "5": { + "class_type": "QwenImage21Cache", + "inputs": { + "model": [ + "1", + 0 + ], + "device": "off", + "dtype": "default" + } + }, + "6": { + "class_type": "SamplerCustomAdvanced", + "inputs": { + "noise": [ + "9", + 0 + ], + "guider": [ + "10", + 0 + ], + "sampler": [ + "11", + 0 + ], + "sigmas": [ + "12", + 0 + ], + "latent_image": [ + "3", + 2 + ] + } + }, + "7": { + "class_type": "VAEDecode", + "inputs": { + "samples": [ + "6", + 0 + ], + "vae": [ + "4", + 0 + ] + } + }, + "8": { + "class_type": "SaveImage", + "inputs": { + "images": [ + "7", + 0 + ], + "filename_prefix": "PacketFoundry-Qwen21Turbo" + } + }, + "9": { + "class_type": "RandomNoise", + "inputs": { + "noise_seed": 49 + } + }, + "10": { + "class_type": "CFGGuider", + "inputs": { + "model": [ + "5", + 0 + ], + "positive": [ + "3", + 0 + ], + "negative": [ + "3", + 1 + ], + "cfg": 1.0 + } + }, + "11": { + "class_type": "KSamplerSelect", + "inputs": { + "sampler_name": "euler" + } + }, + "12": { + "class_type": "ManualSigmas", + "inputs": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + } +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-09.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-09.json new file mode 100644 index 0000000..11d2232 --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-09.json @@ -0,0 +1,132 @@ +{ + "1": { + "class_type": "UnetLoaderGGUFAdvanced", + "inputs": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + } + }, + "2": { + "class_type": "CLIPLoader", + "inputs": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + } + }, + "3": { + "class_type": "TextEncodeQwenImage21", + "inputs": { + "clip": [ + "2", + 0 + ], + "prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA detailed travel photograph of a lively Mediterranean open-air produce market in early morning. In the foreground, three wooden crates hold red tomatoes, striped eggplants and yellow lemons, each type kept in its own crate. An adult vendor wearing a plain white shirt stands behind the stall, handing a small paper bag to an adult customer in a green jacket. A striped blue-and-white canopy casts cool shade over the stall. Beyond them, a narrow pedestrian street slopes toward a distant harbor with two white sailboats. Warm sunlight touches the upper walls but the foreground remains softly shaded. Compose the scene in layers with natural human proportions, believable hands, clear separation between people, baskets and produce, and rich but realistic color. Avoid artificial lens flares, watermarks and unreadable signs.", + "negative_prompt": "", + "resolution": 512 + } + }, + "4": { + "class_type": "VAELoader", + "inputs": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + } + }, + "5": { + "class_type": "QwenImage21Cache", + "inputs": { + "model": [ + "1", + 0 + ], + "device": "off", + "dtype": "default" + } + }, + "6": { + "class_type": "SamplerCustomAdvanced", + "inputs": { + "noise": [ + "9", + 0 + ], + "guider": [ + "10", + 0 + ], + "sampler": [ + "11", + 0 + ], + "sigmas": [ + "12", + 0 + ], + "latent_image": [ + "3", + 2 + ] + } + }, + "7": { + "class_type": "VAEDecode", + "inputs": { + "samples": [ + "6", + 0 + ], + "vae": [ + "4", + 0 + ] + } + }, + "8": { + "class_type": "SaveImage", + "inputs": { + "images": [ + "7", + 0 + ], + "filename_prefix": "PacketFoundry-Qwen21Turbo" + } + }, + "9": { + "class_type": "RandomNoise", + "inputs": { + "noise_seed": 50 + } + }, + "10": { + "class_type": "CFGGuider", + "inputs": { + "model": [ + "5", + 0 + ], + "positive": [ + "3", + 0 + ], + "negative": [ + "3", + 1 + ], + "cfg": 1.0 + } + }, + "11": { + "class_type": "KSamplerSelect", + "inputs": { + "sampler_name": "euler" + } + }, + "12": { + "class_type": "ManualSigmas", + "inputs": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + } +} diff --git a/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-10.json b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-10.json new file mode 100644 index 0000000..b1711c5 --- /dev/null +++ b/runs/3cc53336-2555-5e90-99c3-7b94dcb40acc/workflows/qwen-turbo-10.json @@ -0,0 +1,132 @@ +{ + "1": { + "class_type": "UnetLoaderGGUFAdvanced", + "inputs": { + "unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", + "dequant_dtype": "float32", + "patch_dtype": "float32", + "patch_on_device": false + } + }, + "2": { + "class_type": "CLIPLoader", + "inputs": { + "clip_name": "qwen3vl_8b_int8_convrot.safetensors", + "type": "qwen_image", + "device": "cpu" + } + }, + "3": { + "class_type": "TextEncodeQwenImage21", + "inputs": { + "clip": [ + "2", + 0 + ], + "prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nCreate a highly detailed cinematic architectural illustration of a remote mountain observatory at blue hour after a light snowfall. The building occupies the right two-thirds of the square image: a circular pale-stone base supports one weathered copper dome, whose open viewing slit points toward the upper left. A narrow exterior staircase with dark iron handrails climbs from a small lower entrance to a balcony; every step has a thin undisturbed layer of snow. One adult astronomer in a dark red winter coat stands on the balcony with a small white notebook held closed in one gloved hand, looking toward the sky. Place a warm amber lamp immediately above the entrance, creating a pool of light on the stone without illuminating the whole landscape. In the lower left foreground, a crooked pine tree and two exposed granite rocks frame a winding path that leads toward the building. Behind the observatory, show distant mountain ridges fading from slate blue into violet mist. The sky contains scattered sharp stars and a faint natural Milky Way band, with no oversized planets or fantasy moons. Balance warm interior light against cold ambient snow light, use physically plausible perspective, careful separation of the staircase rails and the figure, and visible differences between copper, stone, glass, fabric and snow. The final scene should feel quiet, inhabited and believable. Do not add titles, labels, logos, watermarks, extra people, extra domes or duplicate staircases.", + "negative_prompt": "", + "resolution": 512 + } + }, + "4": { + "class_type": "VAELoader", + "inputs": { + "vae_name": "qwen_image_2.1_vae_bf16.safetensors" + } + }, + "5": { + "class_type": "QwenImage21Cache", + "inputs": { + "model": [ + "1", + 0 + ], + "device": "off", + "dtype": "default" + } + }, + "6": { + "class_type": "SamplerCustomAdvanced", + "inputs": { + "noise": [ + "9", + 0 + ], + "guider": [ + "10", + 0 + ], + "sampler": [ + "11", + 0 + ], + "sigmas": [ + "12", + 0 + ], + "latent_image": [ + "3", + 2 + ] + } + }, + "7": { + "class_type": "VAEDecode", + "inputs": { + "samples": [ + "6", + 0 + ], + "vae": [ + "4", + 0 + ] + } + }, + "8": { + "class_type": "SaveImage", + "inputs": { + "images": [ + "7", + 0 + ], + "filename_prefix": "PacketFoundry-Qwen21Turbo" + } + }, + "9": { + "class_type": "RandomNoise", + "inputs": { + "noise_seed": 51 + } + }, + "10": { + "class_type": "CFGGuider", + "inputs": { + "model": [ + "5", + 0 + ], + "positive": [ + "3", + 0 + ], + "negative": [ + "3", + 1 + ], + "cfg": 1.0 + } + }, + "11": { + "class_type": "KSamplerSelect", + "inputs": { + "sampler_name": "euler" + } + }, + "12": { + "class_type": "ManualSigmas", + "inputs": { + "sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0" + } + } +} diff --git a/runs/bd801c85-11e3-5033-b076-acd84846b820/metadata.json b/runs/bd801c85-11e3-5033-b076-acd84846b820/metadata.json new file mode 100644 index 0000000..fe26c35 --- /dev/null +++ b/runs/bd801c85-11e3-5033-b076-acd84846b820/metadata.json @@ -0,0 +1,51 @@ +{ + "id": "bd801c85-11e3-5033-b076-acd84846b820", + "kind": "media-generation", + "title": "Predator media: Qwen Image 2.1 Turbo Q4 · ten-image seed/cache trial", + "date": "2026-10-21", + "started_at": "2026-10-21T08:41:18.067000+00:00", + "git_sha": "63be2f9b82ad0dac88a7ceaa28a9cf3154125a18", + "hardware": "predator", + "hardware_label": "Predator · GTX 1060 6 GB · 28 GB RAM · Windows", + "engine": "comfyui", + "harness": "packet-foundry-media-v1", + "model_family": "generation", + "model_sizes": [ + "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M" + ], + "task_kind": "media-generation", + "tags": [ + "image", + "video", + "music", + "retained-failures" + ], + "headline": "10 retained outputs · 10 accepted results · 10 recorded attempts · median Factory trace 202.0s", + "janie_blurb_md": "[Browse the complete campaign and media](/benchmarks/predator-media-20261021.html). [Read the findings](/findings/images-video-music-on-6gb.html).", + "janie_blurb_status": "published", + "caveat": "Technical delivery and media decoding do not establish visual or musical quality. Timings include adapter overhead; cache behavior differs between runs. Failed and cancelled attempts remain in the evidence.", + "caveat_severity": "warning", + "methodology_ref": "packet-foundry-media-v1", + "methodology_deviations_md": "Source commit identifies ComfyUI 0.42.1. Private bridge revisions and workflow hashes, where captured, are recorded separately in the campaign evidence. No clean matched-quality comparison or isolated quantization speedup is claimed.", + "results_table": [], + "cells": [], + "n_calls_total": 10, + "n_errors_total": 0, + "chart_spec": null, + "site_grade": "standard", + "site_grade_reason": "Retained media with explicit verification and quality limitations", + "raw_data_urls": { + "jsonl": "/data/predator-media-20261021/runs/bd801c85-11e3-5033-b076-acd84846b820/run.jsonl", + "md": "/data/predator-media-20261021/runs/bd801c85-11e3-5033-b076-acd84846b820/run.md", + "log": "/data/predator-media-20261021/failures.json", + "gitea_dir": "/data/predator-media-20261021/runs/bd801c85-11e3-5033-b076-acd84846b820/metadata.json" + }, + "synthesis_doc": "/findings/images-video-music-on-6gb.html", + "synthesis_docs_all": [ + "/findings/images-video-music-on-6gb.html" + ], + "related_ids": [], + "status": "complete", + "visibility": "public", + "media_group": "qwen21-turbo-q4km-ten" +} diff --git a/runs/bd801c85-11e3-5033-b076-acd84846b820/run.jsonl b/runs/bd801c85-11e3-5033-b076-acd84846b820/run.jsonl new file mode 100644 index 0000000..068b7dc --- /dev/null +++ b/runs/bd801c85-11e3-5033-b076-acd84846b820/run.jsonl @@ -0,0 +1,10 @@ +{"id": "qwen-turbo-01", "label": "Qwen Image 2.1 Turbo Q4_K_M · Initial request including model loading · seed 42", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 42, "status": "succeeded", "accepted_result": true, "prompt": "A small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph.", "prompt_details": {"word_count": 19, "complexity": "Very short / retained runtime pair", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 42}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 334.589337, "comfy_server_seconds": 320.246, "stages_seconds": {"node_4_observer_receive_interval": 4.436999999976251, "node_2_observer_receive_interval": 6.48399999999674, "node_3_observer_receive_interval": 104.32900000002701, "node_12_observer_receive_interval": 0.031000000017229468, "node_11_observer_receive_interval": 0.014999999955762178, "node_1_observer_receive_interval": 1.6570000000065193, "node_5_observer_receive_interval": 0.10899999999674037, "node_10_observer_receive_interval": 0, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 159.85899999999674, "node_7_observer_receive_interval": 40.75, "node_8_observer_receive_interval": 2.047000000020489}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": [], "same_runtime_retained": null, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "Initial request including model loading", "native_inter_run_gap_seconds": null, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": false, "cold_start_scope": "The first request includes initial/model work; OS disk-cache state was not isolated."}}, "resources": {"sampled_summary": {"sample_count": 162, "inflight_sample_count": 161, "minimum_ram_available_bytes": 1912422400, "minimum_commit_available_bytes": 11296174080, "minimum_system_drive_free_bytes": 18670813184, "maximum_sampled_device_gpu_used_mib": 3983}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent fox figurine standing on a pale wooden desk against a cream background; recognizably matches the simple scene.", "Orange fox, desk, cream background and soft studio light are present.", "Fine fur-like ridges dominate the body; ceramic material is not convincingly conveyed.", "Legs, ears and tail are largely plausible for a stylized figurine. Face is slightly toy-like.", "No text requested or visibly generated.", "Ceramic material appears more like textured wood, fur or a molded toy."], "media": [{"path": "media/qwen-turbo-01.png", "mime_type": "image/png", "size_bytes": 261046, "sha256": "56129ce43e04bb2aca5849e23b9d6a0c9adfdee3047a9329f1fcc360c2d9eb2e", "source_sha256": "c4a363cb2d3fbcb5d51b02226bdb57facc1963d98cac306cefb3ad252700b3d0", "width": 512, "height": 512, "pixel_sha256": "5df320c6bdbdad6dc8dbc07af73f2dc513a55deecad9b2f49b659e85ba5ba2f4", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-01.json", "workflow_sha256": "4369239d25533a14c3b5bb3a15d663b4049d3285e9804cc52483e7ffccaf4852", "source_workflow_canonical_sha256": "1ff73d5851a428f07381767c01cfdf4675c6c0b6f37e9df58ff08a9c74de77bb"} +{"id": "qwen-turbo-02", "label": "Qwen Image 2.1 Turbo Q4_K_M · Same brief, next seed · seed 43", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 43, "status": "succeeded", "accepted_result": true, "prompt": "A small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph.", "prompt_details": {"word_count": 19, "complexity": "Very short / retained runtime pair", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA small orange ceramic fox on a pale wooden desk, soft morning light, plain cream background, realistic studio photograph."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 43}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 96.670077, "comfy_server_seconds": 95.234, "stages_seconds": {"node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 67.14100000000326, "node_7_observer_receive_interval": 27.82800000003772, "node_8_observer_receive_interval": 0.06299999996554106}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "10", "11", "12", "2", "3", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": true, "context": {"workload": "Same brief, next seed", "native_inter_run_gap_seconds": 34.322, "only_seed_changed_from_predecessor": true, "text_changed_from_predecessor": false, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 48, "inflight_sample_count": 47, "minimum_ram_available_bytes": 8770334720, "minimum_commit_available_bytes": 11553648640, "minimum_system_drive_free_bytes": 18671108096, "maximum_sampled_device_gpu_used_mib": 4315}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": true, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent front-facing seated fox figurine in the same basic studio setup, with a visibly different pose and face.", "Orange fox, light wood and cream background are present; composition follows the brief.", "Matte textured surface with fur-like ridges; not strongly ceramic.", "Stylized anatomy is mostly coherent. Eye appearance is uneven, with a pale/cyan mark around the left visible eye.", "No text requested or visibly generated.", "Ceramic appearance is weak.", "Small asymmetric eye detail may be a visual artifact."], "media": [{"path": "media/qwen-turbo-02.png", "mime_type": "image/png", "size_bytes": 283217, "sha256": "238d33136a36561317196ed2a9c0096cbafe4109059ab2d0d8df5ec3079d74bd", "source_sha256": "f8bd3fcfba3462d120142e83fb0d7ec637a9eceebd25f6db11f61bda76254eb1", "width": 512, "height": 512, "pixel_sha256": "ac9f30aded9dc1c28b70d2b87cebd61a4df3d93e3dc62206d1ed31874410fe48", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-02.json", "workflow_sha256": "a48c407f70a6f7ffb6722d35f80c1d304116d197f2307961e7f88bbd31456fa4", "source_workflow_canonical_sha256": "84fabe49d8035de7dc9124d16ea9729ca181a0fdbca0f9e681520453dbb87ef2"} +{"id": "qwen-turbo-03", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 44", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 44, "status": "succeeded", "accepted_result": true, "prompt": "A turquoise alpine lake beneath snowy mountains at sunrise, a yellow canoe near the shore, crisp natural landscape photograph.", "prompt_details": {"word_count": 19, "complexity": "Short landscape", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA turquoise alpine lake beneath snowy mountains at sunrise, a yellow canoe near the shore, crisp natural landscape photograph."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA turquoise alpine lake beneath snowy mountains at sunrise, a yellow canoe near the shore, crisp natural landscape photograph."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 44}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 195.952306, "comfy_server_seconds": 194.865, "stages_seconds": {"node_3_observer_receive_interval": 100.20300000003772, "node_10_observer_receive_interval": 0.014999999955762178, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 66.93800000002375, "node_7_observer_receive_interval": 27.64100000000326, "node_8_observer_receive_interval": 0.045999999972991645}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 22.977, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 96, "inflight_sample_count": 95, "minimum_ram_available_bytes": 4403757056, "minimum_commit_available_bytes": 11454955520, "minimum_system_drive_free_bytes": 18670780416, "maximum_sampled_device_gpu_used_mib": 4315}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["Recognizable and visually coherent mountain-lake landscape with a yellow canoe-like object at the near shoreline.", "Turquoise water, snowy mountains and warm sunlight on peaks are present. The yellow canoe is close to shore.", "Water reflections and rock/snow separation are convincing at this size; fine granular texture is visible.", "The canoe is heavily cropped by the lower edge, showing mostly its bow/top; full canoe geometry cannot be assessed.", "No text requested or visibly generated.", "Canoe is strongly cropped and less clearly legible than a full near-shore canoe.", "Fine speckled texture is visible on the mountain surfaces."], "media": [{"path": "media/qwen-turbo-03.png", "mime_type": "image/png", "size_bytes": 474011, "sha256": "57418cb5f47b5647b8773c0e012572e528e150b0f27f422e78f4f05d5a683a11", "source_sha256": "c5e2e3901c3129df9943c2f995d5b42c7d646c708eb6f9053295778ed77012e7", "width": 512, "height": 512, "pixel_sha256": "ff22ec81f55d0251c62e225baca505db986d0d734728cb05f57b4d2087cd12d1", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-03.json", "workflow_sha256": "d26655caaebfa0bd74939c8f5f1977d3759141277d816b6eb1790a1c18f853c1", "source_workflow_canonical_sha256": "96b30904e6a2fa858c6bc001bf5a0b3e4168fbcbc36fcf53b796c7ec9f07cfd2"} +{"id": "qwen-turbo-04", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 45", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 45, "status": "succeeded", "accepted_result": true, "prompt": "Product photograph of a matte black ceramic teapot with a warm copper handle, resting on dark slate beside two pale linen napkins. Three-quarter view, softbox lighting, crisp material detail, uncluttered background.", "prompt_details": {"word_count": 31, "complexity": "Product materials", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nProduct photograph of a matte black ceramic teapot with a warm copper handle, resting on dark slate beside two pale linen napkins. Three-quarter view, softbox lighting, crisp material detail, uncluttered background."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nProduct photograph of a matte black ceramic teapot with a warm copper handle, resting on dark slate beside two pale linen napkins. Three-quarter view, softbox lighting, crisp material detail, uncluttered background."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 45}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 198.952217, "comfy_server_seconds": 196.417, "stages_seconds": {"node_3_observer_receive_interval": 100.10899999999674, "node_10_observer_receive_interval": 0.01600000000325963, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 67.98399999999674, "node_7_observer_receive_interval": 28.21899999998277, "node_8_observer_receive_interval": 0.061999999976251274}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 16.828, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 98, "inflight_sample_count": 97, "minimum_ram_available_bytes": 4163956736, "minimum_commit_available_bytes": 11323981824, "minimum_system_drive_free_bytes": 18670002176, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent, usable-looking product composition of a black teapot, copper-colored handle, dark slate and pale linen.", "Main object, materials, slate and three-quarter product lighting are present. Only one clearly distinguishable rolled/folded napkin appears, rather than two separate napkins.", "Black matte body, metallic handle and woven linen are well differentiated.", "Teapot body, lid, spout and handle joins are broadly plausible; no obvious broken geometry at 512px.", "No text requested or visibly generated.", "Requested count of two napkins is not clearly met."], "media": [{"path": "media/qwen-turbo-04.png", "mime_type": "image/png", "size_bytes": 280344, "sha256": "e5b3093fdb6bb39fcec4b803b1345733cf212bb274806338d27e140d3b882c3d", "source_sha256": "1c2b9ddeca53a1fd944073a596b050e43f8f9b1b4a7e2fee57608095395b97d3", "width": 512, "height": 512, "pixel_sha256": "2db9b0cf39ba6fdb3f0febf48a4f32d9fbd8012ed303d8d3ec711a4a074f8b81", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-04.json", "workflow_sha256": "434243d6ae778adde215a2487cc10e1ee4123a6826f0f50c21e1b41238930d75", "source_workflow_canonical_sha256": "7592e027ed1ef1f6903c6ca2bdae9faad31a02bf99cd7f67c71689cc3427c0ce"} +{"id": "qwen-turbo-05", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 46", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 46, "status": "succeeded", "accepted_result": true, "prompt": "A cozy neighborhood bakery storefront on a rainy evening. The main sign clearly reads \"MORNING BREAD\" in cream block letters. Warm light glows through the windows, a bicycle leans beside the door, wet paving reflects the orange interior, realistic street photograph.", "prompt_details": {"word_count": 41, "complexity": "Readable sign", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA cozy neighborhood bakery storefront on a rainy evening. The main sign clearly reads \"MORNING BREAD\" in cream block letters. Warm light glows through the windows, a bicycle leans beside the door, wet paving reflects the orange interior, realistic street photograph."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA cozy neighborhood bakery storefront on a rainy evening. The main sign clearly reads \"MORNING BREAD\" in cream block letters. Warm light glows through the windows, a bicycle leans beside the door, wet paving reflects the orange interior, realistic street photograph."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 46}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 201.997778, "comfy_server_seconds": 199.64, "stages_seconds": {"node_3_observer_receive_interval": 102.10899999999674, "node_10_observer_receive_interval": 0.01600000000325963, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 68.81200000003446, "node_7_observer_receive_interval": 28.625, "node_8_observer_receive_interval": 0.06299999996554106}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 16.283, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 99, "inflight_sample_count": 98, "minimum_ram_available_bytes": 4295344128, "minimum_commit_available_bytes": 11412852736, "minimum_system_drive_free_bytes": 18669809664, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent rainy-evening bakery storefront with a legible sign, bicycle, warm windows and wet paving.", "The main sign visibly reads exactly MORNING BREAD in large cream block letters. Bicycle, orange interior and reflected light are present.", "Brick, glazing and wet paving read distinctly; the dark storefront feels photograph-like.", "Storefront perspective and bicycle placement are broadly plausible at this scale; fine bicycle detail is somewhat soft.", "Exact requested wording MORNING BREAD is visibly correct. This is a subjective visual transcription, not OCR or an automated text benchmark.", "Fine bicycle/spoke details are soft at the supplied resolution."], "media": [{"path": "media/qwen-turbo-05.png", "mime_type": "image/png", "size_bytes": 434268, "sha256": "94e64b09278ffff4feb04c3197545dd8d6371b281a514d0e3d1cf8550648d919", "source_sha256": "5a9008f775161ef48b16915597d158f4a0cd80883a7c4d452cbf240bba049ce2", "width": 512, "height": 512, "pixel_sha256": "18129fd4fc85277093ff6d330be029055bfde66f2a18af657ce735b6e928f894", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-05.json", "workflow_sha256": "8768f247abc378db4fd3057cb67987743253489f11b42e75bef673f1b6c39658", "source_workflow_canonical_sha256": "874d22b62ef466732130e9e79fa45e12bf32d748a3b1124d21b16aedac7abf59"} +{"id": "qwen-turbo-06", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 47", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 47, "status": "succeeded", "accepted_result": true, "prompt": "A bright Victorian glass greenhouse filled with lush ferns, hanging orchids and terracotta pots. A narrow brick path curves from the bottom left toward an open green door at the far end. Morning sunlight makes geometric shadows through the iron window frames. Include a single brass watering can beside the path and misty trees beyond the glass. Architectural photograph, balanced natural colors, no people or text.", "prompt_details": {"word_count": 66, "complexity": "Botanical architecture", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA bright Victorian glass greenhouse filled with lush ferns, hanging orchids and terracotta pots. A narrow brick path curves from the bottom left toward an open green door at the far end. Morning sunlight makes geometric shadows through the iron window frames. Include a single brass watering can beside the path and misty trees beyond the glass. Architectural photograph, balanced natural colors, no people or text."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA bright Victorian glass greenhouse filled with lush ferns, hanging orchids and terracotta pots. A narrow brick path curves from the bottom left toward an open green door at the far end. Morning sunlight makes geometric shadows through the iron window frames. Include a single brass watering can beside the path and misty trees beyond the glass. Architectural photograph, balanced natural colors, no people or text."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 47}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 201.931, "comfy_server_seconds": 199.412, "stages_seconds": {"node_3_observer_receive_interval": 102.79699999996228, "node_10_observer_receive_interval": 0.015000000013969839, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 68.43800000002375, "node_7_observer_receive_interval": 28.04699999996228, "node_8_observer_receive_interval": 0.062000000034458935}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 21.517, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 99, "inflight_sample_count": 98, "minimum_ram_available_bytes": 4470677504, "minimum_commit_available_bytes": 11370139648, "minimum_system_drive_free_bytes": 18651901952, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent glass greenhouse with dense foliage, terracotta pots, a brick walkway, an open green door and a brass-colored watering can.", "Most named scene elements are present. The path leads toward an open far door; ferns and hanging plants dominate. Distinct hanging orchid flowers are not obvious, and geometric sunlight shadows are subtle.", "Glass/metal frames, foliage, pots and brick are distinguishable, but foliage and frame textures have noticeable fine speckling.", "Overall greenhouse perspective is convincing; dense plant/frame detail is less clean on close inspection. Watering can is partly occluded/near the left edge.", "No people or text requested; none visibly apparent.", "Requested orchids are not clearly identifiable.", "Geometric cast shadows are weaker than the brief suggests.", "Dense foliage and framing show grainy/speckled texture."], "media": [{"path": "media/qwen-turbo-06.png", "mime_type": "image/png", "size_bytes": 649441, "sha256": "f1ae1c3294bcffe1c0d283e957bcc81c5d48df83fe6e0f316eb4182d57db2af1", "source_sha256": "bd1ac8888861a03d6b9a2436cede0ad7b3009f1535c11b360f794ad47a9c0321", "width": 512, "height": 512, "pixel_sha256": "f74bd726756ed8f29d230b4c506533b708c21ecd53275092c3e54672dbb5a1cd", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-06.json", "workflow_sha256": "c48538fbde2bdf10bb5849c8f29aaa79f9f05da8b1e53518af7b56789aa396a0", "source_workflow_canonical_sha256": "6cc4ea59bd3d96d0811844809d534eda117290b8b705f13f0069b45cb5d2da6b"} +{"id": "qwen-turbo-07", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 48", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 48, "status": "succeeded", "accepted_result": true, "prompt": "An editorial portrait of an adult female ceramic artist in her forties working in a sunlit pottery studio. She wears a blue linen apron over a cream shirt and calmly holds a small unfinished bowl in both hands, fingers naturally curled around the rim. Her face is clear and expressive, with a few silver strands in dark curly hair. Shelves of glazed pottery remain softly out of focus behind her. Frame from the waist upward, eye-level camera, gentle window light from the left, documentary realism.", "prompt_details": {"word_count": 85, "complexity": "Portrait / hands", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn editorial portrait of an adult female ceramic artist in her forties working in a sunlit pottery studio. She wears a blue linen apron over a cream shirt and calmly holds a small unfinished bowl in both hands, fingers naturally curled around the rim. Her face is clear and expressive, with a few silver strands in dark curly hair. Shelves of glazed pottery remain softly out of focus behind her. Frame from the waist upward, eye-level camera, gentle window light from the left, documentary realism."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nAn editorial portrait of an adult female ceramic artist in her forties working in a sunlit pottery studio. She wears a blue linen apron over a cream shirt and calmly holds a small unfinished bowl in both hands, fingers naturally curled around the rim. Her face is clear and expressive, with a few silver strands in dark curly hair. Shelves of glazed pottery remain softly out of focus behind her. Frame from the waist upward, eye-level camera, gentle window light from the left, documentary realism."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 48}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 201.952675, "comfy_server_seconds": 199.784, "stages_seconds": {"node_3_observer_receive_interval": 103.42200000002049, "node_10_observer_receive_interval": 0.015000000013969839, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 68.59399999998277, "node_7_observer_receive_interval": 27.65600000001723, "node_8_observer_receive_interval": 0.0779999999795109}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 13.667, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 99, "inflight_sample_count": 98, "minimum_ram_available_bytes": 4176404480, "minimum_commit_available_bytes": 11358388224, "minimum_system_drive_free_bytes": 18671104000, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent and convincing waist-up studio portrait, with a clear adult face and a bowl held in two hands.", "Blue apron, cream shirt, dark curly hair, pottery shelves, eye-level framing and soft window light from the left all appear. Silver hair strands are not clearly visible. Hands support the bowl from the side and underneath rather than both curling around its rim.", "Cloth texture, skin and unfinished pottery are visibly differentiated; the bowl has a plausible unglazed look.", "Hand shapes, wrist connections and bowl contact look broadly plausible at 512px. Some fingertips overlap or are occluded, so an exact finger-count claim is unwarranted. No obvious extra limb.", "No text requested or visibly generated.", "Specific silver hair detail is not clear.", "The precise requested fingers-around-rim pose is only partly followed."], "media": [{"path": "media/qwen-turbo-07.png", "mime_type": "image/png", "size_bytes": 418100, "sha256": "7b4feea8f758374a5544ad0e9078c8455f32f5a48e8305256ab0839f2afd9472", "source_sha256": "ed1e6c70637edfc74eb3f051c51632087649ce6e88c705b2c4f08f1ee81d756a", "width": 512, "height": 512, "pixel_sha256": "23fe51347e19cff881406c1152d713a2773a46cd9032f54cabc9dc4b5129c108", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-07.json", "workflow_sha256": "23cb0fb14fb4cf31e9a4a67434ac21ab3e74b9858411f7e5724aaf8abbbbb755", "source_workflow_canonical_sha256": "fc32cc9a0f7f8172edfda1400d2b8083ecbf5c86b636b5db6db024c9d20d4142"} +{"id": "qwen-turbo-08", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 49", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 49, "status": "succeeded", "accepted_result": true, "prompt": "A charming small brass robot with a round glass head stands on a wooden step stool inside an enormous old library, carefully returning a red book to a high shelf. The robot has two articulated arms and three visible fingers on each hand. A tiny blue mechanical bird perches on its shoulder. Bookshelves curve toward a domed ceiling painted with constellations; the far aisle is illuminated by warm amber lamps. Render as a detailed hand-painted storybook illustration with teal shadows, golden highlights and a clear focal point. Keep the robot, stool and book visibly separate, with no extra limbs, no captions and no written lettering.", "prompt_details": {"word_count": 105, "complexity": "Stylized story scene", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA charming small brass robot with a round glass head stands on a wooden step stool inside an enormous old library, carefully returning a red book to a high shelf. The robot has two articulated arms and three visible fingers on each hand. A tiny blue mechanical bird perches on its shoulder. Bookshelves curve toward a domed ceiling painted with constellations; the far aisle is illuminated by warm amber lamps. Render as a detailed hand-painted storybook illustration with teal shadows, golden highlights and a clear focal point. Keep the robot, stool and book visibly separate, with no extra limbs, no captions and no written lettering."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA charming small brass robot with a round glass head stands on a wooden step stool inside an enormous old library, carefully returning a red book to a high shelf. The robot has two articulated arms and three visible fingers on each hand. A tiny blue mechanical bird perches on its shoulder. Bookshelves curve toward a domed ceiling painted with constellations; the far aisle is illuminated by warm amber lamps. Render as a detailed hand-painted storybook illustration with teal shadows, golden highlights and a clear focal point. Keep the robot, stool and book visibly separate, with no extra limbs, no captions and no written lettering."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 49}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 210.875124, "comfy_server_seconds": 210.023, "stages_seconds": {"node_3_observer_receive_interval": 107.75, "node_10_observer_receive_interval": 0.01600000000325963, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 72.96900000004098, "node_7_observer_receive_interval": 29.20299999997951, "node_8_observer_receive_interval": 0.061999999976251274}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 14.441, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 104, "inflight_sample_count": 103, "minimum_ram_available_bytes": 4373700608, "minimum_commit_available_bytes": 11362136064, "minimum_system_drive_free_bytes": 18670817280, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent and attractive storybook-style library scene: brass robot, glass-like head, wooden stool, red book, blue shoulder bird and long illuminated aisle are all recognizable.", "The specific main objects, articulated arms, teal/gold palette, warm far lamps and ceiling decoration are strongly represented. The exact three-visible-fingers-per-hand requirement is hard to certify because several small digits overlap or turn away.", "Brass, a reflective glass dome, wood and book covers read distinctly within the painted style.", "Robot, stool, book and shelves remain visibly separate. Arm joints and stool perspective are broadly coherent. Finger-count correctness cannot be established confidently at the supplied resolution.", "No legible caption, title or written lettering is apparent.", "Exact requested finger count remains uncertain; small jointed digits overlap.", "Glass head is mainly reflective/opaque rather than clearly transparent, but still reads as a glass dome."], "media": [{"path": "media/qwen-turbo-08.png", "mime_type": "image/png", "size_bytes": 518714, "sha256": "bd1bcc5079a48be9f59a4d72511eae46abb6e2aa788fe22f72bbac1cc4ba8b7a", "source_sha256": "63abbaf5c49b2baca103bd65761fd4fed09f7e5f173406e91eaaf47d136694f2", "width": 512, "height": 512, "pixel_sha256": "224125f9ca31c3c15dfa975dd61d42fa8f1703d9c012374876d2242054759658", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-08.json", "workflow_sha256": "000bdfef44a50c9cd6275031d4ab57a84f36f216784d1f8c78d637eee60c820a", "source_workflow_canonical_sha256": "3a9b68b7e06cf4b2a375d516a8f5d38b555d7ba5e593354edce22542adf12865"} +{"id": "qwen-turbo-09", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 50", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 50, "status": "succeeded", "accepted_result": true, "prompt": "A detailed travel photograph of a lively Mediterranean open-air produce market in early morning. In the foreground, three wooden crates hold red tomatoes, striped eggplants and yellow lemons, each type kept in its own crate. An adult vendor wearing a plain white shirt stands behind the stall, handing a small paper bag to an adult customer in a green jacket. A striped blue-and-white canopy casts cool shade over the stall. Beyond them, a narrow pedestrian street slopes toward a distant harbor with two white sailboats. Warm sunlight touches the upper walls but the foreground remains softly shaded. Compose the scene in layers with natural human proportions, believable hands, clear separation between people, baskets and produce, and rich but realistic color. Avoid artificial lens flares, watermarks and unreadable signs.", "prompt_details": {"word_count": 128, "complexity": "Crowded composition", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA detailed travel photograph of a lively Mediterranean open-air produce market in early morning. In the foreground, three wooden crates hold red tomatoes, striped eggplants and yellow lemons, each type kept in its own crate. An adult vendor wearing a plain white shirt stands behind the stall, handing a small paper bag to an adult customer in a green jacket. A striped blue-and-white canopy casts cool shade over the stall. Beyond them, a narrow pedestrian street slopes toward a distant harbor with two white sailboats. Warm sunlight touches the upper walls but the foreground remains softly shaded. Compose the scene in layers with natural human proportions, believable hands, clear separation between people, baskets and produce, and rich but realistic color. Avoid artificial lens flares, watermarks and unreadable signs."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nA detailed travel photograph of a lively Mediterranean open-air produce market in early morning. In the foreground, three wooden crates hold red tomatoes, striped eggplants and yellow lemons, each type kept in its own crate. An adult vendor wearing a plain white shirt stands behind the stall, handing a small paper bag to an adult customer in a green jacket. A striped blue-and-white canopy casts cool shade over the stall. Beyond them, a narrow pedestrian street slopes toward a distant harbor with two white sailboats. Warm sunlight touches the upper walls but the foreground remains softly shaded. Compose the scene in layers with natural human proportions, believable hands, clear separation between people, baskets and produce, and rich but realistic color. Avoid artificial lens flares, watermarks and unreadable signs."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 50}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 210.965741, "comfy_server_seconds": 208.566, "stages_seconds": {"node_3_observer_receive_interval": 107.82800000003772, "node_10_observer_receive_interval": 0, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 72.20299999997951, "node_7_observer_receive_interval": 28.43800000002375, "node_8_observer_receive_interval": 0.061999999976251274}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 13.101, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 103, "inflight_sample_count": 102, "minimum_ram_available_bytes": 4364185600, "minimum_commit_available_bytes": 11356860416, "minimum_system_drive_free_bytes": 18670198784, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["Recognizable market transaction under a striped canopy, with three separate produce crates and a harbor beyond. Main scene layers are coherent.", "White-shirt vendor, green-jacket customer, blue-and-white canopy, separate tomatoes/eggplants/lemons and sunlit upper wall over shaded foreground are all present. The paper bag and precise handoff are visually ambiguous. Harbor sailboats are visible but the exact requested count of two is not certifiable at this size.", "Vegetables, wood, clothing and canopy fabric are distinguishable. Fine background detail is smeared/brushy despite a photographic brief.", "Main adult body proportions look plausible. The central touching hands are soft and partially merged/occluded; finger anatomy and grasp cannot be verified. Background masts and architecture become muddled.", "No prominent legible sign, watermark or caption appears.", "Paper-bag handoff is less clear than requested.", "Hands and distant harbor details are imprecise.", "Exact two-sailboat count remains uncertain."], "media": [{"path": "media/qwen-turbo-09.png", "mime_type": "image/png", "size_bytes": 437788, "sha256": "847ec6dd9be4503a94967f96ca4b91d223ffb7638749c4fd08fb318723f544e0", "source_sha256": "5b992fa647b0461c1eb8514854a32b5f44cdd297a129a110f72b1abff6a830ef", "width": 512, "height": 512, "pixel_sha256": "ffaa52b367e79221e07aeed4066dcadc2039859372848f38c9bc028d623b7f50", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-09.json", "workflow_sha256": "049b53d4c92125dbe083563cea3d836330cfd864f19317933f71cb8794bc7db6", "source_workflow_canonical_sha256": "e1fd339039ed8d1dde53d4a0668665f4b78470cce872ff0c3f831156539dd328"} +{"id": "qwen-turbo-10", "label": "Qwen Image 2.1 Turbo Q4_K_M · New brief, retained model runtime · seed 51", "group": "qwen21-turbo-q4km-ten", "kind": "image", "model_id": "qwen-turbo", "model": "unsloth/Qwen-Image-2.1-Turbo-GGUF:Q4_K_M", "seed": 51, "status": "succeeded", "accepted_result": true, "prompt": "Create a highly detailed cinematic architectural illustration of a remote mountain observatory at blue hour after a light snowfall. The building occupies the right two-thirds of the square image: a circular pale-stone base supports one weathered copper dome, whose open viewing slit points toward the upper left. A narrow exterior staircase with dark iron handrails climbs from a small lower entrance to a balcony; every step has a thin undisturbed layer of snow. One adult astronomer in a dark red winter coat stands on the balcony with a small white notebook held closed in one gloved hand, looking toward the sky. Place a warm amber lamp immediately above the entrance, creating a pool of light on the stone without illuminating the whole landscape. In the lower left foreground, a crooked pine tree and two exposed granite rocks frame a winding path that leads toward the building. Behind the observatory, show distant mountain ridges fading from slate blue into violet mist. The sky contains scattered sharp stars and a faint natural Milky Way band, with no oversized planets or fantasy moons. Balance warm interior light against cold ambient snow light, use physically plausible perspective, careful separation of the staircase rails and the figure, and visible differences between copper, stone, glass, fabric and snow. The final scene should feel quiet, inhabited and believable. Do not add titles, labels, logos, watermarks, extra people, extra domes or duplicate staircases.", "prompt_details": {"word_count": 235, "complexity": "Elaborate constraints", "conditioning_texts": {"3.prompt": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nCreate a highly detailed cinematic architectural illustration of a remote mountain observatory at blue hour after a light snowfall. The building occupies the right two-thirds of the square image: a circular pale-stone base supports one weathered copper dome, whose open viewing slit points toward the upper left. A narrow exterior staircase with dark iron handrails climbs from a small lower entrance to a balcony; every step has a thin undisturbed layer of snow. One adult astronomer in a dark red winter coat stands on the balcony with a small white notebook held closed in one gloved hand, looking toward the sky. Place a warm amber lamp immediately above the entrance, creating a pool of light on the stone without illuminating the whole landscape. In the lower left foreground, a crooked pine tree and two exposed granite rocks frame a winding path that leads toward the building. Behind the observatory, show distant mountain ridges fading from slate blue into violet mist. The sky contains scattered sharp stars and a faint natural Milky Way band, with no oversized planets or fantasy moons. Balance warm interior light against cold ambient snow light, use physically plausible perspective, careful separation of the staircase rails and the figure, and visible differences between copper, stone, glass, fabric and snow. The final scene should feel quiet, inhabited and believable. Do not add titles, labels, logos, watermarks, extra people, extra domes or duplicate staircases."}, "actual_positive_text": "Create a single image from the supplied scene description.\n\nSource generation_brief:\nCreate a highly detailed cinematic architectural illustration of a remote mountain observatory at blue hour after a light snowfall. The building occupies the right two-thirds of the square image: a circular pale-stone base supports one weathered copper dome, whose open viewing slit points toward the upper left. A narrow exterior staircase with dark iron handrails climbs from a small lower entrance to a balcony; every step has a thin undisturbed layer of snow. One adult astronomer in a dark red winter coat stands on the balcony with a small white notebook held closed in one gloved hand, looking toward the sky. Place a warm amber lamp immediately above the entrance, creating a pool of light on the stone without illuminating the whole landscape. In the lower left foreground, a crooked pine tree and two exposed granite rocks frame a winding path that leads toward the building. Behind the observatory, show distant mountain ridges fading from slate blue into violet mist. The sky contains scattered sharp stars and a faint natural Milky Way band, with no oversized planets or fantasy moons. Balance warm interior light against cold ambient snow light, use physically plausible perspective, careful separation of the staircase rails and the figure, and visible differences between copper, stone, glass, fabric and snow. The final scene should feel quiet, inhabited and believable. Do not add titles, labels, logos, watermarks, extra people, extra domes or duplicate staircases."}, "parameters": {"1 UnetLoaderGGUFAdvanced": {"unet_name": "qwen-image-2.1-turbo-Q4_K_M.gguf", "dequant_dtype": "float32", "patch_dtype": "float32", "patch_on_device": false}, "2 CLIPLoader": {"clip_name": "qwen3vl_8b_int8_convrot.safetensors", "type": "qwen_image", "device": "cpu"}, "3 TextEncodeQwenImage21": {"negative_prompt": "", "resolution": 512}, "4 VAELoader": {"vae_name": "qwen_image_2.1_vae_bf16.safetensors"}, "5 QwenImage21Cache": {"device": "off", "dtype": "default"}, "6 SamplerCustomAdvanced": {}, "7 VAEDecode": {}, "8 SaveImage": {}, "9 RandomNoise": {"noise_seed": 51}, "10 CFGGuider": {"cfg": 1.0}, "11 KSamplerSelect": {"sampler_name": "euler"}, "12 ManualSigmas": {"sigmas": "1.0, 0.978453, 0.95418, 0.926626, 0.89508, 0.845148, 0.704534, 0.414568, 0"}}, "timing": {"factory_trace_seconds": 226.020651, "comfy_server_seconds": 223.135, "stages_seconds": {"node_3_observer_receive_interval": 115.98399999999674, "node_10_observer_receive_interval": 0.01600000000325963, "node_9_observer_receive_interval": 0, "node_6_observer_receive_interval": 78.40600000001723, "node_7_observer_receive_interval": 28.639999999955762, "node_8_observer_receive_interval": 0.07900000002700835}, "limits": "Native server boundaries, factory attempt traces and observed delivery have different timing scopes. Observer node intervals include loading/preparation/event delivery, not isolated kernel or weight-only times."}, "cache": {"cached_nodes": ["1", "11", "12", "2", "4", "5"], "same_runtime_retained": true, "immediate_warm_claim": false, "encoder_cached": false, "context": {"workload": "New brief, retained model runtime", "native_inter_run_gap_seconds": 10.771, "only_seed_changed_from_predecessor": false, "text_changed_from_predecessor": true, "cold_start_scope": null}}, "resources": {"sampled_summary": {"sample_count": 111, "inflight_sample_count": 110, "minimum_ram_available_bytes": 4460896256, "minimum_commit_available_bytes": 11257696256, "minimum_system_drive_free_bytes": 18669465600, "maximum_sampled_device_gpu_used_mib": 4351}, "limits": "Device-wide GPU observations include display and other workloads. These sampled extrema do not establish dedicated-VRAM residency or absence of shared-memory paging."}, "encoding": {"encoder_cached": false, "token_truncation_verified": false, "interpretation": "Exact Qwen encoder token truncation was not instrumented. The complete actual composed conditioning text is retained."}, "verification": {"full_decode": true, "checksum": true, "acknowledged": true, "trace_attachment": true, "native_experiment": true, "independent_public_copy_decode": true, "visual_review": true, "quality_scope": "Independent subjective inspection of the original 512×512 PNG; no scored quality benchmark and no causal attribution to quantization or schedule."}, "findings": ["A coherent quiet blue-hour observatory scene with one large dome, one red-coated figure, one snow-covered staircase, foreground pine/rocks and distant layered mountains.", "Building placement, open slit toward the upper left, red coat, small light-colored notebook, amber entrance lamp, snow on steps and cold/warm lighting contrast are represented. The figure appears to attend to the notebook rather than clearly look at the sky. Requested stars and faint Milky Way are not visibly present. The exact two-rock count is not cleanly met or certifiable among overlapping foreground boulders.", "Dome reads as dark weathered metal, base as smooth pale masonry/stucco, and snow, coat and rails are distinct. Fine stone material detail is less explicit than the brief.", "Overall building, balcony and staircase perspective are plausible; thin rail details and figure-to-railing separation are imperfect but readable. No obvious extra person, dome or duplicate staircase.", "No title, label, logo or watermark is apparent.", "Stars and Milky Way are absent from the visibly plain sky.", "Figure does not clearly follow the requested skyward gaze.", "Exact rock-count and some fine architectural/material details are weaker than the broad composition."], "media": [{"path": "media/qwen-turbo-10.png", "mime_type": "image/png", "size_bytes": 402891, "sha256": "7bc1033039962507ee2aa153ebfacaa4cdd1f77e93e152bd7241ebae686b85bc", "source_sha256": "f051ec1ab055f23d3fd6513e3428d807734a846f5e51881403cf419ccb33c161", "width": 512, "height": 512, "pixel_sha256": "9672ee9777924c1b99ccff8952362c0119f99a87bb1e22c7bc82a24b579ccb6f", "metadata_handling": "PNG re-encoded losslessly without ancillary metadata; decoded pixels unchanged.", "public_copy_full_decode_passed": true, "source_content_preserved": "Decoded pixels byte-identical"}], "workflow_path": "workflows/qwen-turbo-10.json", "workflow_sha256": "11fe5261e6abff03fd6a82be45d93b3827c421e3cc3e684f03b72df24694e0b0", "source_workflow_canonical_sha256": "7120b879556ee04ac859a6126f8d1ab24771a28bf208d58fff10b120bf50e68a"} diff --git a/runs/bd801c85-11e3-5033-b076-acd84846b820/run.md b/runs/bd801c85-11e3-5033-b076-acd84846b820/run.md new file mode 100644 index 0000000..33fd5d4 --- /dev/null +++ b/runs/bd801c85-11e3-5033-b076-acd84846b820/run.md @@ -0,0 +1,9 @@ +# Predator media: Qwen Image 2.1 Turbo Q4 · ten-image seed/cache trial + +10 retained outputs · 10 accepted results · 10 recorded attempts · median Factory trace 202.0s + +Technical delivery and media decoding do not establish visual or musical quality. Timings include adapter overhead; cache behavior differs between runs. Failed and cancelled attempts remain in the evidence. + +ComfyUI source: `63be2f9b82ad0dac88a7ceaa28a9cf3154125a18`. Per-run prompts, workflow hashes, model versions, timing boundaries, cache evidence, resource observations and verification are in this group’s run.jsonl. Media-generation records use the campaign schema, not the LLM token-throughput schema. Private raw operator logs are excluded; sanitized failed attempts remain. + +[Methodology](https://benchmarks.weeyuga.com/methodology.html#packet-foundry-media-v1) · [All media](https://benchmarks.weeyuga.com/benchmarks/predator-media-20261021.html) · [Findings](https://benchmarks.weeyuga.com/findings/images-video-music-on-6gb.html)