Re-read /home/zwitschi/StabilityMatrix/Data/README.md - it now documents 141 GB of models and four
verified, core-node image/texture graphs - and wired the ones that matter into the toolchain.
comfyui_generate.py gains a 'texture' subcommand that rebuilds the API graph for each documented workflow,
because the files in Workflows/ are ComfyUI UI format and /prompt cannot accept them:
albedobase AlbedoBase XL v3.1, 30 steps, CFG 6.0, dpmpp_2m/karras, baked VAE
krea2 Krea-2 Turbo NVFP4, 8 steps, CFG 1.0, euler/simple, ConditioningZeroOut, texture-fix VAE
qwen21 Qwen-Image 2.1 int8, 25 steps, CFG 1.0, euler/simple, TextEncodeQwenImage21, texture-fix VAE
Default pair is albedobase + krea2 - both commercially usable - generated from the same prompt and seed,
scored by high-frequency detail, and compared before the winner takes --out. Smoke-tested at 512 px: both
rendered, albedobase won 18.91 against 17.51, and the resulting limestone albedo is a plausible material
sheet.
Tools/README.md documents the profiles with their settings, a licence row per profile (Qwen-Image 2.1 is
research-only and must not ship), and the two limits the workflows carry: they produce albedo only, and
ComfyUI 0.44 has no seamless sampler, so tiling has to be checked by offsetting the image 50%. Five tests
cover the graph shapes; the suite is at 108.
Tools/README.md now covers the generative tools next to each other:
- comfyui_generate.py drives the LOCAL ComfyUI at 127.0.0.1:8188, which this repository does not start. A new
subsection says so and says what to do: start Stability Matrix, launch ComfyUI, re-run. It points at
/home/zwitschi/StabilityMatrix/Data/README.md for the package version, the launch arguments and the shared
80 GB model library, and notes that THIRD_PARTY_NOTICES.md must carry the model and licence before a
generated texture ships.
- generate_facade_textures.py and generate_character_textures.py are listed as the deterministic procedural
path: no model, no network, byte-identical on every run.
- The tool's own docstring carries the same Stability Matrix instruction.
The default second checkpoint was sd_xl_base_1.0.safetensors, which is NOT installed; it is now
epicrealismXL_pureFix.safetensors, verified against the running server. The pair was smoke-tested end to end
at 512 px: both checkpoints rendered, both scored, juggernautXL won at 20.37 against 12.93, and the winner
landed on --out with the loser kept beside it.
A single vision model has been wrong in both directions on this project - it missed real defects and
disputed frames that were correct - so agreement between different model families is worth more than one
confident opinion.
judge_screenshot.py now runs two models by default (openai/gpt-4o-mini + qwen/qwen3.7-flash, both served
on OpenRouter; claude-3.5-haiku is gone, which the first attempt discovered with a 404). combine()
reconciles them: agreement carries the verdict with the mean confidence, disagreement reports 'uncertain'
and keeps both reasons rather than letting the more confident model win. --model still forces a single
judgement, and every line prints which model said what.
comfyui_generate.py generates the same prompt and seed with two checkpoints, scores both by
high-frequency detail (mean absolute Laplacian, which catches blurry or flat results), keeps the loser as
<out>__<model>.png with its sidecar, and moves the winner onto --out. --alt-model restores the single
run.
Tests cover the reconciliation - agreement, disagreement, unknown verdicts, and no verdicts at all - and the
suite is at 103.
Two tools that turn character work into something measurable.
- Tools/comfyui_character_blueprint.py generates a four-view sheet (front, side, back,
head) plus a contact sheet from one description, with fixed seeds and a JSON record
of prompts and prompt ids. First sheet: 'GigLine walker', lean pedestrian in a navy
tee, jeans and sneakers.
- Tools/character_match.py compares a rendered character against a sheet: figure
aspect, fill and coverage deltas, mean CIE76 colour distance, dominant-palette
distance, and a side-by-side montage. The deltas stay separate on purpose - a single
similarity score between a photoreal sheet and a low-detail body would be a number
with no meaning.
- comfyui_generate.py grew wait_for_image() so both tools share the polling path
instead of duplicating it.
- 13 unit tests for the colour maths, masks and palette; 95 tool tests pass.
First real result, recorded rather than smoothed over: aspect delta 1.24, fill delta
0.12, mean deltaE 31.9, palette deltaE 25.4. The montage shows the reason - the model
is a nude body and the blueprint wears a tee and jeans, so clothing dominates every
metric. Figure segmentation is explicit (--render-box): auto-detection picked backdrop
blobs on every attempt, and a background plate is the proper fix.
Tools/comfyui_generate.py drives the local ComfyUI instance for generated textures
and writes a sidecar with model, prompt, seed and workflow, so a generated asset can
be regenerated or rejected instead of being an unexplained PNG. Models list and
generation both verified against 127.0.0.1:8188.
Stable Diffusion is not usable for flat skin swatches, and that is worth recording:
three attempts produced an anatomical close-up twice and a leather-scale surface once.
Skin detail is generated procedurally instead - tileable octave noise for pores, plus
a strand sheet for hair - in Tools/generate_character_textures.py, with sidecars.
Hair geometry exists (cap conforming to the head, weighted to the head bone, second
material slot in Blender) but Unreal's import collapses the second material slot, so
the cap renders with the skin material and reads as a bald skull. It is behind
--hair and off; the next step is a separate skeletal mesh attached as a second
component.
zwitschi