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