First GPU pass on the rev-3 depth pipeline surfaced two issues; both fixed
and re-rendered clean:
1. Blown near-field highlights (regression from 018909b). Mapping the
nearest pixel to pure white (nearPlane=nearRaw) made the depth
ControlNet render overexposed near floors/ceilings at strength 0.85.
fpv.js: nearPlane=0 (anchor white at the camera; nearest real surface
≈gray 214, headroom intact) + minSpan 5→8. Re-bake confirms the blown
ceiling is gone while recesses still reach black and geometry holds.
2. Phantom doors on closed walls — a PROMPT problem, not the depth map
(present in old maps too). Opening NOUNS in the positive prompt summon
a passage onto solid stone: threshold's "gatehouse, portcullis,
entrance" painted a gate on a blank wall, and at cfg 1.0 the negative
can't cancel it. bake_atlas.py: THEME_BODY_CLOSED (full theme flavor,
zero opening nouns) for closed facings + opening-nouns negated on
closed facings only. r0_N (threshold) now renders clean ashlar wall.
("Bricked-up archway" backfires — "archway" alone re-summons the arch.)
Also: sweep_strength.py re-exec-under-venv guard (the venv python is a
symlink to the base interpreter, so the realpath check wrongly skipped
the hop and the montage died after every render); now env-sentinel
guarded. Findings in .dev/2026-06-15-diffusion-pipeline-reliability.md.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
--alt-work renders each test frame from a second depth dir alongside the
new full-range maps, so the GPU run proves the rev-3 normalization fix
(018909b) beats the washed-out rev-2 maps rather than just looking
better. Persisted the rev-2 maps under tools/out/compare/rev2_depths so
the A/B is reproducible. Doc updated with the one-shot command.
Workstation note: host 192.168.1.26 is up but ComfyUI (:8188) wasn't
running this iteration — needs a manual start before the sweep.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
sweep_strength.py renders one open + one closed FPV frame across a
ControlNet strength matrix at fixed seed/prompt and montages them beside
the depth map — the GPU experiment to find the right strength now that
rev-3 depth maps carry a full-range signal (018909b). Findings/plan in
.dev/2026-06-15-diffusion-pipeline-reliability.md.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
