Add reviewed MakeHuman head prototype (vision gate rejected)

This commit is contained in:
zwitschi committed 2026-11-04 17:04:20 +01:00
1 parent d1898b3221
commit 958d3ae0ff
13 files changed
+197 -14

No files matched your search

+11 -2
View File
@@ -80,8 +80,6 @@ the rest of §C remains the broad review checklist.
### 3. Head and face prototype
### 3. Head and face prototype
- [x] Build one `npc_casual` head/face prototype as a connected head shell plus intentional feature
geometry — not a sphere with floating facial balls. Use loop-based eye sockets/lids, a
readable nose bridge/tip/ala, a mouth with upper/lower lip volumes, ears with concha depth,
@@ -95,6 +93,17 @@ the rest of §C remains the broad review checklist.
- [x] Compare two or three subdivision/loop variants at 1.85 m front/three-quarter/profile and a
1.5 m portrait. Pick the variant that reads best before changing all 12 characters; keep the
rejected variants local.
- [-] Rebuild a standalone MakeHuman-derived head from the researched CC0 base and save it to
`assets/models/blender_src/makehuman_head.glb`; record provenance/review in
`docs/makehuman-head-review.md` and the preview sheet. (Final build: 12,641 tris / 7,690
split verts; assembled `npc_casual` 15,532 tris / 13,310 verts / 14 bones, `inspect` clean.
Latest attempt: skin eyelid mask, recessed globes, flat iris/pupil discs, surface-anchored
brows. Vision verdict: still **REJECT** — globe depth is inconsistent between front,
three-quarter and profile, the lid does not visibly clip the globe, and the iris reads
decal-locked at three-quarter; prototype stays local and un-integrated.)
- [ ] Build true upper/lower eyelid masses that overlap the globe in front, three-quarter and
profile views (not a socket hole), then re-run the vision gate before integrating the head.
Compared `head_proto` at seg 24 (670 verts), 48 (1,558 verts) and 64 (2,470 verts) at 1.85 m
and 1.5 m, front/three-quarter/profile. Vision-judged: **s48 wins** — s24 is too faceted
(nose/lips melt, jagged silhouette), s64 is visually identical to s48 at both distances