REGISTRY entry + the registered 46.7 MB GLB (add -f per the working-files rule:
registered v02 artifacts are tracked even though the v0N dirs are ignored).
sha256 verified against the registry line. Recipe 07_fill_crotch.py already
tracked; known consequences (hip shelf bulges, flat donor texel) documented in
the registry entry, not defects to chase.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The exp05 verdict ("weights alone cannot clear the bar on this mesh") was
measured against a hard partition that was itself causing much of the tearing.
exp01-exp05 gave every vert to exactly ONE digit -- 08_finger_weights.py had an
explicit `elif lo in FING: continue # other digit: hard wall` -- which
guarantees the fused inter-digit bridges tear by the full finger separation,
because a single edge ring absorbs the whole gap.
Replace that with graded blending (WEB_BLEND_R0): a vert's fraction toward its
nearest other digit ramps 0 -> 0.5 as r = d_own/(d_own+d_other) goes 0.35 -> 0.5,
so both sides of the equidistance valley reach 50/50 and the field is continuous
across the boundary. Applied after the smoother (whose hard wall would erode beta
exactly where it must survive) and evaluated in the neighbour digit's own
arc-length frame. WEB_BLEND_SKIP excludes the thumb: its transform is opposition,
not curl, so its frame does not correspond to a finger's.
Measured against an identical baseline (same input, blend the only variable),
real tears (>=1mm rest length) drop 23-60% with NO regression on flat, the pose
that ships: fist_r 1178 -> 530, grip_r 561 -> 222, fist_l 979 -> 633,
grip_l 420 -> 322; total >5x 2487 -> 1307; p99.9 better on every pose;
flat unchanged at 0/1. fin_bones confirms the mechanism rather than just the
count -- the middle<->ring and pinky<->ring families leave the top classes while
the thumb/palm ones are untouched to the edge (182 -> 182, 148 -> 148).
This does NOT make fist/grip shippable: 222-633 real tears still reads as a
destroyed hand in clay renders, and the residual is now ~53% thumb-pad-fused-to
-palm, which is topology and needs mesh surgery or the v02 rebake. Flat and
relaxed are the shippable poses; fist/grip belong to the morph lane for now.
Also here:
- README: the solver's input is v02/..._exp03.glb, NOT exp01. exp01 is pre-hand
-fit (converter steps 2b/2c); its finger groups sit on the wrist and overlap
the real finger by 1.6cm, so a solve from it silently zeroes every _02/_03 bone
-- rigid stick fingers and a torn flat -- while weight sums stay 1.0 and every
assert passes. Cost three wasted bakes and one false "the solver regressed".
- Seed assert demanded >=100 seeds while the radius loop caps at
SEED_AXIS_R_MAX, which left pinky (88 seeds at 16mm) can never satisfy; the two
constants were mutually unsatisfiable. Now >=60, and it is documented as a
sanity gate rather than a quality bar.
- Detwist poses tested at last: real but marginal (fist_r 38.7x -> 28.7x,
grip_r 30.0x -> 17.3x, left hand flat). A knob, not a fix.
- edge_stretch_cmp.py / skin_bone_territory.py / handpose_trim_hand_obj.py:
judge tears by rest length and absolute posed growth, not raw ratio; audit
whether a bone owns any verts at all (thumb_01 owns ZERO in exp05); and trim
an arm-sized skin dump to the hand before rendering.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
`07_fill_crotch.py` and its README autopsy were written in an earlier session
and had been sitting uncommitted. Committing them unchanged, separately from
today's hand-weight work, so neither is misattributed to the other.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The morph lane's blocking bug was believed to be coincident duplicate verts hidden
from the solver's convergence gate. Welding was necessary but nowhere near
sufficient: the previous relaxation could satisfy every stretch bar without posing
anything, so the numbers it reported were not measuring the thing they claimed.
Measured on Ariki_Female_QuatSkin_LowPoly_40, the tracked "left hand is fully
clean" baseline was a morph that moved fingertips 0.3cm, and the right hand was a
rigid 6.5cm translation of the whole hand with its shape intact. Gated Laplacian
diffusion of a delta field has a null space — constants — and edge stretch cannot
see any of it: a rigid translation stretches no edge and neither does a collapse to
zero. Both exits report perfect bars.
Six defects fixed, each with its measurement in the code comments:
weld_roi 356 duplicate groups solved twice, deltas up to 2.55cm apart
stitch_components hand is 6 overlapping sheets with 1-10mm gaps; Dijkstra
cannot cross one, so each bone saw only its own sheet
refit_fingers skeleton finger chain ran to 19.8cm; the flesh ends at 14.0cm,
so curl_02/curl_03 drove almost no weight (shipped skeleton
untouched — this is solver-local scaffolding)
solve_weights chain-arc partition of unity; the old 8mm isotropic kernels
left ZERO of 17,480 verts owned above 0.85, and a 3-way blend
averages the curl away. Plus a support cutoff: outside every
kernel, renormalized 1e-81 noise had handed a mid-palm vertex
index_02_r=0.50 and flung it 29cm
build_roi joint-sphere ROI, so the rim is a wrist band and not a fractal
of interior chart holes
relax strain-only edge projection + ROI-border seam constraints,
replacing the diffusion described above
All 8 shapes now pass every bar with real deformation: p99.9 <= 1.58x, zero edges
over 5x, zero needles with the sliver floor removed, seam <= 3.2mm, cross-hand
independence exactly 0.0000cm, mesh fingertip travel 4.6-6.2cm on fist and
3.5-4.7cm on grip (both hands). Gates now report mesh travel and seam alongside
stretch, because stretch alone cannot gate this lane.
Verified in anim_hand_test_bed: fist and grip read as a real curl on both hands,
static and mid-dance, no fins/shards/needles. It is a loose fist rather than a
clenched one — her fingers are ~4-5cm past the knuckles.
Demo GLB stays out of git (122.6 MB, ariki-game/scratchpad/). Ship decision, the
thumb-axis refit, and the TDR crash from 8 dense targets are open — see README.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Two-finger mako rig authoring source: v5 working file plus a cleaned
variant, ahead of the TwoFinger body export.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Lena with the leaf bikini sculpted into the mesh — a separate character
descending from her own Tripo original, not a lena_base_v02. Full-res
(1,029,360 v) at 1.777 m, now Ariki_Female_QuatSkin.glb + LOD1.
First nearest-surface graft in the rig lane: AccuRig had to be fed a 20:1
decimated bait, so the index-exact graft the two earlier ships used was
impossible. The bait recovers AccuRig's 9.346 mm offset in closed form, which
makes it a required input forever — copied into the ship folder, and rig-work's
"disposable" rule amended to say so. Mesh moved 0 mm, 0 unweighted verts.
lena_base_v01 goes superseded, not rolled-back: she lost the default slot but
Ariki_Female_QuatSkin_Nude.glb is untouched, so per rule 9 her folder stays put.
Also lands the leaf-cut lane (work/lena_leafbikini): hue-keyed seam detection
that removes the leaf shell and leaves both holes open by request, registered
as a lane milestone with its sha256. Corrects the rig-work trap note — baits do
carry embedded textures; it is AccuRig that strips them.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Marvelous Designer authoring sources for lena_hunter_skirt_v8 — the .zprj/.zpac are
hand-draped and no script regenerates them, so they are KEEP tier, not scratch — plus
the exported garment (fbx/obj/mtl), the colorway and cloth textures, and the QC
screenshots (front/back/presim) the drape was judged on.
tools/tailor/avatar/Mako_Fullhead_QuatSkin_Avatar.fbx is the MD avatar built from the
shipped male body. Worth landing even though it does not work yet: Mako's FBX imports
into MD but yields 0 arrangement points against Lena's 109, which blocks every male
garment. Having the exact file in-repo is the starting point for diagnosing that.
Mako lane intermediates (decimated fullhead, AccuRig glb, rigged blend) join their
recipes under work/mako.
All heavy files ride LFS (.glb/.blend/.fbx/.obj/.zprj/.zpac already in .gitattributes).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The v02 chain, which exists because bisecting the shard-eye defect proved it lives in
the hires MASTER rather than in any v02 step: the early heal chain ran whole-mesh welds
and reset custom split normals, and the eye/lash shells only read as an eye through
Tripo's authored normals. Every descendant inherits it, including the shipped body.
48_decimate_headsafe body/hands only, head bit-identical (feathered ramp)
49_seams_v02 24_seams with a density-aware hip landmark — the v01 rule fired
at u=0.610, mid-belly, on the head-protected vert count
50_seams_headuv body-only unwrap; the head KEEPS its original Tripo charts.
SLIM collapsed the undecimated lash/brow slivers to points and
ANGLE_BASED packed at half v01's texel density; the face was
the best-mapped region of the source atlas, so it is reused
51_reatlas_v02 31_reatlas + centroid splats for sub-texel triangles — the
skipped set IS the lashes, which rendered as grey glass
52_head_transplant the PRISTINE ORIGINAL head onto the decimated nude body
53_rig_transfer 54_crotch_refill
55_fullres_v02.blend is pinned in .lanekeep as THE archival master: full-res nude body
+ pristine original head, crotch refill and texture despeckle applied, 883,404 v /
1,761,640 f. It supersedes 34_v04 as the lane root (34_v04's head has the shard eyes).
Also: tools/graft_hands.py, the Marvelous Designer hunter-skirt configs v1-v8, the
hunter cloth texture generator, and Mako's measurement card.
Per .agents/rules/working-files.md the per-attempt .blend files under work/lena/v02
are SCRATCH ("never committed") — the .py recipes here are the history and regenerate
any of them from the pinned master. See the ignore rule landing next.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
jeremy