Go to file
type-two 7c5fc53224 Registry writer: make the four non-file-drop consumers one-click, and keep 3GOD as a depot
Most GODVERSE consumers do not scan a directory — they load from a literal array in source
(djsim/procity PED_NAMES, thriftgod HERO_FLOOR) or a JSON manifest, so a dropped file does
nothing until its NAME is in that list. tools/registry.py makes that edit, and the export hub
now offers it instead of only reporting the file:line.

Paranoid by construction, because the targets are live shipping games:
· dry-run by default; write:true is required to touch source
· idempotent — an existing entry is a no-op, verified against a real djsim ped
· backs up to <file>.bak-<stamp> before every write
· refuses if the anchor matches anything other than exactly once, rather than guessing which
  of several arrays was meant
· handles djsim keeping TWO live copies of PED_NAMES (rigs.js owns the street, index.html has
  a second) — updating one and not the other is a silent half-fix
· works locally and over ssh, since djsim lives on ultra

Bug I introduced and caught while testing: the first implementation inserted after the opening
bracket, i.e. PREPENDED. procity's loadPedFleet fills fixed slots BY PED_NAMES INDEX, so that
silently reassigns every existing character — and my own note on that target said "appending is
safe, reordering is not" three lines above the code that did it. Now scans to the matching
close bracket (string-aware) and appends. Verified: leading entries byte-identical, new name at
the tail, then reverted from backup with git clean.

Also corrected procity's path — its rigs.js is under web/js/citizens/, not crowd/, and declares
`export const`; the two games diverged despite sharing the loadPedFleet shape.

On 3GOD: NOT retired. It looked like a spent planning experiment, but thriftgod fetches its
props from it at runtime in production (index.html:1176 DEPOT = 'https://digalot.fyi/3god') and
procity does the same — 773 live assets. It is two things bolted together: a dumb asset CDN
(/a, /dl, /t, /api/list) which stays, and an authoring UI (upload/tag/thumb/fetch/meta) which
wardrobegod now supersedes. Absorb the bench, keep the depot, publish into it.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-24 21:55:04 +10:00
tools Registry writer: make the four non-file-drop consumers one-click, and keep 3GOD as a depot 2026-07-24 21:55:04 +10:00
web Export hub, unified wardrobe, warm thumbs, body-type heights 2026-07-24 21:00:46 +10:00
.gitignore Phase 5: the dressing room — thumbnails, a manifest, and the rig/LOD/dress bench on one page 2026-07-24 20:25:44 +10:00
blender_ops.py Phase 5: the dressing room — thumbnails, a manifest, and the rig/LOD/dress bench on one page 2026-07-24 20:25:44 +10:00
README.md WARDROBEGOD v1 — wardrobe generator bench (bodies, fits, garment gen pipeline) 2026-07-20 20:53:43 +10:00
server.py Registry writer: make the four non-file-drop consumers one-click, and keep 3GOD as a depot 2026-07-24 21:55:04 +10:00
STATE_AND_PLAN.md Update STATE_AND_PLAN with all phases done + the verified export reality table 2026-07-24 21:02:02 +10:00

WARDROBEGOD — one skeleton, infinite fits

The wardrobe generator bench: load bodies (FBX/GLB/OBJ), build a clothes library, generate new garments on the farm, fit them, assemble dressed NPCs. Sibling of NPCFACTORY (the dress-up/clip bench) and MeshGod imagelab — this one owns clothing.

python3 server.py                 → http://localhost:8150
WG_HOST=0.0.0.0 python3 server.py → expose on the tailnet
MB_TOKEN=mbt_… python3 server.py  → enables the generator tab (MODELBEAST bearer token)

No build step: stdlib server + three.js from CDN. Heavy lifting = local headless Blender (blender_ops.py, tested on 5.1.2; override path with WG_BLENDER).

What it does

  • Library panes: bodies (scans library/bodies/ + ~/Documents/anatomy + thriftgod's web/assets/models), garments, generated images, finished outfits. Upload via file inputs.
  • Stage: orbit viewer; FBX auto-converts to GLB (cached in .glbcache/); HUD reads tris / height / rig state (bones + clips); first clip auto-plays.
  • Body ops: scale-to-height (bases arrive 1.0m tall — normalize to ~1.72m); decimate-keep-weights (the SAFE local path for rigged meshes — never farm /finish).
  • Rigid attach (tier 1): hats/bags/shoes → pick a bone, slide offsets, preview live. Bake permanently via the fleet /rig endpoint (m1ultra :8011) when happy.
  • Deforming FIT (tier 2): weight-transfers the rigged body's skinning onto a garment mesh (Blender data_transfer, nearest-face interpolated, optional inflate-mm of air), parents it to the armature. FIT → wardrobe item = garment+skeleton GLB (reusable); fit → dressed glb = body+garment in one.
  • Assemble: body + ticked fitted garments → one dressed GLB in out/ — drop straight into thriftgod / djsim / procity, or publish to 3GOD.

Generating NEW clothing (the pipeline, $0 on MODELBEAST)

  1. flux (flux_local, Klein) — product-shot prompt template: garment laid flat, white bg
  2. cutout (bg_remove_local) — the biggest quality lever before 3D
  3. then ONE of:
    • → rigid 3D (trellis_mac, ~5 min): hats, shoes, bags, glasses — bone-parent these. NOT for floppy cloth: reconstruction eats thin geometry (fleet traps ledger).
    • → hanging texture: flat garment PNG for rack planes (djsim fittings.js garment({image})) and the paper-doll wardrobe. Flat clothes love this route.
    • deforming wearables: model/retopo against the base body (or socket-cut a clothed torso, character_kit_modular style), then FIT. Never raw-gen these.

House rules

  • Bases (anatomical models) stay in the library. Only dressed outfits ship into games.
  • FIT requires a rigged body. Unrigged base → rig first (MIRPAMO on ultra, or Mixamo).
  • One skeleton (mixamorig) across the fleet: any garment fitted on one body fits every body on that skeleton, and every clip in the bank drives it.