Fable: round-5 update — fal.ai spend authorized (John topped up $22); Lane E runs glass-case/bus-shelter fallbacks

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
m3ultra 2026-07-14 20:38:25 +10:00
parent fd8b0bfcfc
commit 039c179c0e

View File

@ -42,7 +42,10 @@ land-blocker.
3. **OSM plan import uses cached/fixture data ONLY.** Lane A builds the second plan source 3. **OSM plan import uses cached/fixture data ONLY.** Lane A builds the second plan source
against thriftgod's `city_source.json` cache (or a checked-in fixture extract). NO live against thriftgod's `city_source.json` cache (or a checked-in fixture extract). NO live
Overpass/network calls from the repo — the asset-free/zero-network law extends to citygen. Overpass/network calls from the repo — the asset-free/zero-network law extends to citygen.
4. **fal.ai spend (glass-case/bus-shelter) stays parked; pushing stays John's.** Audio stays 4. **fal.ai spend is AUTHORIZED (John, 2026-07-14: account topped up to $22).** Lane E runs
the fallback gens — glass-case + bus-shelter, plus one optional magazine-rack attempt —
at Hunyuan v2 ~1417¢/gen, ≤2 attempts each (expected total well under $1). Nothing else
on fal without a per-item reason logged in AUDIT.md. Pushing stays John's. Audio stays
greenfield-parked (needs an asset-sourcing decision from John). greenfield-parked (needs an asset-sourcing decision from John).
5. **v1.1 tag**: Lane F tags after C1 (re-map) and E1 (bake) land and the tour still looks 5. **v1.1 tag**: Lane F tags after C1 (re-map) and E1 (bake) land and the tour still looks
right. v2 flags do NOT need to be done for v1.1. right. v2 flags do NOT need to be done for v1.1.
@ -196,11 +199,19 @@ qa green.
(metres, +Y up, base origin, ≤5k where they're not focal), thumbs, manifest, publish, (metres, +Y up, base origin, ≤5k where they're not focal), thumbs, manifest, publish,
AUDIT provenance. Do NOT add manifest entries no lane consumes — check the other lanes' AUDIT provenance. Do NOT add manifest entries no lane consumes — check the other lanes'
round-5 notes or ask in yours. round-5 notes or ask in yours.
3. **Not doing** (still parked): fal.ai spend, MESHGOD_BATCH full run, audio assets. 3. **fal.ai fallback run (NOW AUTHORIZED — decision #4).** Generate `glass-case` and
`bus-shelter` via fal Hunyuan v2 (the two TRELLIS glass rejects), plus ONE optional
magazine-rack attempt if the first two go well. ≤2 attempts per prop; expected total
<$1 of the $22 balance. Then the normal funnel: normalize (bake-to-lowpoly if heavy per
task 1) → eyeball thumbnails (no melted glass/legible branding) → publish → manifest.
Log every paid gen + cost in AUDIT.md provenance. If a prop fails both attempts, it keeps
its primitive — do not burn extra attempts.
4. **Not doing** (still parked): MESHGOD_BATCH re-runs beyond the above, audio assets.
**Acceptance**: 6 hero props ≤8k tris with baked textures live on the depot, visually clean, **Acceptance**: 6 hero props ≤8k tris with baked textures live on the depot, visually clean,
C's interior re-measure confirms the drop; new shortlist normalized + published with C's interior re-measure confirms the drop; glass-case/bus-shelter resolved (shipped or
consumers identified; validate --depot 0 errors; qa green. documented-failed at ≤2 attempts) with costs logged; new shortlist normalized + published
with consumers identified; validate --depot 0 errors; qa green.
## Lane F (Integration) — v1.1 tag, v2 seams, gates that grow ## Lane F (Integration) — v1.1 tag, v2 seams, gates that grow