security-game/TODO.md

TODO

Active task list. Work is grouped by milestone and workstream; every item is small enough to complete and verify on its own. The product plan, gap analysis and release roadmap live in PLAN.md; durable decisions and gotchas live in MEMORY.md; AGENTS.md defines the workflow.

Current state (2026-11-05): v0.9.8 is published. Three venues, the shift loop, the tutorial/campaign, the access/escalation system and the 12-character procedural library are shipped. The character plan changed twice and landed on the CC0 Quaternius base; the owner accepted the 62-joint runtime rig (ADR-0018), but the base does not yet render or animate correctly in Godot.

Roadmap (detail in PLAN.md):

Version	Type	Scope
0.9.9	patch	Quaternius pilot base: rig/export fix, pose composition, face/silhouette gate
0.9.10	patch	12-role roster on the approved base, gear fit, budget evidence
0.9.11	patch	Scheduled venue, prop, foliage, lighting and set-dressing polish
0.9.12	patch	Performance/Web/accessibility verification before the minor
0.10.0	minor	Public-readiness release: playtests, balance, legal, platform matrix, docs
0.10.1–0.10.3	patch	Onboarding, balance and stability follow-ups
0.11.0	minor	Live body-camera replay, fourth venue, localization foundation
0.12.0	minor	Signed/storefront builds, frozen save contract, launch languages
0.13.0	minor	Content complete and feature freeze for 1.0.0
0.13.1–0.13.2	patch	Release candidates
1.0.0	major	Stable release with a documented compatibility promise

Priority now: the 0.9.9 rig-and-export correction. Do not start the 12-role roster (0.9.10) before the single-variant pilot renders, animates and passes the owner face gate.

The unversioned tracks (3D review evidence, maintenance, Gaussian splatting) stay at the top of this file because their work is not release-scoped; each milestone below is versioned and must follow the release procedure in AGENTS.md.

The MIT Blender skill pack lives locally at .agents/skills/blender-skills/ (gitignored). Every 3D workstream below names the skills to load; they are guidance only, and the project contract, blender_assets.py audit/fit/inspect and the capture gates stay authoritative.

3D model and local map re-evaluation

Goal: re-judge every shipped 3D asset against the newly available Blender skill guidance, rank the real gaps, and fix only what the rankings justify (sections A–G); and re-evaluate the three venues against licensed real-world data — OSM footprints and building massing, reference-derived photoreal facades and the photo-driven geometry pipeline (sections H–M). Evidence comes from three separate sources: licensed real-location photos, generated concept boards, and build captures. This is an unversioned development track: it must not change the npc_base contract, the renderer or the public 0.10.0 release.

Sequencing (updated 2026-11-05): the old block is cleared. The local map rebuild (H–M) is done, the 96/96 review ran against the new maps, and character work moved to the versioned 0.9.9/0.9.10 milestones. This section is now the evidence, scorecard and measurement source, not an open-ended backlog: the venue/prop/lighting fixes are scheduled into 0.9.11, the budget/LOD checks into 0.9.12, and the character items (C) are superseded by 0.9.9/0.9.10. Fixes are made from the milestone checklists; the scorecard keeps the measurements and verdicts.

Map rebuild status (2026-11-03): H–M are complete — the Joke Shop is now a 6.0 m low hall (dark stucco, white signature fascia, tied-back curtains, hotel tower) and Hollywood/TV Studio share the limestone-ashlar Masonic facade. A fresh capture_npc_review.gd run produced 96/96 frames against the new maps (none blank) and a vision spot-check confirms the characters still read cleanly. The remaining §L signage-texture candidates and the deferred 3D facade-reveal depth are enhancements, not rebuild blockers.

Reference material is collected first (section A): local, gitignored, development-only photo research may use any source the owner authorises (DuckDuckGo image search, street-view frames, neighbouring properties, buildings across the street) without the licence gate. The licence and privacy gates still apply the moment a frame is promoted into docs/screens/refs/ or any committed asset — local tmp/research/ is the only licence-free zone.

Scope: the shipped 3D catalogue — 8 library characters, the frozen npc_base fallback, 18 gear pieces, 20 library props, 6 generated props and the three venue families — plus the local map pipeline that re-evaluates those venues (OSM data/licensing, reference imagery, the photoreal facade kit, AI-assisted composing, Gaussian-splatting evaluation and map QA).

Exit (asset review, A–G): docs/assets/3d-model-review.md scores every catalogue asset; every P0/P1 is fixed or explicitly accepted with a measurement; inspect/audit/fit/proportions, the 24-NPC budget and the capture gates are green; no runtime contract, dependency or release artifact changed except the rebuilt assets themselves.

Exit (local map, H–M): one local block is reconstructed from licensed data with reference-derived photoreal facades, passes performance/navigation/visual QA and has complete provenance; or the pipeline is rejected with the findings recorded.

A. Local reference material collection (development-only, gitignored)

Owner override (2026-11-03): for local, development-only photo research the licence gate does not apply. Anything downloaded for this track stays under tmp/research/ (gitignored) and is never committed, shipped, or used as a committed asset. The moment a frame is promoted to docs/screens/refs/<venue>/ it must pass the sidecar licence + privacy gate; local research is the licence-free zone only.

• Add a DuckDuckGo image-search fetcher: tools/refcollect.py search <query> --slug <block> --limit N downloads top results into tmp/research/<block>/ with a source-URL manifest (no licence filter, marked local_only/unchecked-local), and sheet composes a local contact sheet. Requests go through a token bucket plus polite jittered backoff that honours Retry-After, and send a full Firefox header set. Tests in tools/test_refcollect.py. Run note: DDG rate-limits aggressively; space batches and run the per-venue collection below when the limit clears.
• Collect Hollywood block reference: facade, marquee/blade-sign neighbour, Walk of Fame and arrival staging are covered by tmp/research/elcapitan/, hollywood-block/ and the jimmy-kimmel-* sets (day, night, entrance and stage frames).
• Collect Joke Shop / Comedy Store block reference: tmp/research/comedy-store/ covers the facade, canopy, street and visitor views; ramada-hotel/ adds the neighbouring hotel and comedy-store-hillside/ the hillside/houses behind the roofline.
• [-] Collect street-view frames of both blocks: street-level DDG frames now cover the facades, corners and opposite frontages for both blocks; true Google Street View panorama capture is deferred (needs a Google API key or a manual browser capture — recorded in MEMORY).
• Build a local contact sheet per block: sheet.jpg exists for every collected set (elcapitan, hollywood-block, jimmy-kimmel-stage, jimmy-kimmel-entrance, comedy-store).
• Record which local frames changed a decision: added the frame→decision mapping to docs/design/venues.md#hollywood-boulevard (recent elcapitan-2026 geosearch → 2009 frame stays the geometry reference; jimmy-kimmel-entrance/07.jpg → the four entrance concepts) and docs/design/venues.md#the-joke-shop (comedy-store/ → the four concept pairs + map plan; comedy-store-hillside/ → the day-concept anchor that clears the image-recitation filter). Frames stay gitignored; only the provenance notes are committed.

B. Baseline, scorecard and routing

• Load blender-director plus the per-family specialists before touching geometry, and record the skill stack and the advice that changed a decision in docs/assets/3d-model-review.md. (Routing, QA severity/verdict, texel-density tiers, character loop criteria and budget rules recorded in the review doc; MCP-only steps translated to headless flatpak runs.)
• Re-run the neutral NPC baseline (48 frames), the final-context review (96 frames), the prop gallery and the 24-NPC budget on the current build; keep raw frames in tmp/ and do not replace committed sheets until a fix lands. (48/48 and 96/96 frames captured under xvfb-run, none blank; raw sets in tmp/model-baseline and tmp/model-review; committed sheets untouched. 24-NPC smoke still 6.94 ms / 168 meshes / 131,616 triangles.)
• Create docs/assets/3d-model-review.md with a per-asset scorecard: silhouette, anatomy/proportion, topology/edge flow, UV/texel density, textures/materials, rig/weights, LOD/impostor, collision/grounding, export contract, provenance, skill route, verdict (P0/P1/P2/accept). (Doc created with character/gear/prop/venue tables, verdict columns and a ranked candidate list; unmeasured cells are marked — instead of being guessed.)
• Collect the baseline measurements in one table: blender_assets.py inspect, audit, fit, npc_proportions.py, texel density, triangles/meshes, 24-NPC frame time. (All seven captured; texels and inspect --json --out were added to make the measurements repeatable. All 46 assets audit/inspect clean; texel density 64–122 px/m authored, 287–440 effective fabric.)
• Review the scorecard with the project owner, rank the P0/P1 gaps and decide fix vs accept per item; record the decision and the reason in the review document. Deferred until the full map rebuild (H–M) lands and a fresh 3D review runs against the new maps.

C. Character re-evaluation (superseded 2026-11-05)

The procedural character review is complete and its evidence stays in docs/assets/3d-model-review.md. The Quaternius base shipped in 0.9.9; the remaining roster follow-ups are Milestone 0.9.10. The open items below are kept as evidence for what the new base must not regress (face read, hands, weights, cloth, texel density, crowd variety).

Skill route: character-artist, retopology, rigging, uv-workflow, texture-workflow, materials, hair-groom, asset-optimization, lookdev, qa-review.

• Face/head review: the eight library characters ARE built with the 0.9.4 face detail in tools/blender_assets.py._face() — sculpted chin plate + cheekbones (_shape_head_mesh), socketed sclera + thin upper lid, nose bridge/tip/wings, rounded mouth and ear concha. The plain sphere head in tools/npc_pipeline.py is the frozen npc_base fallback only. The ranked P1 "flat faceted cheek/brow + slit eyes" is therefore a readability/scale question on the existing sculpt (head seg=18/rings=12, small offsets), not missing geometry. Next step if the owner funds it: a single-variant npc_casual prototype that raises head subdivision and deepens the cheek/brow offsets. Fix is gated on the owner scorecard review.
• Hands/grip review: the hands ARE detailed — tools/blender_assets.py._hands_and_shoes() builds a wider shallower palm, four two-segment bent fingers with rounded tips and a thumb; gear contact is the open palm plus the HAND_ITEM_FORWARD (0.04 m) offset. Decision: keep the open palm + forward offset; a finger-curl rig stays out of scope — the current rig has no finger bones, so a curl rig is a contract decision, not a modelling tidy-up. Optional extra finger bones are now allowed by the cap/contract extension, but require an ADR and weight/budget verification. Recorded as ranked gap #6 (accept; revisit only if face work opens the budget).
• Weights/pose evidence captured and vision-reviewed, and the P1s fixed: wrist/hand detachment (wrist + hand now 100% Hand, hand raised, sleeve cuff follows the hand), hip/thigh collapse + knee pinch (thigh/shin rigid with a short knee blend). Vision-verified: hip fix clean, knee crease minor residual. The guard hip-pouch clip and the residual knee crease remain as 0.10.0 refinements.
• UV/texel density: measured (64–122 px/m authored, 287–440 px/m effective fabric on the 4× repeat; table in docs/assets/3d-model-review.md). Decision: accept as-is — the maps are neutral/runtime-tinted (not painted skin), so density controls only subtle weave/noise, and fabric already sits in the standard tier. No atlas/UDIM/resolution change is justified until a portrait shows texture starvation; CHAR_MAX_TRIS/CHAR_MAX_VERTS stay unchanged.
• Cloth and hair: hem clipping is already handled by the 0.76 stride scale for npc_dress/ npc_coat (v0.9.0), hair/beard are the five procedural styles, and headgear clearance is verified clean by blender_assets.py fit (0.9.4). Decision: cloth simulation stays out of scope for the 8-variant library and the Compatibility renderer — rigid procedural meshes plus stride scaling are the accepted approach; a baked cloth-sim pipeline would only be reconsidered if a later close-up shows visible hem/coat stiffness.
• Fix only the ranked P0/P1s, one asset family per commit; refresh the comparison sheet and re-run inspect, audit, fit and npc_proportions.py after each fix. Scheduled: the remaining character items are covered by Milestone 0.9.10; this row is evidence only.
• Crowd-variety reference: five licence-cleared Commons photos fetched to tmp/research/crowd-variety/ (CC0 ×2, CC BY 2.0, CC BY 3.0, PD) covering umbrella/dense crowds and a pedestrian mall — general build/hair/outer-layer variety context, reference only (never traced). The AI proposal set (generated crowd-variety concepts) will carry the specific 4-variant detail when the character-development roster starts.
• Generate a face/hand concept board for the highest-ranked P1 and compare it with the 0.9.4 baseline before rebuilding anything; keep it development-only and record model and prompt.
• After each accepted fix, capture front/three-quarter/profile comparisons, add them to the review sheet, and refresh the README character row.
• qa-review sign-off: all eight variants ship with notes (face realism landed, hands/ cloth/gear accepted, props reviewed); the weights/pose column is the only open item (blocked on the harness) and the full foliage carded-cluster is the one remaining owner-gated P1. Recorded in docs/assets/3d-model-review.md §qa-review sign-off.

D. Gear and prop re-evaluation

Skill route: prop-artist, hard-surface, vegetation-artist, materials, collision-proxy, lod-pipeline, qa-review.

• Gear: review the 18 pieces for silhouette, straps/buckles, helmet and lanyard readability and rebuild only what fails the scorecard. (Attach/clearance verification is done — the fit report is recorded in docs/assets/3d-model-review.md with the back-gear-vs-hip-radius measurement caveat; the visual silhouette read is what remains, pending the owner scorecard.)
• Props: review the 20 library props plus the 6 generated props for bevels, chamfers, silhouette, ground contact and shared material language (prop-artist, hard-surface); list the rebuilds that are actually worth the triangle budget. (Vision-reviewed 2026-11-03 via tools/describe_image.py: fire_hydrant rebuilt and shipped; planter_box/ round_planter foliage and street_tree/palm_tree are the foliage-language P1 (gated on the owner); newspaper_box slot and bollard collar are minor rebuild candidates; bench/ trash_bin are ship-ready.)
• Foliage: replace the flat-shaded _shrub()/_tree() clumps with vegetation-artist-style carded clusters; measure triangles and keep the 60k LIBRARY_TRIANGLE_BUDGET. (Partial 2026-11-04: _shrub() clumps and the palm fronds now alternate two green tones (foliage/foliage_dark, palm_leaf/palm_leaf_dark) so the crown is no longer one flat colour, and a textured_solid bug that broke planter_box/round_planter imports was fixed. The two-tone read is subtle at gallery distance; the full carded-cluster pass remains the owner-gated P1.)
• Materials and emissives: lookdev pass on bulbs, signs, lanterns, the marquee and the sedan at day and late; confirm the flashlight stays a tight spot with no beam geometry.
• Collision: verify every prop collider and navigation obstacle matches its silhouette and add the missing collision-proxy shapes where walking is blocked.
• Build a real-photo comparison for street furniture (bollard, bench, bin, hydrant, planter, lamp) and rank the silhouette and proportion mismatches.
• Rebuild only the mismatched pieces through the assisted Blender pipeline (blender_assets.py inspect → scripted retopo/UV assist → atlas → audit/fit); document the exact commands in docs/assets/asset-tooling.md.
• Refresh sheet_props.jpg/sheet_npc_gear.jpg, the README asset rows and docs/blender_previews/README.md after the rebuilds.
• Exit: a prop/gear capture sheet, budgets and collision checks pass with no unreviewed asset.

E. Venue and camera re-evaluation

Skill route: environment-artist, archviz, set-dressing, lighting, camera-cinematography, lookdev, rendering, qa-review.

• [-] Facade/venue review: rank the top five shape-language fixes for Hollywood, the Joke Shop and the TV Studio (environment-artist, archviz) and check them against the map-budget tests. Ranked 2026-11-03: (1) TV Studio renders a generic pilaster/concrete box instead of the six-column Masonic Temple — tv_studio lacks masonic_facade, so _build_building() runs while the Jimmy Kimmel starting point is the same facade; (2) Joke Shop blade sign is a flat slab + label, not the reference's circular sign on a roof pole; (3) Joke Shop canopy fascia lacks the deep white-signature band; (4) Hollywood entrance signage shows EL/CAPITAN/LIVE banners where the real building carries the Jimmy Kimmel Live marquee (owner content call); (5) Joke Shop red-curtain bays and entrance-recess rhythm read flat under the canopy. Fix 1 landed: tv_studio now reuses masonic_facade (architecture masonic, ashlar field, six columns, eight steps, landing + guard spawn at y=1.0) while keeping the arrival staging (bulb_canopy, sedan, handrail, barriers); Godot 276/276, screenshot smoke 21/0, 13-studio-arrival.png recaptured. Fix 3 landed: the Joke Shop canopy signature band is now white (Color(0.90, 0.88, 0.80)) over the black fascia, matching the concept's "white celebrity signatures"; a test asserts the CanopySignatures label is white, Godot 276/276, 12-joke-shop.png + the three-up board recaptured. Note: fix 2 (blade sign) is already a circular disc on a roof pole in the code (37f925c); the ranking note was stale. Fix 5 landed: the red curtain bays are now seven tied-back drapes (hourglass cinch, dark recess between bays, scalloped hem) instead of four flat red slabs; vision-judged as "tied-back/gathered fabric" via tools/describe_image.py. Joke Shop map is now 25,828 tris / 518 meshes (under the 150k/700 budget); Godot 277/277, 12-joke-shop.png + board recaptured.
• Lighting: lighting/lookdev pass on sconces, canopy, marquee and street lamps with day/late captures. (Attenuation is now gated: test_venue_lights_use_physical_falloff asserts every OmniLight3D/SpotLight3D in all three venues uses attenuation = 1.0, so a constant-falloff light cannot return unnoticed.) Vision review 2026-11-03 (tools/describe_image.py on the night + flashlight captures): the flashlight (light_energy 4.2, spot_angle 14°) washes the facade flat when pointed at the door from close range and backlights the nearest NPC into a silhouette (staged-shot artifact, but worth a shorter/softer cone or lower energy for the 06-night-flashlight stage); the El Capitan Theatre red marquee reads as a flat saturated block and the event banners read as flat emissive decals rather than glowing signs — both are Compatibility renderer limits (no bloom) plus the marquee's near-pure-red body; street lamps pool weakly on near-black ground. No code change yet; these are owner-scorecard inputs.
• Set dressing: review entrance and sidewalk prop clusters (set-dressing) for occlusion, navigation obstacles and readable storytelling clusters. Vision review 2026-11-03 (tools/describe_image.py on the Joke Shop capture): the facade is well dressed, the sidewalk under-dressed. Findings: (1) the queue story is weak — the security_podium sits at x=6.8 while the entrance is at x=-8.8, so the host/podium and the two stanchion/crowd_barrier pieces do not read as a line at the door; (2) the rope barrier reads as thin bare geometry at night; (3) the west palm (x=-13.0) is inside the 27 m facade x-range so its fronds visually overlap the roofline from the beauty camera; (4) the sidewalk has no light pool or A-frame/poster cluster. These are P2 composition items, kept for the owner scorecard rather than moved unilaterally.
• Camera height: already decided in the 0.9.0 pass — keep the 1.7 m gameplay camera; the ~1.55 m concept frames are art reference only (recorded in DONE.md 0.9.0 Animation).
• Build the venue three-up boards (real photo | generated concept | current capture): tools/three_up.py compose renders labelled strips into docs/screens/boards/ (caption sidecar per board, rows emits README rows). Boards committed: El Capitan day, El Capitan night, Joke Shop day. Tests in tools/test_three_up.py.
• Link each committed board from docs/design/venues.md#hollywood-boulevard and docs/design/venues.md#the-joke-shop.
• Use the generated concept lighting and palette as the target for the lighting/lookdev pass, and record the deviations that stay accepted.
• After the pass, regenerate the venue captures and update README plus both venue documents.
• Hooded/winter portrait: capture_realism.gd now force-spawns the requested --portrait-type when the role is event-driven (types 5–7 do not appear in the routine day crowd), so the deferred close-up can be staged. Captured --portrait-type=5 and vision-reviewed: the hood reads as a hoodie (drawstrings + face opening) but is snug — no slack volume behind the head/drape at the neck — and the face stays fully lit with no inner-rim shadow. Acceptable; the snug hood is a minor scorecard note, not a P0/P1.

F. Reference images, concept boards and screenshots

Skill route: lookdev, lighting, environment-artist, qa-review; tooling: tools/scene_concepts.py, tools/screen_designs.py, tools/asset_pipeline.py.

Licensed photos are evidence of what a place looks like, generated boards are art-direction proposals, and captures are proof that the build matches. Keep the three labelled and separate, and never ship a generated or downloaded image without provenance.

• Create docs/screens/refs/ with a README and a sidecar template: source URL, author, licence/attribution, capture or generation date, what it is evidence for, and whether it is shipped or development-only. (docs/screens/refs/README.md defines the layout, naming, licence and privacy gates and the three-way separation from concepts and captures; SIDECAR-TEMPLATE.md carries the photo table, redaction/permission notes, derived-decision log and the generated-board reference table. Linked from docs/screens/README.md.)
• Hollywood board: five licence-cleared photos (facade day, blade sign, 1922 elevation, Ovation across street, premiere staging) with a sidecar and per-image purpose, linked from docs/design/venues.md#hollywood-boulevard and credited in CREDITS.md. Cropped the foreground crowd out of the facade shot; EXIF stripped. (docs/screens/refs/hollywood/.)
• Joke Shop board: three licence-cleared photos (hillside/canopy block, frontage with palms, canopy signature band) with a sidecar, linked from docs/design/venues.md#the-joke-shop and credited in CREDITS.md. The CC BY-SA The Comedy Store.jpg from the old link list was deliberately not used; the public-domain frames cover the same evidence. (docs/screens/refs/joke-shop/.) Removed 2026-11-03: the committed Comedy Store photo board was replaced by the local-only tmp/research/comedy-store/ research set (owner override; see §3D model and local map re-evaluation → A).
• TV Studio: no separate board — the venue is the same El Capitan Entertainment Centre block as the Hollywood day shift, restaged for a late-night arrival. The Hollywood board covers both venues and the premiere-staging photo is the arrival evidence; the reuse is now stated in docs/design/venues.md#hollywood-boulevard, docs/design/gameplay.md, PLAN.md and docs/project/development-status.md.
• Fill the Hollywood gaps that have a clean licence: El Capitan Theatre neighbour and marquee, Walk of Fame terrazzo band, star close-up, and a night street-lighting context frame (cropped to remove people and a plate). The board now has nine photos. (CC BY-SA night frame is redistributed unmodified for study only; the review is recorded in the sidecar.)
• Generate one concept board per venue with tools/scene_concepts.py generate (palette, facade rhythm, lighting), keep raw output gitignored, commit only the selected board, and record the model, prompt and spend in the sidecar.
• Build a three-up sheet per venue (real photo | generated concept | current build capture) and use it as the reference for the character, prop and venue verdicts. (Done for the venue exteriors via docs/screens/boards/; character/prop verdicts still use the Blender preview sheets.)
• Apply the screenshot rule to every workstream: each accepted fix updates the relevant row in README.md plus the matching docs page (docs/blender_previews/README.md, docs/screenshots/README.md, docs/assets/3d-model-review.md or the venue map doc).
• Re-run tools/check_screenshots.sh and python3 tools/check_docs.py after gallery changes and record both results in the handoff.
• Keep AI-generated images out of assets/ until they are picked, cropped, credited and (for textures) audited; CREDITS.md and ADR-0015 still govern provenance.

G. Gates and documentation

• Add the measurable checks the review needs (texel density, weight/pose sanity, collider coverage, LOD/impostor check) as stdlib or GDScript gates where possible; CI stays offline. (Texel-density gate in tools/test_blender_assets.py checks all five slots per character and the 40–400 px/m authored band plus the 250 px/m fabric floor; the pose→bone contract is covered by test_pose_reaches_the_skeleton_bones; LOD/impostor already has test_avatar_mesh_budget_and_lod_swap. Collider coverage is still open — split below.)
• Add a collider/proxy check: every library prop that blocks walking has a StaticBody3D + NavigationObstacle3D, checked against the catalogue. (world_builder.gd tags every library node with a library_prop meta and declares the non-blocking marquee/handrail/lantern/door set; two tests in test_world.gd verify placed props (body + shape + avoidance radius, all three venues) and library dressing (body unless documented exempt). Mutation-checked: removing _add_prop_collision fails 3 tests, dropping handrail from the exemption list fails with the exact node. Suite is 275/275.)
• Add a coverage test so every catalogue asset has a scorecard row and verdict. (tools/test_model_review.py: every CHARACTERS/GEAR/PROPS entry needs a row in docs/assets/3d-model-review.md, and character rows need a non-empty verdict.)
• Document the new gates: docs/assets/blender-assets.md now lists texels, the test_model_review.py coverage gate and the test_world.gd collider/proxy gate; skill routes live in AGENTS.md §Local Blender skills and each TODO.md section; the scorecard and evidence table are in docs/assets/3d-model-review.md.
• Record accepted limits/deferred rebuilds: docs/design/known-limitations.md already records the stylized low-poly style, no finger-curl rig and the >26 m impostor; MEMORY.md holds the durable decisions (texel density accepted, cloth sim out of scope, open palm + forward offset, face subdivision prototype gated on the owner). The corrected face note is also in docs/assets/blender-assets.md.

Local map workstreams (H–M). Build a local, data-backed map pipeline that can replace or supplement the stylized venue geometry with licensed OpenStreetMap footprints, 3D building massing, reference-derived photoreal facades and AI-assisted composition. This remains part of the same unversioned track, not a release gate. The shared reference/provenance gates in A/F still apply; the map pipeline has its own exit in M.

Local-map skill route: environment-artist, archviz, hard-surface, geometry-nodes, procedural-modeling, vegetation-artist, set-dressing, lighting, materials, texture-workflow, lookdev, scene-assembly, qa-review.

H. Local map data, licensing and scope

• First target block + role: Hollywood (El Capitan, 6840 Hollywood Blvd) and the Comedy Store (8433 Sunset Blvd); the result supplements/informs the stylized VENUE_SPECS (reference-derived materials, facades and massing), never replaces them with imported OSM geometry. Recorded in both venue map docs.
• ODbL audit: OpenStreetMap is used as reference only (hand-built geometry informed by OSM ways + Commons photos), so the shipped geometry is not an ODbL derived database and no share-alike obligation attaches. Attribution is still kept via the OSM way links in docs/design/venues.md#hollywood-boulevard / docs/design/venues.md#the-joke-shop. If derived OSM geometry ever ships, it needs the attribution + share-alike review in a separate ADR.
• OSM extract workflow: no offline extract is required while OSM is reference-only. If the pipeline later imports footprints, pin an Overpass query + extract date and record it in a provenance sidecar (see below); no city-scale extract is imported into the runtime.
• 3D building datasets: OSM Buildings/Overture/CityGML are not needed for the reference-only approach. If massing import returns, OSM Buildings LOD1 is the first candidate (CC0-ish footprints); CityGML/Overture need a licence review per source.
• Provenance sidecar: map sources are attributed in the venue map docs (OSM ways, Commons photos, local-only research sets). Any future derived map asset adds a sidecar with source URL, licence, capture/extract date and attribution text before it is committed.

I. Local map reference imagery

Recent-photo research (2026-11-03): Commons geosearch around 34.10118/-118.34009 (500 m) plus category/date sorting shows usable imagery from the last 12 months: Dolby Theatre billboards (2026-09-30, CC0, 6 frames), Dolby street level and a fresh Walk of Fame star (2026-07-17, CC0), a Michael Jackson impersonator at the El Capitan frontage (2026-02-25, CC BY-SA 4.0), the Celia Cruz Square marker and a 2025-09-28 star series (CC0), and a Thunderbolts premiere (2025-04-28, CC BY-SA 2.0). No recent frame shoots the El Capitan facade straight-on in daylight; the best facade evidence is still the 2009 CC BY frame. The 8 downloaded candidates are in tmp/research/elcapitan-2026/ (gitignored) with meta.json; nothing was committed.

• Wire tools/refboard.py (standalone, stdlib-only): search --source search|geo|category lists candidates with ISO date, age-in-months and a commit/review/skip licence gate (--allow-ccbysa after a recorded review); fetch downloads usable rows into tmp/ with spaced requests and writes meta.json; rows emits SIDECAR + README table rows from the manifest. Six recent CC0 photos of the block were re-fetched into tmp/research/elcapitan-2026-tool/. Tests in tools/test_refboard.py (licence gate, recent filter, review-row exclusion, row emission).

• Add a --max-age-months filter (default 12) and a --must-reuse mode that matches the folder's existing sidecar entries so a refresh only reports what changed.

• Research-diff step: keep the last-12-months board as the "current reality" evidence and the older photos as the geometry evidence, labelled per photo; no sidecar date field needs to change, only the purpose column.

• Write a reference shot list per block: facade materials, windows, signage, rooflines, street furniture, curb cuts and lighting.

• Download and catalogue committed reference images from Wikimedia Commons, local/own photography or other clearly licensed sources; record author, licence and attribution with the §F sidecar template. Local-only research that ignores licence limits (DuckDuckGo, street-view frames, neighbours, across-street buildings) is collected under §A into tmp/research/ and is never committed.

• Separate development-only references from shipped textures; commit only images with clear redistribution rights and keep the rest under tmp/ or the ignored reference folders.

• Build the per-block real-photo board under docs/screens/refs/<block>/ using the sidecar template from the 3D review §E, and link it from the venue map documents.

• Define the photoreal target and a material-derivation workflow (albedo/normal/roughness, AO, weathering) that matches the existing Compatibility renderer (texture-workflow, materials, lookdev), with the recent 2026 street frames as the current-reality reference and the 2009 facade frame as the geometry reference.

• Remaining Hollywood gap: a licensed night photograph of the hero facade and blade sign (the night frame on the board is a different block, near Vine), plus a sidewalk-level shot of the steps/door recess and the east retail neighbour. For the committed board the licence gate stays (no unlicensed Street View or scraped social media); local-only copies of the same viewpoints go through section A instead.

J. Local map geometry and AI-assisted composing

• OSM footprint import and 3D building massing: deferred — the §H decision keeps OSM as reference-only, so no footprint/massing import runs; the procedural VENUE_SPECS plus the reference-derived materials carry the geometry instead.
• Facade kit approach: superseded for this pass — §K delivered the banding/mass via procedural geometry plus photo-anchored tiles (limestone_ashlar, dark_stucco). A reusable Blender facade kit (window bays, cornices, storefronts) is a future enhancement, not a blocker for the current maps.
• Joke Shop building-appearance pass: dark_stucco tile (1024 px) promoted via the picks pipeline and wired into _build_comedy_store_building, replacing the grey paver tile so the hall reads as flat matte charcoal instead of masonry; photo-anchored day concept generated from the hillside PD frame (the branded frontage photo trips the model's image-recitation filter — recorded in the concept README); day capture promoted and the three-up board rebuilt. Test asserts the hall wall carries dark_stucco.
• Building-appearance realism pass, measured against the reference set: bring the masonic facade's material response (limestone value/roughness, joint lines, weathering streaks), the banner/canopy and marquee materials, and the night emissive balance to the level of the 2009 facade photo and the 2026 street frames; capture before/after and record each delta. (Ashlar field: limestone_ashlar tile, reference-anchored on the facade photo, 768 px over ~4.6 m of wall = ~167 px/m, promoted via the picks pipeline, warm-shifted tint behind VENUE_SPECS. Banner cloth: warm self-light registered with _marquee_emissive (0.14 base; 0.28 night / 0.028 day). Marquee sign face + bulbs: warm-incandescent already in the prop PALETTE, now test-verified to brighten at night via _marquee_emissive. Day and night captures promoted, both three-up boards rebuilt.)
• Upgrade the facade from flat material bands to 3D photoreal detail (real cornices/reveals, not just the ashlar/stucco tiles). Recorded as a deferred enhancement — the tiles already give the material read; the remaining gap is geometric reveal depth (texture-projected or Blender modules). Verify texel density, tiling and night look when it lands.
• Integrate behind WorldBuilder.VENUE_SPECS: the photo-anchored tiles are wired into _build_masonic_building / _build_comedy_store_building with colliders, navigation obstacles and the existing day/night materials.

K. Comedy Store map recreation

Plan (2026-11-03): rebuild the Joke Shop venue geometry from the new local-only reference set (tmp/research/comedy-store/, gitignored) and the four photo-anchored scene pairs now in docs/screens/concepts/joke_shop/. Evidence wins for shape; the concepts win for material/mood. Nothing is committed until the licence gate on any promoted reference and the map-budget tests pass. The current _build_comedy_store_building() stays as the fallback.

• Extract the real proportions and reconcile: vision-read tmp/research/comedy-store/ frames via tools/describe_image.py and recorded the reconciled table in docs/design/venues.md#the-joke-shop §Measured proportions. Key finding: the game facade is proportionally taller than the real (3.2:1 vs ~6:1) — the main shape-language deviation; canopy pitch, blade-sign diameter, entrance and roofline all match closely. Curtain-bay count is crop-dependent (3–7), so 7 is accepted as a readable stylization.
• Rank the shape-language gaps and write the top-five fix list (measurements from the proportions pass, §docs/design/venues.md#the-joke-shop): 1. Facade height — game 8.4 m vs real ~5.5–6 m (≈3.2:1 vs ~6:1 ratio). Fixed 2026-11-03: facade_height lowered to 6.0 m; the wall now reads as a low single-storey hall with the canopy capping the top (vision-verified). 2. Rounded west corner — the real building has a ~3–4 m radius corner; the game uses a hard 90° corner, and the entrance sits at the rounded corner. 3. Canopy depth/slope — game 4.9 m at 11° vs real ~4–4.5 m with a shallower pitch. 4. Neighbouring hotel/parking mass — absent; the hotel reference frame shows a low blocker mass east of the club. 5. Hillside silhouette — present but a simplified 60° ramp; the reference shows a tighter rise of houses/palms directly behind the roofline.
• Canopy + blade sign: kept as procedural geometry in _build_comedy_store_building (no separate library prop needed) — the circular 3.9 m white disc on a roof pole and the sloping black canopy with the white signature fascia both match the reference and were vision-verified against the entrance concept.
• Add the neighbouring hotel mass: WestHotelTower now rises behind the west cream block (12 × 18 m pink slab with a parapet and a 5×4 window grid, stepped back from the podium) matching the Ramada Plaza reference; Joke Shop map is 26,828 tris / 580 meshes (under the 150k/700 budget), Godot 277/277. The parking-ramp/drop-off detail stays out (low-detail blocker mass only).
• Facade field banding: dark_stucco field, deep black canopy + shadow, white signature fascia, seven tied-back red curtain drapes and the west recessed entrance all landed and were vision-verified; the remaining shape gaps (facade height, rounded corner) are recorded in the top-five list.
• Gates after the rebuild: Godot 277/277 (map budget 26,828 tris / 580 meshes under 150k/700; navigation/collider tests pass), tools/check_screenshots.sh 21 good / 0 bad; the three-up board and docs/design/venues.md#the-joke-shop were refreshed.
• Update docs/project/development-status.md, docs/assets/asset-tooling.md and CHANGELOG.md when the rebuild lands; record any accepted deviations from the reference.

L. Generated location imagery

Skill route: lookdev, lighting, environment-artist, texture-workflow; tooling: tools/scene_concepts.py, tools/asset_pipeline.py, tools/screen_designs.py.

Tool plan for reference-anchored generation (plan first, implement after this section's review):

• Add scene_concepts.py generate --reference <photo.png>: attach a licence-cleared real photo as a source image to the OpenRouter image request (multimodal chat endpoint with a base64 data URL), with a prompt suffix that preserves the photographed facade geometry. Record the reference file, its licence and page in the .refgen.json provenance sidecar next to the generated frame.
• Extend generate_textures.generate_image() with an optional reference_path argument (base64 data URL of the photo); existing call signatures unchanged, and the path is gated by scene_concepts._reference_source requiring a refs-board SIDECAR.md row (or an explicit --reference-unchecked). Verified live on the El Capitan day concept anchored on the CC BY 2.0 2009 facade photo; tests in ReferenceGateTests.
• Add a sidecar-driven variant pass: scene_concepts.py anchor --reference <photo> --treatments day,night,weathered generates same-framing variants with treatment clauses, one .refgen.json provenance sidecar per frame. Verified live on the facade photo (night + overcast frames committed; the weathered treatment reads as mild overcast — noted on the board). Tests in AnchorTests.
• Concept frames generated and committed: El Capitan facade day/night/weathered + entrance day-v2; Comedy Store facade/street/strip/visitor pairs (gemini + gpt) + entrance day/night + across-street; Jimmy Kimmel entrance day/night × two models. Each frame carries a .refgen.json sidecar (model, prompt, source photo).
• Palette/lighting/material targets derived: limestone_ashlar (masonic field), dark_stucco (comedy hall), the warm banner/marquee emissive and the awning tile are the material targets; the deltas (3D reveal depth) are recorded in §J as the deferred facade-detail enhancement.
• Generate signage, awning and poster texture candidates with asset_pipeline.py; run the role audit and the CREDITS.md provenance check before anything is used in the kit.
• Compare generated concepts against the licensed photos: where they disagree, the photos win for shape and the generated board only informs colour, mood and detail density. Human review stays the acceptance gate, and AI output is never treated as licensed map data.

M. Local map QA and exit

• Compare the block against screenshots/map docs and keep the stylized fallback: the three-up boards (photo | concept | capture) and docs/design/venues.md#the-joke-shop record the evidence chain; the procedural _build_comedy_store_building() remains the fallback.
• Budget/navigation/collider + day/night captures: Godot 277/277 (Joke Shop 26,828 tris / 580 meshes under 150k/700; navmesh, step collider and prop-collider tests pass), tools/check_screenshots.sh 21 good / 0 bad.
• Gallery + before/after: the Joke Shop day gallery frame and three-up board were recaptured after each facade change; the photo/concept/capture strip is the before/after record.
• tools/check_screenshots.sh re-run (21/0) and the Joke Shop three-up board refreshed.
• Docs: docs/design/venues.md#the-joke-shop (mapping, measured proportions, simplifications), docs/assets/asset-tooling.md (describe_image.py) and docs/project/development-status.md updated. CHANGELOG.md is untouched (unversioned track; no release).
• Exit: one local block (the Comedy Store) is reconstructed with reference-derived facades and massing — 6.0 m low hall, white signature fascia, tied-back curtains, hotel tower — passes performance/navigation/visual QA with complete provenance; the rounded west corner and hillside detail are recorded as accepted simplifications.

Maintenance and documentation hygiene

Unversioned housekeeping that keeps the planning sources trustworthy. Do these when they block a task, not on a schedule.

Status (2026-11-05): v0.9.8 published; three venues; 277 headless + 162 Python tests; the Hollywood and Joke Shop facade-realism passes landed (limestone_ashlar, dark_stucco, banner night emissive, marquee emissive test); reference boards (hollywood, joke-shop) with sidecars; three-up boards; tools/refboard.py + reference-anchored generation (--reference, anchor) + tools/three_up.py. Gaussian splatting stays deferred at the end of this file.

Memory and status accuracy

• Trim MEMORY.md from ~1088 lines to the durable set: dropped the superseded 2026-11-01 handoff, the v0.2.x host/disk/NSIS/release anecdotes and duplicated release notes; kept the gotchas (async/event wiring, import order, colliders, RNG, contracts, Godot --import cache, image-recitation filter, capture-harness windowing, bisect zero-area triangles) and the easter-egg note. Now ~800 lines.
• Re-check the remaining memory claims against code before deleting them; MAX_LIVE_NPCS is still 16 for gameplay while the separate budget smoke spawns 24, so do not "fix" that. (Verified phase timing is data-driven via shift_phases, fixed the stale REQUIRE_SIGNIN_VIEW contradiction and the old 272/108 counts.)
• Fix stale status claims as they are found; add a documentation check or grep note only if the same class of drift returns. (Counts and the sign-in state corrected on 2026-11-03.)
• Keep the PLAN.md gap-analysis baseline and the README status line in sync when a test count, version or feature milestone changes (baseline bumped to 277/162 on 2026-11-05).

Imagery and provenance upkeep

• Keep docs/screens/refs/ sidecars, concept-board records and CREDITS.md in sync whenever an image is added, replaced or dropped; photo-anchored concept frames must keep their .refgen.json provenance sidecar (source photo, licence, page, model) next to the frame.
• Run python3 tools/check_docs.py after any gallery or reference-board edit so every tracked image stays linked from its folder README.
• Use tools/refboard.py search/fetch/rows for new Commons research instead of ad-hoc API scripts; downloads stay under tmp/ and only licence-cleared rows (commit/review gate) reach a board after a sidecar entry.
• Delete or ignore development-only photos and raw generations once a board is committed; the repository carries the picks and their provenance, not the working set.

Godot import-cache hygiene

• When a new committed texture (assets/textures/..., picks/) fails to import under the existing .godot/ cache, use the scratch-copy import (rsync --exclude .godot ... then godot --headless --import and copy back the .import + .godot/imported/*.ctex/.md5); never hand-edit the uid= in a copied .import. Record any new instance in MEMORY.
• Prefer tools/scene_concepts.py promote (or a manual move into game_ready/v2/ plus a PICKS entry) so new picks follow the same path as limestone_ashlar and dark_stucco.

Agent skill upkeep

• Add the tracked project skills gig-line-blender-asset-contract and gig-line-release-qa under .agents/skills/, with references for the asset contract/verification and release/QA runbooks.
• Keep the third-party blender-skills pack local/gitignored while allowing project-authored skills to be committed; reference the skills from tools/README.md, AGENTS.md and MEMORY.md.
• Revisit the skills after the next Blender or release work changes the contract, commands or gates; stale skills are worse than no skill.

Blender skills upkeep

• Re-check the installed Blender pack (version.json) when updating it and record the version in AGENTS.md if it changes; keep .agents/ gitignored and .agents/.gdignore in place.
• Prune the skill routes in AGENTS.md if a skill stops matching the project (for example if a future second-rig path makes rigging/animation guidance authoritative).

Gaussian splatting evaluation (deferred)

Postponed for further development on 2026-11-03. The local-map track ships the facade kit and reference-anchored material passes first; this photo-driven pipeline stays an unversioned, offline study with no runtime or release dependency until an ADR closes it. All licence and privacy gates below remain in force whenever the track resumes.

Photo intake and provenance

Gaussian splatting is evaluated as a photo-driven pipeline: licensed or permissioned photos become a reference, texture source, mesh proxy or distant backdrop. It is an unversioned local track with no runtime dependency or release gate until the ADR closes it, and it extends the reference-image licence and privacy gates above rather than replacing them.

• Define the accepted photo channels: project/team captures, volunteer or user-supplied photos with written permission, and downloaded images from clearly licensed archives (Wikimedia Commons, Openverse, Mapillary and similar). Record which channel produced each reconstruction.
• Write a one-page capture brief for user-supplied photos: overlapping arc or orbit around the facade, 60–80% view overlap, 12 MP+ originals, fixed exposure/white balance, HDR and filters off, no motion blur, no people or plates in frame where avoidable, overcast or even light, and the facade height/doors/steps included in several frames.
• Build an intake checklist and provenance sidecar per photo set: contributor, consent or property permission, licence and attribution text, source URL, capture date, camera/lens, location, faces/plates/interiors present, redactions applied, file hashes and whether the set is shipped or development-only.
• Set the licence gate: own photos need an explicit contributor licence; CC0/CC BY are preferred; CC BY-SA triggers a share-alike review for derived textures and meshes; reject Street View, social-media scrapes, unlicensed commercial imagery and AI-upscaled copies of unknown originals. Attribution must be feasible in CREDITS.md before use.
• Set the privacy gate: blur or remove faces, minors, licence plates, screens, private interiors and anything behind a closed door; obtain property permission where required; keep a redacted copy as the reconstruction input and never commit the originals.
• Treat downloaded photo sets as sparse and of unknown intrinsics. Use them for reference, palette and texture study, or as extra training views; do not expect a usable splat from 10–30 casual web photos of a facade.
• Keep photos, masks, COLMAP databases, checkpoints and splats outside git; commit only derived, licensed, provenance-stamped assets that pass the review gate.

Reconstruction and alignment

• Pin a reproducible offline toolchain (for example COLMAP or GLOMAP for camera poses plus gsplat, OpenSplat, Nerfstudio or Brush for training), with exact versions, GPU/VRAM/disk needs and known failure modes recorded in the sidecar.
• Reconstruct two quality tiers: a dense 100–400-photo arc/orbit capture and a sparse 10–30-photo downloaded set. Record camera-registration rate, training time, splat count, file size and held-out PSNR/SSIM/LPIPS for both.
• Mask sky, people, vehicles and moving foliage before training; verify the masks do not erase facade edges, signage or window detail, and keep the redacted input variant.
• Align the splat to game coordinates from the OSM footprint plus surveyed points (door width, column spacing, step rise, storefront height) and record alignment residuals in a Blender scene containing both the splat and the current venue geometry (scene-assembly).
• Judge quality from held-out views and a render from the game's day/night camera poses (camera-cinematography, lookdev, qa-review), not from training views alone; reject splats that only look correct from the capture path.
• Maintain a coverage map per block: which facades, corners and rooflines have usable views, and which areas must fall back to the facade kit or stylized geometry.

Map enhancement paths

• Path A — Reference and QA (lowest risk): render the splat from game camera poses and compare it with the realism captures; derive dimensions, signage placement, facade rhythm and material colours. No splat data reaches the runtime.
• Path B — Texture derivation: project splat renders or reference photos onto the facade kit to produce albedo, normal, roughness and AO candidates; de-light and bake them (texture-workflow), then verify UV scale, tiling and readability under the Compatibility renderer. Deriving directly from a committed reference photo requires its licence to allow derived works and a CREDITS.md entry; CC BY-SA sources stay out unless reviewed.
• Path C — Mesh proxy: extract a mesh from the splat or dense reconstruction, retopo and dress it in Blender (retopology, prop-artist), assign LODs (lod-pipeline), and keep it inside the venue triangle/material budgets with mesh-based colliders (collision-proxy).
• Path D — Distant backdrop: test a baked panorama, impostor or billboard generated from the splat for skyline and opposite-street depth where the player cannot walk; measure file size, loading and visible pop at the venue boundary.
• Path E — User-photo supplement: let a small supplied photo set improve one facade or landmark (a signage patch, colour reference or hero detail) without rebuilding the venue; define the minimum evidence and the same licence/privacy/review gate as a full capture.
• For every path record quality gain versus effort, licence compatibility of the derived output, and whether the result may ship or stays development-only. Prefer the cheapest path that closes the visible gap.

Runtime evaluation and decision

• Probe Godot 4.7.2 Compatibility and Web support for splat rendering in a throwaway scene: addon or importer availability, shader/sort cost, memory, load time and fallback behaviour. Do not add a splat dependency to the release project during evaluation.
• Set hard limits before any runtime experiment: maximum per-splat file size and total memory, allowed frame-time delta on the 24-NPC budget scene, acceptable Web load time, and no compute-shader requirement for the Compatibility/Web path.
• Keep collision and navigation mesh-based even if a splat is displayed; a splat must never become the playable floor or the access/occlusion authority.
• Write the ADR with a per-path go/no-go, the accepted licence/provenance rules, whether each result ships or stays development-only, rejected alternatives and the evidence bundle.
• Exit: one licensed photo capture measurably improves a venue through a documented path and passes visual, performance and provenance review; or the approach is rejected with reproducible findings and removed from the release path.

Milestone 0.9.10 — NPC roster rollout (patch)

Goal: extend the approved base to the full 12-role roster in asset-family batches, without regressing gear fit, budget or readability.

Roster build

• Map the 12 roles to the three full Quaternius GLBs (owner chose the full swap): hoodie for casual/coat/hoodie/crowd_c, casual tee for jacket/dress/crowd_a/crowd_b, worker for vest/firefighter/guard/crowd_d. npc_firefighter, npc_guard and npc_crowd_d drop the upstream hivis vest and hard hat. All twelve built, AO-baked and committed (a6ccd86).
• Hair part library. Every base ships one hair shell; QUATERNIUS_HAIR_PARTS reshapes it in place (short identity, buzz squashed toward the hairline, long stretched from the crown), and each role picks a style through the existing CHARACTERS[...]["hair"] field. The T-pose cache is keyed on the style. The roster now reads as three silhouettes instead of one.
• Outfit part library — intake landed (Suit pilot). The owner downloaded the Ultimate Modular Men pack (CC0, 15 families, FBX only) into tmp/quaternius-intake/. tools/quaternius_parts.py make-base composes a family into a skinned base GLB, undoing five FBX artefacts (per-file armatures, _end/neutral_bone, node transforms, the shared jacket/trouser fabric, seam degenerates). Suit.glb is the first base: npc_coat is now a dark formal suit instead of a tinted hoodie. The pitfalls are recorded in docs/assets/npc-model-sourcing.md and MEMORY.md.
• Outfit families rolled out. Seven more bases intaken (Adventurer, Beach, Casual_2, Farmer, King, Punk, Suit) with a slot map each: the roster now runs eleven distinct outfits across twelve roles. patch_slots buries a hidden triangle for slots a family cannot paint (Farmer has no hair shell), so the five-slot contract holds on helmeted and hatted heads.
• Swat base faces backwards, and three yaw attempts failed. It is authored facing Godot −Z. Tried and rejected: (a) rotating the exported armature node — the skin turns with it and the character renders headless; (b) rotating armature + mesh objects — collapses into a spike; (c) rotating mesh data plus the rest bones after transform_apply — still headless. Next idea: yaw each of the four FBX parts in its own import (before any join), or re-author the base from the pack's All together/Humans_Master scene where facing is known. npc_guard stays on the worker base; Swat.glb is intaken but unused.
• SpaceSuit is not intaken: it has no skin, eye or hair geometry at all and no role fits an astronaut on Hollywood Boulevard. Weapons (pistol, sword) is unused by design — the game has no weapon system.
• Heavy outfits cap the texel density: npc_dress (King, 11k tris) runs at 212 px/m on the 4× fabric weave where the lighter bases reach 320. The test floor moved to 200 px/m; a per-slot atlas would recover the difference.
• Intake tooling: inventory / extract / verify / make-base in tools/quaternius_parts.py, all exercised against the real pack.
• Bake per-slot albedo/normal/ORM maps and a per-slot atlas; the five slots currently share one whole-body UV set, which caps how much the cloth/pants tints can differ.

Owner review findings (2026-11-06)

The owner signed the roster off with follow-ups and rejected three review rows (faces at portrait, hands and hand-held gear, head gear on the skull). All follow-ups landed and the roster was accepted on 2026-11-06; the review page and its sheets were then retired.

• Worker torso holes on npc_guard: the build deleted Worker_Vest, which on the modular Worker is the torso. Fixed by mapping it to cloth instead of deleting it.
• Worker crown hole on npc_firefighter/npc_crowd_d: the deleted Worker_Yellow hat was the crown of the skull. Fixed by keeping it and splitting the material at z 1.65 (crown → hair, vest stripes → cloth).
• Unibrow on npc_guard/npc_vest/npc_crowd_d: the Worker Eyebrows bar and its oversized eye quads are cut out, and the hoodie's two brows are transplanted onto the Worker head, lifted 18 mm clear of the eye line (WORKER_HEAD_DONOR).
• Worker crown reads as a hard hat: the hoodie's hair is transplanted onto the Worker head and the hard-hat faces above the collar are cut away, so guard/responder/crowd_d have hair.
• Worker roles (vest, firefighter, guard, crowd_d) had no visible eyes: the Worker's own quads sit inside the head. The hoodie's eye pair is now transplanted with a 4 mm forward offset, so all four roles show two eyes and two brows (owner request).
• Owner: remove all hats except the firefighter helmet, and remove the performer cone. Caps and hoods are gone from ACCESSORIES, the per-character gear list and the primitive fallbacks; the performer keeps only the baton and board in hand. The head-gear lift/scale compensation those hats needed is deleted with them, so firefighter_helmet is the only head item and no role can hide its face.
• Worker faces: the dark "unibrow" band turned out to be the cap brim sitting on the brow, not head geometry. HEAD_GEAR_LIFT went 0.07 → 0.103 (the gear is authored around the head centre at 1.62, the Quaternius head centre is 1.69) and GEAR_SCALE grows the cap/hood so they clear the 0.228 m hair shell. The cap now reads as a cap and both eyes and brows show.
• Hair poking through head gear: fixed by the same lift + cap/hood scale; the firefighter helmet dome covers the hair, and the cap reads as a cap rather than a visor.
• Eyes now have a sclera and a pupil, done as geometry rather than a texture: the existing eye quads became the light sclera and a shrunken copy pushed along the sclera's own normal became the dark pupil (_add_quaternius_eye_whites()). Two extra draw calls per head, no new UV work.
• Review page re-created for the second pass, then retired after the owner accepted the roster (2026-11-06). The durable evidence is docs/assets/3d-model-review.md, the measurements in docs/assets/blender-assets.md, and the capture_npc_gear.gd / capture_npc_baseline.gd scripts.
• [-] Arm/torso intersection on npc_dress/npc_jacket/npc_crowd_a/npc_crowd_d: the baked rest now aims the upper arm 7° out from vertical. Verified clear on the casual base; a light clip remains on worker, so this needs either a wider angle or a per-base shoulder offset.
• Worker crown still reads as a hard hat: the shape is upstream geometry, so it needs the modular hair/cap library (or a carved brim) instead of a tint.
• Eyes and brows still read as one heavy dark mass at portrait distance (checklist row 3).
• Head gear still does not sit on the skull convincingly (row 5): the firefighter helmet brim is 1.55× the head width and the cap sits high.
• Carried role props floated beside the body: _add_props() placed the performer hat/board and the protester pole/sign at fixed rig offsets. They now ride the hand proxies (_hand_for()), and the performer cone was resized from a 0.44 m base to 0.25 m so it sits on the 0.20 m head instead of swallowing it.
• Hand-held gear is still authored for the old hand (row 4): the npc_vest clipboard is oversized on the shorter Quaternius arm.
• [-] Build one asset family per commit, AO-baked and re-imported; never run a no-op full-library rebuild. Use the concept + vision-judge loop for head/outfit decisions and record model, prompt and spend. (Roster family landed in one commit because the three sources share one pipeline; the 12 × 6 baseline sheet is captured but not yet owner-approved.)
• Keep every variant inside 24,000 tris / 24,000 glTF vertices and the five tint slots; no new bone beyond the accepted 62-joint set without an ADR.
• Refresh CHARACTERS, the manifest roles/tint data, CREDITS.md and the review doc rows as each family lands.

Prop and gear fit pass (new NPC attach points)

Goal: every gear piece and role prop is measured against, and re-authored for, the Quaternius skeleton. The 0.9.9 roster swap moved the hand from y 0.78 to the wrist at y 1.017, the head centre from 1.62 to 1.69 and the shoulder width from 0.27 to 0.15, so anything authored for the procedural body is now the wrong size or in the wrong place. Nothing in this workstream changes the runtime contract.

• Extend tools/blender_assets.py fit to measure hand gear (it skipped hand outright). It now reports grip drop, the lowest point at rest against the Quaternius wrist (y 1.017), the hip-line flag (0.95), the authored size and size_vs_head — the proportion number that makes "reads oversized" measurable.
• Record the nine hand-item measurements (table in docs/assets/blender-assets.md). Only the clipboard was actually mis-authored: its grip sat in the middle of the board, so it hung through the thigh line. It now grips the top edge like a carried clipboard (0.22 x 0.30 m).
• Re-author the clipboard (top-edge grip, 0.22 x 0.30 m) and rebuild + re-bake AO. The umbrella (0.90 m) measures on its real-world size and keeps its grip, so it stays.
• Re-fit the role props in NpcAvatar._add_props(): the delivery parcel (0.34 m slab across the torso) is now a 0.26 m carried box in the hand, and the confused visitor carries a folded map instead of a tray hovering at the hip.
• The primitive cap/hood fallbacks in _add_accessory() are deleted with the hat removal; the remaining hand-item primitives are unreachable while the gear GLBs exist and stay as the documented fallback.
• Retire the gear GLBs the roster no longer uses: cap and hood are gone from GEAR, the manifest and the asset tree (beanie was already retired). The builder branches stay for a future design, guarded by a comment.
• docs/assets/blender-assets.md documents the 15-piece matrix and the hand-item fit table; architecture.md, asset-tooling.md and the preview index were updated with the new counts. CREDITS.md still cites the upstream pack, which is unchanged.
• Owner feedback: the back gear hung at the waist. The back attach moved from y 1.10 to 1.26 (shoulder-blade height on the Quaternius torso); backpack, shoulder bag and SCBA tank re-captured on the character sheet.
• Re-measure fit for the back and chest gear on the slimmer torso: backpack inner x 0.150 against a 0.190 clothing radius (straps cross the body, as intended), shoulder_bag x clearance +0.072, scba_tank inner x 0.150, lanyard radial clearance +0.012 on the chest. The backpack straps were re-fitted for this torso in fb3e75a; nothing further needed.
• Rebuilt docs/blender_previews/sheet_npc_gear.jpg for the 15-piece set (all renders regenerated) and re-ran the 24-NPC budget smoke (175,264 tris, 0.87 ms).
• New src/test/capture_npc_gear.gd captures every gear piece worn by a real avatar through the runtime attach path, one frame per item, and the gear-on-character sheet (retired with the review pages) collected them. It caught two things the asset renders could not: sunglasses were hidden inside the hair shell, and the outfit's own gear double-attached over the tested piece (the capture now clears it first).
• Owner feedback on grip height applied: every hand item drops into the palm (HAND_ITEM_DROP, camera 0.085 m, flashlight 0.065, cup 0.050, umbrella 0.045, clipboard and board 0.035, phone/radio/wand 0.030), the carried role props drop 0.06 m, and the lanyard attach moved from y 1.30 to 1.38 for the taller Quaternius sternum.
• Owner verdict on the hand-held gear: accepted (2026-11-06).

Exit: every gear piece and role prop has a measured fit against the Quaternius attach points, the hand-held items are within +/-10 % of their authored real-world size, and the review page row 4 is signed off.

Gates and rollout

• Re-bake AO per rebuilt GLB and refresh the stale .godot import cache (scratch-copy import) before judging any face or body change.
• Run inspect/audit/fit/texels and npc_proportions.py report; record the 12-variant measurement table.
• [-] Capture the 12 × 6 baseline and get owner approval before replacing the committed GLBs. Owner signed off on 2026-11-06; the review page was retired afterwards.
• Re-run the 24-NPC budget smoke and the Web/Compatibility capture; record frame time, mesh/triangle counts and Web load before/after.
• Retire the procedural variants that are no longer used, keeping one tested fallback; update docs/project/development-status.md, docs/assets/3d-model-review.md, docs/blender-previews sheets and the README character row.
• Release 0.9.10 with the full procedure (version, changelog, local build, checksums, tag).
• Shrink the release artifacts: assets/ alone is ~273 MB in the Linux binary. Excluding tmp/, build/, docs/, tools/, src/test/ and .agents/ from the export presets already cut 426 MB to 273 MB; the remaining bulk is texture resolution.

Exit: all 12 roles use the approved base or an explicitly kept fallback, every variant passes the contract and gear gates, the 24-NPC and Web budgets are measured and documented, and the baseline sheet is owner-approved.

Milestone 0.9.11 — Venue, prop and lighting polish (patch)

Goal: close the scheduled subset of the unversioned 3D review backlog. Only ranked P1/P2 items whose fix is visible at gameplay distance are in scope; the review doc stays the evidence and scorecard source, this milestone is the actionable list.

Props, gear and foliage

• Rebuild the ranked prop candidates from §D: the foliage-language P1 (planter_box/round_planter foliage, street_tree/palm_tree), plus the newspaper_box slot and bollard collar if the budget allows.
• Replace the flat _shrub() clumps with leaf cards (_leaf_card()), the foliage-language P1. Each shrub is now one low-poly inner mass wrapped in 18–53 two-triangle cards scattered on the canopy shell and tilted off its normal, so the silhouette is ragged instead of rock-like — and it is cheaper: planter_box 1,324 → 818 tris, round_planter 1,240 → 808, street_tree 2,592 → 1,506. Props total 13,808 → 11,784 tris (budget 60k).
• Review the 18 gear pieces for silhouette and strap/buckle readability and rebuild only what fails; re-run fit after every change.
• Verify prop colliders and navigation obstacles against the final silhouettes.

Venue, lighting and set dressing

• Joke Shop / Hollywood shape language: fix the ranked remaining items (west rounded corner, canopy pitch, facade reveal depth) only where a capture shows a readable gain.
• Flashlight tuned by measurement. The late-night capture proved it did nothing: energy 4.2, 20 m range and a 14° cone changed the frame by 0.1/255. Now 10 energy, 45 m, 24°, tilted 5° down — the aimed region goes 13.5 → 29.7 and the near pavement 3.7 → 11.7 (0–255), and the committed 06-night-flashlight shot shows the beam on the steps and the NPC in front.
• Marquee/event-banner emissive read under the Compatibility renderer: measure the banner luminance against the reference night board and adjust only if it still clips.
• Street-lamp pools. Added a lamp_late shot that frames a lamp, and made the lamps StreetLamp_* so they tune apart from the sconces. Attenuation must stay 1.0 (the venue test requires physical falloff), so the pool comes from range and energy: 16 m/2.2 → 10 m/8.0 takes the pool 14.1 → 16.0 and the gap between poles 2.7 → 1.3, pool/gap 5.2 → 12.3.
• Refresh docs/screenshots/ for the whole roster and the new flashlight; check_screenshots.sh reports 21 good / 0 bad.
• Set dressing: fix the Joke Shop queue story (podium/stanchion line at the door), rope barrier thickness, the west palm overlap and the empty sidewalk light pool; apply the same review to Hollywood and the TV Studio.
• Build one real-photo comparison for street furniture and record the accepted mismatches.

Evidence and release

• sheet_props.jpg and sheet_npc_gear.jpg rebuilt; the 21-image venue gallery regenerated.
• Re-run tools/check_screenshots.sh and python3 tools/check_docs.py; record both results.
• Release 0.9.11 with the full procedure.

Exit: every scheduled P1/P2 item is fixed or explicitly accepted with a measurement, the capture gallery and asset sheets are current, and no unreviewed asset remains in the shipped catalogue.

Milestone 0.9.12 — Performance, Web and accessibility (patch)

Goal: prove the new character and venue content still meets the frame, memory, load and accessibility targets on every target before the minor release.

Performance and platform

• Profile the 24-NPC budget scene on Compatibility/Web and Desktop HQ; record frame time, draw calls, texture memory and pck size against the pre-character baselines.
• Verify LOD/impostor thresholds, the 60k library and 150k/700 venue budgets, and the Web load time after the character and texture changes.
• Confirm the Compatibility/Web build still avoids compute shaders and Vulkan-only features; spot-check the HQ fallback message.
• Re-run the save/session regression suite on an isolated user-data directory after the content changes.

Accessibility and localization readiness

• Verify font scaling at maximum, the longest strings, caption coverage, reduced motion, colour-independent escalation cues and the hold-to-confirm emergency path.
• Run the string catalogue through a pseudo-locale/length check; move remaining hardcoded UI strings into content/strings.en.json.
• Fix only the issues the checks expose; new accessibility features are 0.11.0 scope.

Release

• Regenerate the full capture gallery and the screen-design judge from the current build; record model, spend and verdict.
• Release 0.9.12 with the full procedure.

Exit: the character and venue changes do not regress frame time, memory, Web load, save/session reliability or accessibility behaviour, and both automated gates are green.

Milestone 0.10.0 — Public-readiness release (minor)

Goal: ship the new characters, the polished venues and the public-readiness gates as the 0.10.0 minor, after the patch milestones above have landed. Content is frozen at 0.9.12; only release-blocking fixes and balance changes may land here.

External playtest and onboarding

• Run at least five first-time-player sessions covering the tutorial and one full venue shift; record where players get stuck, misread evidence or over-escalate.
• Triage by severity and fix every P0/P1 onboarding, evidence or accessibility issue; record the rest as known limitations or post-0.10 work.
• Re-run the tutorial/campaign smokes after each fix and capture a first-hour walkthrough.

Balance and fairness

• Run the scripted venue sweeps across easy/normal/hard and all four phases using the docs/process/release-qa.md#balance-acceptance targets; tune event weights, reaction tables, compliance and report strictness where targets fail.
• Review emergency calls against independent behaviour labels, not _read_npc_truth() as its own oracle; no routine behaviour may make emergency the dominant winning move.
• Verify reputation grades, debrief text and campaign progression still teach de-escalation-first after tuning; record the evidence and deliberate exceptions.

Legal, signing and storefront

• Resolve the macOS signing/notarization decision and record the quarantine instructions.
• Clear or explicitly block on AI-provider commercial rights, storefront/rating approval and the privacy/mental-health representation review.
• Confirm every shipped image, texture, character and audio asset has provenance and credits.

Platform and release engineering

• Smoke-test the target matrix: Linux portable/AppImage, Windows portable/installer, macOS unzip/launch, Web static hosting and the Desktop HQ Vulkan fallback.
• Run tools/build_releases.sh from a clean checkout, verify artifact names, sizes, checksums and download links, and do a PUBLISH_DRY_RUN.
• Run the complete headless suite, Python tooling tests, tutorial/campaign/session/scenario smokes and the 24-NPC budget smoke.
• Confirm no P0/P1 defects remain; document P2 items as known limitations.

Documentation and visual sign-off

• Update README.md, PLAN.md, docs/project/architecture.md, the ADRs, docs/project/development-status.md, docs/process/getting-started.md, docs/process/release-qa.md, CHANGELOG.md and this roadmap.
• Regenerate the gallery and realism baselines from the release-candidate build, not the dev tree; verify every README/docs image resolves from the tagged build.
• Run the manual smoke checklist on every target and link the evidence.
• Release 0.10.0: version, changelog, clean build, checksums, tag, push and post-release record.

Exit: 0.10.0 is content-locked, externally played, legally reviewed, documented, reproducibly built and approved for public download.

Milestone 0.10.1 — Onboarding and accessibility fixes (patch)

• Fix the onboarding, tutorial and evidence-reading issues deferred from the 0.10.0 playtests that are cheap and non-structural.
• Fix accessibility issues found at maximum font scale, in captions or in the colour-blind cue set.
• Re-run the manual smoke checklist on the affected paths and release 0.10.1.

Exit: first-hour comprehension and accessibility defects from the 0.10.0 evidence are closed.

Milestone 0.10.2 — Balance and fairness follow-up (patch)

• Re-run the venue sweeps against the 0.10.0 targets and tune the remaining failing cells.
• Review shift-review explanations, reputation grades and campaign difficulty progression.
• Record the balance evidence and release 0.10.2.

Exit: every balance-acceptance cell is inside target or explicitly accepted with a reason.

Milestone 0.10.3 — Stability and quality-of-life (patch)

• Fix save/checkpoint edge cases, UI focus traps, gamepad remapping and pause/resume defects reported after 0.10.0.
• Add the small quality-of-life items that need no new system (default focus, key hints, settings persistence).
• Release 0.10.3 with the full procedure.

Exit: no known P1 stability defect remains before the 0.11.0 feature work starts.

Milestone 0.11.0 — Systems and content depth (minor)

Goal: add the first post-1.0-track feature set on top of a stable 0.10.x line: a live incident replay, a fourth venue and the localization foundation.

Live body-camera replay

• Replace the text-only incident timeline with a replay of recorded positions, actions and timestamps; define the recording format, memory cap and deterministic playback.
• Keep the replay review-only (no new escalation mechanics) and verify it on Compatibility/Web.
• Add a skip path and a captioned summary so replay is never a hard gate on the shift review.

Fourth venue

• Evaluate and select the fourth real-site venue (candidate: the TCL Chinese Theatre forecourt at 6925 Hollywood Blvd) in an ADR covering layout, access policy, gameplay distinctness and asset budget.
• Build the venue under content/venues/ with licensed reference, sidecar provenance and the three-up board; enter it from venue selection and the campaign.
• Add venue-specific events, access artifacts and escalation cases; extend the balance matrix to the new cells.

Localization foundation

• Extract every remaining hardcoded string into the JSON catalogue and make the pseudo-locale length check a CI gate.
• Add one full secondary language (text, layout, font coverage) with the translation workflow documented; no machine-translated release without review.
• Verify layout at maximum font scale in both languages.

Release

• Re-run the capture gallery, screen-design judge, budget and platform smokes for the new content; release 0.11.0 with the full procedure.

Exit: the replay, the fourth venue and one secondary language ship without regressing the 0.10.x performance, accessibility or balance baselines.

Milestone 0.11.1 — Replay and venue fixes (patch)

• Fix the replay, new-venue and localization defects found in wider play.
• Re-run the new venue's balance cells and the replay capture gate; release 0.11.1.

Exit: no P1 defect remains in the 0.11.0 feature set.

Milestone 0.11.2 — Content and balance polish (patch)

• Tune the new venue events and the campaign pacing; fix the remaining content readability issues.
• Refresh the affected captures, boards and asset rows; release 0.11.2.

Exit: the 0.11.x line is stable enough to freeze its contracts.

Milestone 0.12.0 — Platform and distribution (minor)

Goal: make the game distributable on the launch platforms with signed builds, a frozen save contract and complete localization.

Signing and storefronts

• Sign and notarize the macOS build; record the exact commands and certificate handling.
• Produce the storefront packages and metadata (rating, privacy, credits, screenshots) for the chosen stores; complete the AI-provenance and commercial-rights review.
• Add the storefront-facing extras that are in scope (achievements, cloud-save decision, controller/Steam Deck verification) behind explicit feature flags.

Contracts and hardware

• Freeze the save/checkpoint format as version 2 with export/import and forward-migration tests; document the compatibility promise.
• Verify the target-hardware matrix (minimum GPU, Steam Deck, Web browsers, integrated graphics) and publish the measured results.
• Finish the launch languages and the accessibility statement for the shipped feature set.

Release

• Release 0.12.0 with the full procedure and the storefront artifacts.

Exit: signed, storefront-ready builds exist for every launch platform with a frozen save contract, complete localization and measured hardware support.

Milestone 0.12.1 — Distribution fixes (patch)

• Fix signing, installer, storefront-metadata and save-migration defects found in the storefront dry runs; release 0.12.1.

Exit: every storefront build installs, launches and migrates saves cleanly.

Milestone 0.13.0 — Content complete (minor)

Goal: finish the planned content and lock it for the 1.0.0 release candidate.

Content and presentation

• Finalise all planned venues, events, dialogue and access artifacts; close the content backlog with an explicit accept list.
• Complete recorded audio (VO and music) if the 0.11.0 audio decision funded it; otherwise record the accepted procedural-only direction.
• Finalise the tutorial and first-hour flow against the playtest evidence.

Release candidate preparation

• Run a full accessibility audit against the shipped features and fix the remaining issues.
• Run the balance acceptance protocol one final time and freeze the tuning.
• Declare content and feature freeze; only release-blocking fixes may land after this point.
• Release 0.13.0 as the content-complete milestone.

Exit: all planned content is implemented and frozen, and no known P0/P1 feature gap remains.

Milestone 0.13.1 — Release candidate 1 (patch)

• Build release candidates on every platform from a clean worktree; run the complete smoke matrix, save migration, performance and accessibility checks.
• Fix release-blocking defects only; release 0.13.1.

Exit: a release candidate exists that passes every gate except final legal/storefront approval.

Milestone 0.13.2 — Release candidate 2 (patch)

• Verify the 0.13.1 fixes on the target matrix and re-run the full gallery and judge.
• Freeze the release notes, credits, privacy and known-limitations documents; release 0.13.2 as the final candidate.

Exit: the candidate is build-reproducible, documented and awaiting only the final 1.0.0 release decision.

Milestone 1.0.0 — Release (major)

Goal: publish the first stable release with a documented compatibility promise.

• Confirm legal, privacy, rating and storefront approval is complete in writing.
• Confirm the hardware matrix, accessibility audit and localization sign-off are complete.
• Confirm no P0/P1 defects remain and every P2 is a documented known limitation.
• Update README.md, PLAN.md, docs/project/development-status.md, CHANGELOG.md, CREDITS.md, PRIVACY.md and the release notes for 1.0.0.
• Run the full release procedure: version 1.0.0, changelog, clean build, checksum verification, tag, push and publish.
• Record the post-release state in MEMORY.md and open the 1.0.x maintenance list.

Exit: 1.0.0 is published on every launch platform with complete documentation, verified checksums and no open release blocker.

Future / deferred (not scheduled)

• Seasonal events or a fifth venue after the 0.11.0 fourth venue proves the venue pipeline.
• Real voice/dialogue audio if the 0.13.0 audio decision does not fund it; ambience, cues and radio crackle stay procedural until then.
• Swap the primitive prop builder for an image-to-mesh model API behind the same assets/models/generated/<prop>.glb contract.
• Rare easter eggs/cameos (e.g. Bob Dobbs) only after core content and QA are stable.
• A second Chinese Theatre/forecourt style venue if the fourth venue is not enough.