Merge remote-tracking branch 'origin/main' into capture-harness-redesign

2026-07-25 22:13:49 +02:00
parent a45d4a5f58 ae45d5e21a
commit 28a0b3e2af
24 changed files with 2357 additions and 70 deletions
@@ -95,11 +95,14 @@ Always mirror the category/subcategory path across stages so you can trace
 Final game-ready assets are copied to `spikes/3dpipeline/models/` for spike
 testing. For production, character GLBs go to `client/assets/characters/`
 (already in active use — see its `bodies/`, `clothing/`, `hair/`,
-`skeleton/`, etc. subdirectories). `client/assets/models/` does not exist —
+`skeleton/`, etc. subdirectories). Non-character categories (furniture,
-there is no established production path yet for non-character categories
+props) go to `client/assets/models/{furniture,props}/` — register every
-(furniture, props); that's a Phase-5 concern (Player control & in-world
+promoted asset in `client/assets/models/manifest.json` (schema documented
-rendering, per CLAUDE.md's Development Cascade). Until then, keep
+inline in that file; mirrors the character manifest's per-item pattern).
-non-character output staged under `.tmp/glb-gen/postproc/`.
+Promotion = repo-canonical (D-241): an asset entering the manifest AND
 `client/assets/models/` **is** the promotion act — nothing stays "promoted"
 in `spikes/`. Naming, mask-sidecar, and texture conventions are in
 `docs/assets/visual/conventions.md`.
 ## Post-process in Blender
@@ -26,3 +26,4 @@ INSERT INTO ticket_deps (blocker_record_id, blocked_record_id, created_at, updat
 INSERT INTO ticket_deps (blocker_record_id, blocked_record_id, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FS5XQW73B1P8WKR2BCZBFXZW', '06FS5XW0ZMPVYA2CPBBTP88TQC', '2026-07-24 07:12:41.953', '2026-07-24 07:12:41.953', NULL, '900579ff4a0f777db3fee1b48be50ad2', 2) ON CONFLICT(blocker_record_id, blocked_record_id) DO UPDATE SET updated_at=excluded.updated_at, deleted_at=excluded.deleted_at, hash=excluded.hash, canonical_version=excluded.canonical_version WHERE excluded.updated_at > ticket_deps.updated_at OR (excluded.updated_at = ticket_deps.updated_at AND excluded.hash > ticket_deps.hash);
 INSERT INTO ticket_deps (blocker_record_id, blocked_record_id, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FS5XXY5GNCQEXYAYQEQD0CHM', '06FS5XZF4A978WS1J66Q2JBK8C', '2026-07-24 07:12:42.876', '2026-07-24 07:12:42.876', NULL, '6215bc0bc9966f904af9212895fb2a93', 2) ON CONFLICT(blocker_record_id, blocked_record_id) DO UPDATE SET updated_at=excluded.updated_at, deleted_at=excluded.deleted_at, hash=excluded.hash, canonical_version=excluded.canonical_version WHERE excluded.updated_at > ticket_deps.updated_at OR (excluded.updated_at = ticket_deps.updated_at AND excluded.hash > ticket_deps.hash);
 INSERT INTO ticket_deps (blocker_record_id, blocked_record_id, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FS5XT949C8D2JZRWDX2CRZN8', '06FRXKFEJMDEPQ0PZ5Q862DHKG', '2026-07-24 07:12:43.096', '2026-07-24 07:12:43.096', NULL, '1813bd9c252aa447c9d7ad88c604d6cd', 2) ON CONFLICT(blocker_record_id, blocked_record_id) DO UPDATE SET updated_at=excluded.updated_at, deleted_at=excluded.deleted_at, hash=excluded.hash, canonical_version=excluded.canonical_version WHERE excluded.updated_at > ticket_deps.updated_at OR (excluded.updated_at = ticket_deps.updated_at AND excluded.hash > ticket_deps.hash);
 INSERT INTO ticket_deps (blocker_record_id, blocked_record_id, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FBPTXPE8FR7TGEM7X6QVPV4M', '06FBPTXPPHC9FZ4T0GNKR615PR', '2026-07-25 18:11:40.311', '2026-07-25 18:11:40.311', NULL, '10922d908dd3b4fb42c989bb75391935', 2) ON CONFLICT(blocker_record_id, blocked_record_id) DO UPDATE SET updated_at=excluded.updated_at, deleted_at=excluded.deleted_at, hash=excluded.hash, canonical_version=excluded.canonical_version WHERE excluded.updated_at > ticket_deps.updated_at OR (excluded.updated_at = ticket_deps.updated_at AND excluded.hash > ticket_deps.hash);
@@ -2115,3 +2115,492 @@ Batch-5 Si pass correction (2026-07-25): the 2026-06-16 refinement block above i
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FRJD6EKPBQE08NA8VHP692W8', 'status', 'backlog', 'in_progress', NULL, '2026-07-25 18:04:06', '2026-07-25 18:04:06.736', '2026-07-25 18:04:06.736', NULL, '581321f1ce2faaef8636c8b2f5182716', 2) ON CONFLICT(hash) DO NOTHING;
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FSM3MQ5GFSPKKFJF6GGZN8CM', 'status', 'backlog', 'ready', NULL, '2026-07-25 18:04:07', '2026-07-25 18:04:07.273', '2026-07-25 18:04:07.273', NULL, '56f337047d088a7547c30ce3f4186497', 2) ON CONFLICT(hash) DO NOTHING;
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FRJD6EKPBQE08NA8VHP692W8', 'assigned_to', NULL, 'hoshe', NULL, '2026-07-25 18:04:09', '2026-07-25 18:04:09.374', '2026-07-25 18:04:09.374', NULL, '91edbb31a5c8128989e1fb19f2eedc70', 2) ON CONFLICT(hash) DO NOTHING;
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FBPTXPJ910NNR5T2R9F9NWQR', 'description', '(description follows in first append)
 ---
 Refinement (2026-06-16, /whats-next Si pass) — FULL scope (user decision). Two deliverables + conventions.
 (1) CATALOG: instantiate the 5 docs/assets/visual/*.md files (sprites, furniture, tilesets, icons, effects) with per-category status tables. Template: docs/assets/_templates/sprite.md (SPR/FUR/TILE/ICO/FX prefixes); status vocab planned/in-progress/placeholder/final (docs/assets/README.md). Audio side (docs/assets/audio/README.md) is the working model.
 (2) glb-gen PRODUCTION PATH: SKILL.md:115 names client/assets/models/ which does NOT exist. Create it + furniture/ + props/ subdirs (mirror the spike) + .gitkeep. Define client/assets/models/manifest.json schema MIRRORING the character manifest (docs/architecture/character-asset-organization.md) — per-asset entries: name, category, source-model path, subdirectory. Promotion convention per D-241 (promotion = repo-canonical). Update SKILL.md to point at the new path + reference the manifest.
 (3) CONVENTIONS (author NOW, full scope): door state-frame suffixes (e.g. <model>_open / <model>_closed), Godot TileSet resource organization (<tileset>/<tile_id>.tres), multi-tile footprint metadata, and renderer/textures/ naming keyed to D-235 tokens (current renderer/README.md:80-95 covers sprites only: <model>_<direction>_<resolution>.png). Author these in docs/assets/visual/conventions.md (single home) and cross-link from the SKILL.
 OVERLAP: T-1053 (rewrite araminta briefing) softly depends on this catalog — note the edge. T-1049''s per-class stories will reference these conventions — do T-1050 first / concurrently so T-1049 can cite them.
 ---
 HELD 2026-06-16 (separated from the cascade batch; needs re-scoping before resuming).
 The asset pipeline this ticket assumes is sprite-centric (docs/assets/visual/sprites.md
 "Entities", _templates/sprite.md, SPR/FUR/TILE/ICO/FX prefixes, the /sprite-gen
 render-3D->2D-sprites step). Direction correction: we are NOT producing sprites — Trellis
 produces 3D objects (.glb) displayed in-world; the only flattened 2D content is TEXTURES.
 Re-scope this ticket around a 3D-model + texture catalog (not a sprite catalog) before
 activating. See the sprite-vs-3D root-cause note (this session).
 Batch-5 Si pass correction (2026-07-25): the 2026-06-16 refinement block above is INVALIDATED by the HELD note that follows it — the sprite-centric scope (SPR/FUR prefixes, sprite catalog, /sprite-gen) contradicts the recorded direction (Trellis produces 3D .glb models; only flattened 2D content is textures). Do not implement as written. A re-scope pass (3D-model + texture catalog, manifest mirroring character-asset-organization, D-241 promotion, D-235-keyed texture naming) is running as part of batch 5; T-1052 (style bible) and T-1051 (title-only) get refinement passes in the same sweep. Sequencing per T-1049''s own note stands: this ticket''s conventions land first, T-1049/T-1053 consume them.', '(description follows in first append)
 ---
 Refinement (2026-06-16, /whats-next Si pass) — FULL scope (user decision). Two deliverables + conventions.
 (1) CATALOG: instantiate the 5 docs/assets/visual/*.md files (sprites, furniture, tilesets, icons, effects) with per-category status tables. Template: docs/assets/_templates/sprite.md (SPR/FUR/TILE/ICO/FX prefixes); status vocab planned/in-progress/placeholder/final (docs/assets/README.md). Audio side (docs/assets/audio/README.md) is the working model.
 (2) glb-gen PRODUCTION PATH: SKILL.md:115 names client/assets/models/ which does NOT exist. Create it + furniture/ + props/ subdirs (mirror the spike) + .gitkeep. Define client/assets/models/manifest.json schema MIRRORING the character manifest (docs/architecture/character-asset-organization.md) — per-asset entries: name, category, source-model path, subdirectory. Promotion convention per D-241 (promotion = repo-canonical). Update SKILL.md to point at the new path + reference the manifest.
 (3) CONVENTIONS (author NOW, full scope): door state-frame suffixes (e.g. <model>_open / <model>_closed), Godot TileSet resource organization (<tileset>/<tile_id>.tres), multi-tile footprint metadata, and renderer/textures/ naming keyed to D-235 tokens (current renderer/README.md:80-95 covers sprites only: <model>_<direction>_<resolution>.png). Author these in docs/assets/visual/conventions.md (single home) and cross-link from the SKILL.
 OVERLAP: T-1053 (rewrite araminta briefing) softly depends on this catalog — note the edge. T-1049''s per-class stories will reference these conventions — do T-1050 first / concurrently so T-1049 can cite them.
 ---
 HELD 2026-06-16 (separated from the cascade batch; needs re-scoping before resuming).
 The asset pipeline this ticket assumes is sprite-centric (docs/assets/visual/sprites.md
 "Entities", _templates/sprite.md, SPR/FUR/TILE/ICO/FX prefixes, the /sprite-gen
 render-3D->2D-sprites step). Direction correction: we are NOT producing sprites — Trellis
 produces 3D objects (.glb) displayed in-world; the only flattened 2D content is TEXTURES.
 Re-scope this ticket around a 3D-model + texture catalog (not a sprite catalog) before
 activating. See the sprite-vs-3D root-cause note (this session).
 Batch-5 Si pass correction (2026-07-25): the 2026-06-16 refinement block above is INVALIDATED by the HELD note that follows it — the sprite-centric scope (SPR/FUR prefixes, sprite catalog, /sprite-gen) contradicts the recorded direction (Trellis produces 3D .glb models; only flattened 2D content is textures). Do not implement as written. A re-scope pass (3D-model + texture catalog, manifest mirroring character-asset-organization, D-241 promotion, D-235-keyed texture naming) is running as part of batch 5; T-1052 (style bible) and T-1051 (title-only) get refinement passes in the same sweep. Sequencing per T-1049''s own note stands: this ticket''s conventions land first, T-1049/T-1053 consume them.
 Refinement (batch-5 Si pass, 2026-07-25) — SUPERSEDES the 2026-06-16 block above (sprite-centric, invalidated by D-244). READY.
 CONTEXT: D-244 (confirmed, 2026-06-16) already re-scoped the catalog structure and repointed
 renderer/README.md, sprite-gen SKILL.md, docs/assets/README.md, docs/assets/visual/{README,palette}.md,
 and docs/assets/_templates/sprite.md (marked legacy) at the 3D-model reality. This ticket is now
 narrower than its original draft: instantiate what D-244 named but did not fill in, plus author
 conventions.md (net-new — no D-244 amendment touched it).
 DELIVERABLE (a) — Category catalog files, per docs/assets/README.md''s already-updated structure:
  - docs/assets/visual/models.md   — 3D objects/props (.glb via /glb-gen). Row template TBD (see
    conventions.md below) since _templates/sprite.md is explicitly legacy/retired for this use.
  - docs/assets/visual/textures.md — PBR surface maps (BaseColor/Normal/Roughness) for 3D models.
  - docs/assets/visual/artwork.md  — flat 2D art (paintings/flags/billboards/signage) via /sprite-gen''s
    repurposed 2D-artwork role.
  - docs/assets/visual/icons.md    — UI icons (13 SVGs already exist in client/assets/icons/ — this file
    should reflect that as-built inventory, not start at zero).
  - docs/assets/visual/effects.md  — shaders/particles/overlays.
  Each file follows docs/assets/audio/README.md''s model: per-category status table (planned/in-progress/
  placeholder/final per docs/assets/README.md''s vocab), current counts. Update docs/assets/visual/README.md''s
  category table counts once files exist (currently all show 0, which will stay accurate at initial creation
  — do not invent placeholder rows).
  ACCEPTANCE: 5 files exist, cross-linked from visual/README.md (already has the links — verify they resolve),
  each with a real (possibly empty) status table matching the audio/README.md pattern.
 DELIVERABLE (b) — glb-gen production path:
  - Create client/assets/models/ with furniture/ and props/ subdirectories (mirroring the proven
    spike layout at spikes/3dpipeline/models/{furniture,props}/) + .gitkeep in each.
  - Define client/assets/models/manifest.json, schema mirroring docs/architecture/character-asset-organization.md''s
    per-asset-entry pattern (that doc''s clothing/accessory entries are the closest precedent — a flat
    per-item record, not per-body-type variants since environment props aren''t body-fitted). Minimum
    fields: name, category (furniture|props), source_model (path under client/assets/models/), subdirectory,
    status (planned/in-progress/placeholder/final — reuse the docs/assets/README.md vocab so catalog and
    manifest never disagree). Promotion = repo-canonical per D-241 — an asset entering manifest.json AND
    client/assets/models/ IS the promotion act; nothing stays "promoted" in spikes/.
  - Update .claude/skills/glb-gen/SKILL.md — replace the "client/assets/models/ does not exist... that''s
    a Phase-5 concern" paragraph (lines ~96-102) with the new path + a pointer to the manifest. NOTE:
    this contradicts nothing already in D-244/T-750 — T-750''s own epic description lists furniture/props
    production as Phase-4 parallel work, so populating the production path (not populating the world with
    them — that''s Phase 5 per T-961) is in-scope now.
  ACCEPTANCE: client/assets/models/{furniture,props}/.gitkeep exist; manifest.json schema documented
  (either inline in SKILL.md or as a comment/example in the file itself, following the character
  manifest''s convention); SKILL.md''s stale "does not exist" paragraph is corrected.
 DELIVERABLE (c) — docs/assets/visual/conventions.md (NEW FILE — no prior art to amend):
  - Model naming: <category>_<descriptor>.glb (e.g. table_baroque.glb, chair_modernist.glb) —
    mirrors the spike''s existing baroque_table.glb/modernist_chair.glb pattern loosely; pick one
    order and state it (category-first matches the character-asset clothing convention
    <category>_<descriptor>; the spike used descriptor-first — this is a naming decision to make,
    not silently pick — see lead note below on who rules it).
  - Mask sidecar: <model>_mask.png, single-channel per the character-asset convention (Section 3
    of character-asset-organization.md) UNLESS the item needs multi-region tinting, in which case follow
    the RGBA channel-routed convention from that doc''s Section 11 (T-1089) rather than reinventing one.
  - Texture naming keyed to D-235 tokens: PBR maps for building-exterior surfaces should name by
    the ratified ObjectTag vocabulary (D-235 amendment, wiki/economics/object_tag_vocabulary.toml) —
    e.g. concrete_wall_basecolor.png, pitched_roof_normal.png — NOT the old renderer/README.md
    <model>_<direction>_<resolution>.png scheme, which is direction-indexed for the retired 4-direction
    sprite render and does not apply to PBR texture maps at all (there is no "direction" for a texture,
    only map type: basecolor/normal/roughness).
  - Door state handling — DO NOT port a state-frame suffix convention. The 2026-06-16 draft proposed
    <model>_open/<model>_closed GLB variants; this is sprite-era thinking (baked state = baked frame).
    For a 3D object the door''s open/closed/locked/sealed state (T-979/T-998''s DoorSpec: Open/Closed/
    TemporalWindow/Locked/Sealed) is a RUNTIME transform/animation on one door .glb, not separate baked
    models — but the actual mechanism (AnimationPlayer swing vs. procedural hinge rotation vs. a
    door-model convention with a named "open" bone) is a Phase-5 rendering decision (player control +
    in-world rendering, per the cascade), not a Phase-4 asset-catalog concern. This ticket
    should record ONLY the negative constraint (no state-frame suffix files) and defer the positive
    mechanism to whichever Phase-5 ticket handles door rendering. Do not invent a runtime convention here.
  - Godot TileSet resource organization — OUT OF SCOPE, removed from ticket. client/scripts/rendering/
    tile_renderer.gd''s TileSet is an explicitly-labeled placeholder ("programmatic TileSet with placeholder
    colored rectangles," D-014) and the cascade rule states current in-world rendering code is
    "left as-is until Phase 5 — do not build on it or rip it out before then." A <tileset>/<tile_id>.tres
    convention would be building on Phase-5-owned code from Phase 4. Do not author this section.
  - Multi-tile footprint metadata: still relevant (a table/couch spans multiple sim tiles regardless
    of sprite vs 3D) — keep this item, express as a manifest.json field (e.g. footprint_tiles: [w, h])
    rather than a filename convention.
  ACCEPTANCE: conventions.md exists, covers model naming / mask sidecars / D-235-keyed texture naming /
  multi-tile footprint metadata; explicitly states the door-state and TileSet items are deferred (with
  reasons) rather than silently omitted, so a future refinement pass doesn''t have to re-derive why.
 SEQUENCING (unchanged from 2026-06-16 note): T-1053 (araminta briefing rewrite) softly depends on this
 catalog. T-1049''s per-class stories reference these conventions — this ticket lands first / concurrently.
 Lead note (batch-5 activation): the model-naming-order call (category-first vs descriptor-first) is
 delegated to araminta (art seat) this batch — her ruling lands in conventions.md with a one-line
 rationale; surfaced to Jeroen in the batch report for cheap veto. VERDICT: READY.', NULL, '2026-07-25 18:11:31', '2026-07-25 18:11:31.588', '2026-07-25 18:11:31.588', NULL, '538f0887a4125773b49a88dcdd0b99ff', 2) ON CONFLICT(hash) DO NOTHING;
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FBPTXPV39JX54HYP83RZJFDM', 'description', '(description follows in first append)', '(description follows in first append)
 Refinement (batch-5 Si pass, 2026-07-25) — READY.
 CURRENT STATE: docs/assets/visual/palette.md exists as a STUB (correctly re-scoped by D-244 — no
 sprite-era content survives in it) — 4 bullets: camera angle (D-019, now superseded by D-148 for
 gameplay — palette.md''s bullet is stale and must be corrected), entity colors (D-033), environmental
 neutrality (D-045), "functional warmth." It explicitly defers to "art direction workshop synthesis and
 first visual sprint" — this ticket IS that synthesis pass for Phase 4.
 WHAT D-235 ALREADY FIXES (do not re-litigate): the building-exterior material/roof/facade/street
 vocabulary (WallMaterial/RoofForm/FacadeRhythm/StreetSurface, ratified as the 28-template
 architecture_trait_catalog.toml / wiki/economics/object_tag_vocabulary.toml) and the fallback-to-generic
 mechanism. D-235 is the LOGICAL token layer ("what material is this wall") — Phase 4+ scope per its own
 Implementation note ("textured render + the bulk of the theme library are Phase 5+"). T-1052 is the
 RENDERED-LOOK layer on top: given a WallMaterial token, what does concrete_wall actually look like
 (hue range, texture grain, weathering baseline) when art is eventually produced against it.
 WHAT T-1052 MUST STILL AUTHOR (net-new, no existing decision covers this):
  - Correct the camera-angle bullet: D-148 (30 deg low-angle Camera3D, supersedes D-019) is the GAMEPLAY
    camera; D-019''s -72.5 deg survives only for the offline renderer (2D-artwork / retired sprite mode,
    per renderer/README.md). palette.md''s current bullet conflates these — fix it to name both and
    which applies where.
  - A concrete color/material register for the D-235 token vocabulary: not new tokens, but the RANGE
    each token maps to (e.g. "stone_wall reads as warm grey-tan, not cool grey" / "corrugated_metal
    weathers to rust-orange, not staying silver") — the thing D-235 explicitly parked for Phase 5+ art
    production but which a style bible can pre-specify so future asset generation has a target.
  - Reconcile with docs/design/visual-grammar-v01.md and docs/workshops/art-direction-mood-board/
    workshop-outcomes.md (both pre-cascade, v0.1-era, sprite-framed) — extract what survives (color
    relationships, "functional warmth" identity, D-033 palette) vs what''s superseded by D-235/D-244
    (era-based material ladder — explicitly retired by D-235''s own text: "Era is NOT a material filter
    here"). Do not leave two contradictory style documents live without a note in one pointing at the
    other''s supersession.
  - Texture "grain"/rendering treatment for the D-244 3D-object reality: since objects are lit 3D meshes
    now (not flat sprite fills), the style bible should say something about the toon/PBR balance —
    precedent exists in client/assets/characters/shaders/ (toon.gdshader, toon_masked.gdshader,
    outline.gdshader) and the glb-gen postprocess convention (roughness=1.0/specular=0.0 "for toon
    compatibility" per .claude/skills/glb-gen/SKILL.md''s setup_materials() note). Confirm whether
    environment props share this toon treatment with characters, or diverge — this is a real open
    question the style bible should answer, not assume.
 ACCEPTANCE: docs/assets/visual/palette.md is no longer a stub; camera-angle bullet corrected;
 color/material register for D-235 tokens documented; toon-vs-PBR treatment for props stated;
 supersession relationship to visual-grammar-v01.md / workshop-outcomes.md is explicit (either by
 archiving/marking those superseded, or by palette.md stating exactly what it inherits from each).
 Lead note (batch-5 activation): the toon-vs-PBR treatment call for environment props is delegated to
 araminta (art seat) this batch — her ruling lands in palette.md with rationale; surfaced to Jeroen in
 the batch report for cheap veto. VERDICT: READY.', NULL, '2026-07-25 18:11:34', '2026-07-25 18:11:34.859', '2026-07-25 18:11:34.859', NULL, '3a08af26c8f04bdd4ecef0b4a2fcd115', 2) ON CONFLICT(hash) DO NOTHING;
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FBPTXPPHC9FZ4T0GNKR615PR', 'description', '(description follows in first append)', '(description follows in first append)
 Refinement (batch-5 Si pass, 2026-07-25) — GAPS. Formally deferred rather than refined into an
 actionable brief-writing task now.
 WHY: T-1051 was filed as a child of T-1049''s per-class breakdown (the fable-ous.md audit T-1049 cites,
 2026-06-12). T-1049 itself is HELD pending the same sprite-to-3D re-scope as T-1050 (its "Refinement
 2026-06-16 — READY" block still frames the per-class stories, briefs, and catalog files in sprite-era
 language, and its own text says "sequence T-1049 to consume T-1050''s conventions where they overlap" —
 so T-1049 cannot be un-held before T-1050 lands). T-1051 is one layer further down that same
 dependency: it cannot be meaningfully scoped (what does a "one-page brief" even template against?)
 until (1) T-1050 lands conventions.md + the catalog structure a brief would reference, AND (2) T-1049
 is re-scoped and un-held (it owns the per-class story breakdown T-1051 is part of).
 Zero repo hits exist for lamp posts, barns, TVs, or billboards beyond the T-750 epic description text
 (confirmed by T-1049''s own audit matrix) — there is no existing partial work this ticket would be
 resuming, so there is no cost to deferring it cleanly.
 DISPOSITION: blocker edge wired (T-1051 blocked-by T-1049). When T-1049 is re-scoped (a future Si
 pass, after T-1050 lands), that re-scope explicitly decides whether lamp posts/barns/TVs/billboards
 get real briefs or a formal deferral note (T-1049''s existing block already leaned toward "explicitly
 record which classes are deferred" for cars via Q-067 — the same treatment likely applies to these
 four, but that is T-1049''s call when re-scoped, not this ticket''s in isolation).', NULL, '2026-07-25 18:11:38', '2026-07-25 18:11:38.173', '2026-07-25 18:11:38.173', NULL, '8567ca57e67996983e28229cb470dbd4', 2) ON CONFLICT(hash) DO NOTHING;
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FBPTXPJ910NNR5T2R9F9NWQR', 'status', 'backlog', 'in_progress', NULL, '2026-07-25 18:11:42', '2026-07-25 18:11:42.942', '2026-07-25 18:11:42.942', NULL, '97221c4a7a5aa2ff6e522b450335292d', 2) ON CONFLICT(hash) DO NOTHING;
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FBPTXPV39JX54HYP83RZJFDM', 'status', 'backlog', 'in_progress', NULL, '2026-07-25 18:11:42', '2026-07-25 18:11:42.948', '2026-07-25 18:11:42.948', NULL, '613156e91577043373205250f21a72cd', 2) ON CONFLICT(hash) DO NOTHING;
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FBPTXPJ910NNR5T2R9F9NWQR', 'assigned_to', NULL, 'justine', NULL, '2026-07-25 18:11:45', '2026-07-25 18:11:45.571', '2026-07-25 18:11:45.571', NULL, '7fe3851d234f1cafb2b4fc70c1f5eed8', 2) ON CONFLICT(hash) DO NOTHING;
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FBPTXPV39JX54HYP83RZJFDM', 'assigned_to', NULL, 'araminta', NULL, '2026-07-25 18:11:46', '2026-07-25 18:11:46.095', '2026-07-25 18:11:46.095', NULL, '0d1336d72b0807fe85c054696e914350', 2) ON CONFLICT(hash) DO NOTHING;
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FBPTXPV39JX54HYP83RZJFDM', 'status', 'in_progress', 'review', NULL, '2026-07-25 18:17:34', '2026-07-25 18:17:34.364', '2026-07-25 18:17:34.364', NULL, 'a903761710732a5f62b742cc7ca4acac', 2) ON CONFLICT(hash) DO NOTHING;
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FBPTXPJ910NNR5T2R9F9NWQR', 'status', 'in_progress', 'review', NULL, '2026-07-25 18:19:48', '2026-07-25 18:19:48.053', '2026-07-25 18:19:48.053', NULL, '2fa6ce22c32b396a2ea7ac1dd6f64dc0', 2) ON CONFLICT(hash) DO NOTHING;
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FSHSYM96N2M9GJTQMF58E4WC', 'description', 'Flagged by stig during the T-1189/T-1192 round (2026-07-25): test_step_canvas_viewer.gd has no injected disk-cache-root seam (unlike test_step_canvas_request.gd, which injects a disposable root), so viewer tests that exercise the disk-cache path against a real-looking body_id (e.g. GJ1c) read/write the REAL user://atlas_cache/ shared across runs — a stale real entry from earlier manual sessions can leak into test behavior. Current mitigation (used by the new T-1189/T-1192 tests and the existing sweep smoke test): distinctive synthetic body_ids. Proper fix: give the viewer suite the same injectable cache-root seam the request suite has (constructor/setter injection through StepCanvasRequest), and migrate the synthetic-body_id tests onto it. Small, test-only.', 'Flagged by stig during the T-1189/T-1192 round (2026-07-25): test_step_canvas_viewer.gd has no injected disk-cache-root seam (unlike test_step_canvas_request.gd, which injects a disposable root), so viewer tests that exercise the disk-cache path against a real-looking body_id (e.g. GJ1c) read/write the REAL user://atlas_cache/ shared across runs — a stale real entry from earlier manual sessions can leak into test behavior. Current mitigation (used by the new T-1189/T-1192 tests and the existing sweep smoke test): distinctive synthetic body_ids. Proper fix: give the viewer suite the same injectable cache-root seam the request suite has (constructor/setter injection through StepCanvasRequest), and migrate the synthetic-body_id tests onto it. Small, test-only.
 First slice landed 2026-07-25 (main, 9ac98542d) via a live incident: PR #211''s gate failed on test_step_canvas_viewer''s two before-any-canvas tests because user://atlas_cache/ is machine-shared (every worktree gate, live capture driver, and play session write the same directory) and the T-1183 disk lookup short-circuits before test_mode''s no-op IPC — a concurrent GJ380c Global live capture (T-1157 eyeball work) warmed the exact key the tests assume cold. Fix: StepCanvasViewer.disk_cache_root_override threading into StepCanvasRequest''s existing test-injection seam; viewer + legend suites route construction through _make_viewer() helpers pinning isolated roots. REMAINING for this ticket: screen-constructed viewers (RegionalScreen builds its viewer internally — no override path), and any future suite constructing viewers directly should use the helper pattern; consider a gdUnit-wide guard (e.g. a shared test base or a lint) so new suites cannot silently ride the shared root.', NULL, '2026-07-25 18:45:59', '2026-07-25 18:45:59.847', '2026-07-25 18:45:59.847', NULL, 'dd331004eee53f809c993e6845b52cbd', 2) ON CONFLICT(hash) DO NOTHING;
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FBPTXPJ910NNR5T2R9F9NWQR', 'status', 'review', 'done', NULL, '2026-07-25 18:55:08', '2026-07-25 18:55:08.780', '2026-07-25 18:55:08.780', NULL, '544c6b5e89dafda431434ef0a9525881', 2) ON CONFLICT(hash) DO NOTHING;
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FBPTXPV39JX54HYP83RZJFDM', 'status', 'review', 'done', NULL, '2026-07-25 18:55:08', '2026-07-25 18:55:08.787', '2026-07-25 18:55:08.787', NULL, '1ee2a69f3be3f8cc551a4d9c725b18f5', 2) ON CONFLICT(hash) DO NOTHING;
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FBPTXPE8FR7TGEM7X6QVPV4M', 'description', '(description follows in first append)
 Filed 2026-06-12 from the fable-ous.md audit (S-43; citations adversarially verified).
 T-961 (two-sentence description, no acceptance criteria, no brief links) is the only asset-production ticket under T-750, while the epic names eight classes; cars, barns, TVs, billboards appear in NO ticket at all (verified against the full corpus). Refine into child stories per class — station walls, rural walls, doors, floors, furniture/props, street furniture (lamp posts), signage/screens (billboards/TVs), rural structures (barns) — each gated on a one-page brief; explicitly record which classes are deferred (cars: see Q-067 audit note). Decide Q-119 (generic-parent placeholder set as first milestone) before sequencing.
 Audit per-class status matrix (verified 2026-06-12):
 # docs/assets/ survey (13 files + downloads/)
 | File | One-line summary |
 |---|---|
 | README.md | Master index: audio/visual/video pipelines, status vocabulary (planned/in-progress/placeholder/final), structure that promises visual/{sprites,furniture,tilesets,icons,effects}.md |
 | _templates/audio.md | Row + detailed-entry template for audio assets |
 | _templates/sprite.md | Row + detailed-entry template for sprites (ID prefixes SPR/FUR/TILE/ICO/FX, status, method, directions columns) |
 | _templates/video.md | Row + detailed-entry template for video |
 | audio/README.md | ACTIVE production-status catalog — per-category counts table (12 assets tracked), generation tools, bus architecture |
 | audio/palette.md | Sonic style bible |
 | audio/ambient.md / sfx.md / ui.md | Per-category audio asset tables |
 | visual/README.md | STUB — links 5 category files that do not exist, all counts 0 |
 | visual/palette.md | STUB — 4 bullet decisions (D-019/D-033/D-045, "functional warmth"), defers to art-direction workshop |
 | video/README.md / palette.md | Stubs |
 | downloads/ | Purchased character sources (Quaternius Universal Base Characters, Modular Character Outfits FBX) — characters only |
 # Per-asset-class matrix
 | Class | Written brief | Style guide | Naming/res conventions | Gen path | Produced assets | Tickets |
 |---|---|---|---|---|---|---|
 | Walls (station) | art-direction workshop-outcomes.md §3.3 (v0.1.1 set, pre-cascade) + docs/design/placeholder-art-spec-v01.md §2.1 | visual-grammar-v01.md §1/§4.3, workshop-outcomes §1.8 era palette (partly superseded by D-235), D-235 WallMaterial vocab | renderer/README.md:80-95 (`<model>_<dir>_<res>.png`, 1024→256→64 chain) | working — renderer/models/wall_structural.tscn rendered | wall_structural ×4 dirs in client/assets/sprites/; wall_bar_green model unrendered; 2 textures | T-961 (backlog), T-988 (token vocab, backlog) |
 | Walls (rural) | none found | D-235 tokens only (timber/rammed-earth etc.) | renderer/README.md (generic) | same path, unused | none | T-961 text mention only |
 | Doors | placeholder-art-spec-v01.md §2.1 (procedural rects, v0.1) + workshop-outcomes §3.3 `door_sliding` | none for generated world | none for open/closed state frames | same path, unused | none | T-979/T-998 (DoorSpec — logical only), T-961 |
 | Floors | placeholder-art-spec-v01.md §2.1 + workshop-outcomes §3.3 (4 floor sprites) | D-228 FloorMaterial axis (open vocab) | 64px visual tile (placeholder-art-spec §1) | same path, unused | none — client/assets/tilesets/ is .gitkeep only | T-959 (tile fill, logical), T-961 |
 | Styled buildings | D-232 + D-235 + wiki/economics/architecture_trait_catalog.toml (T-1005 done, ~35 visual bundles) | D-235 material/color-register grammar — tokens only, no rendered art direction | token→fallback-parent chain (catalog header :34-44) | no model/texture path defined for building exteriors | none | T-977, T-988, T-994, T-1004 (all backlog), T-1005 (done) |
 | Furniture | docs/assets/visual/furniture.md MISSING (linked); workshop-outcomes §3.3 lists 8-10 objects (pre-cascade) | era table §1.8 only | character-asset-organization.md explicitly excludes furniture ("follow Trellis spike conventions") | proven in spike | spikes/3dpipeline/models/furniture/ ×3 GLB (unpromoted); zero in client/assets | T-961, T-995 (ObjectTag vocab, backlog) |
 | Items | docs/design/icon-set-v01.md (HUD/inventory icons only) | D-086 icon constraints | SVG icon_ prefix (client/assets/icons/) | n/a for icons | 13 SVGs in client/assets/icons/ | none for in-world item sprites |
 | Cars/vehicles | none found | none | none | spike only | spikes/3dpipeline/models/props/vw_beetle.glb (spike) | none — Q-067 open ("driveable vs NPC traffic?") |
 | Lamp posts | none found (zero repo hits) | none | none | unused | none | T-961 text mention only |
 | Barns | none found (only T-750 epic text) | none | none | unused | none | none |
 | TVs | none found (only T-750 epic text) | none | none | unused | none | none |
 | Billboards | none found (only T-750 epic text; other hits = LOD impostors) | none | none | unused | none | none |
 | Characters (reference class — the working model) | docs/architecture/character-asset-organization.md + docs/design/character-visuals-spec.md | yes | full (manifest.json, seg_ naming, body-type keys) | proven end-to-end | 1002 files in client/assets/characters/ | done |
 # Generation path (verified files)
 image-gen (.claude/skills/image-gen/SKILL.md + scripts/image_connector.py, Gemini) → glb-gen (SKILL.md + tooling/db/trellis_connector.py + tooling/trellis-batch.sh + Blender postprocess scripts) → sprite-gen (SKILL.md + scripts/render.sh) → renderer/ (project.godot, render_scene.tscn, render_export.gd; D-019 camera, 3-light rig, 1024/256/64 chain) → client/assets/sprites/. renderer/output/ holds only .gdignore (intermediates gitignored).
 # Produced assets in client/assets (non-.import files)
 audio 16 · characters 1002 · fonts 3 · icons 13 · planets 7 · sprites 9 (npc_generic ×4, wall_structural ×4, .gitkeep) · theme 1 · tilesets 1 (.gitkeep only)
 ---
 Refinement (2026-06-16, /whats-next Si pass) — READY. This is a PLANNING task: break T-961 into per-class child stories.
 CLASS LIST (8 stories, parented under T-961, each gated on a one-page brief): station walls, rural walls, doors, floors, furniture/props, street furniture (lamp posts), signage/screens (TVs/billboards), rural structures (barns).
 - CARS (9th class named in T-750): DEFERRED — Q-067 needs a phase decision (Phase-4 deliverable vs Phase-5+) before any car story. File a deferral note, not a story.
 - Q-119 (generic-parent placeholder set as milestone-0): NOT a blocker — sequencing only. On each relevant story note that the generic parents (generic_wall/roof/facade/street + generic_door/floor/furniture; D-235 / wiki/economics/architecture_trait_catalog.toml:33-41) are milestone-0 IF Q-119 resolves "yes".
 BRIEF TEMPLATE: none exists. Create docs/assets/_templates/asset-brief.md (scope · visual reference · naming/resolution conventions · generation path · acceptance), or adopt docs/_templates/design.md. Each story states where its brief + catalog file live under docs/assets/visual/.
 READ before filing: docs/assets/README.md, docs/design/{placeholder-art-spec-v01,visual-grammar-v01}.md, docs/workshops/art-direction-mood-board/workshop-outcomes.md, D-235/D-228/D-232, docs/architecture/character-asset-organization.md (the reference model).
 NOTE: T-1050 (sibling) owns the catalog + naming conventions these stories reference — sequence T-1049 to consume T-1050''s conventions where they overlap.
 ---
 HELD 2026-06-16 (separated from the cascade batch; needs re-scoping before resuming).
 The asset pipeline this ticket assumes is sprite-centric (docs/assets/visual/sprites.md
 "Entities", _templates/sprite.md, SPR/FUR/TILE/ICO/FX prefixes, the /sprite-gen
 render-3D->2D-sprites step). Direction correction: we are NOT producing sprites — Trellis
 produces 3D objects (.glb) displayed in-world; the only flattened 2D content is TEXTURES.
 Re-scope this ticket around a 3D-model + texture catalog (not a sprite catalog) before
 activating. See the sprite-vs-3D root-cause note (this session).', '(description follows in first append)
 Filed 2026-06-12 from the fable-ous.md audit (S-43; citations adversarially verified).
 T-961 (two-sentence description, no acceptance criteria, no brief links) is the only asset-production ticket under T-750, while the epic names eight classes; cars, barns, TVs, billboards appear in NO ticket at all (verified against the full corpus). Refine into child stories per class — station walls, rural walls, doors, floors, furniture/props, street furniture (lamp posts), signage/screens (billboards/TVs), rural structures (barns) — each gated on a one-page brief; explicitly record which classes are deferred (cars: see Q-067 audit note). Decide Q-119 (generic-parent placeholder set as first milestone) before sequencing.
 Audit per-class status matrix (verified 2026-06-12):
 # docs/assets/ survey (13 files + downloads/)
 | File | One-line summary |
 |---|---|
 | README.md | Master index: audio/visual/video pipelines, status vocabulary (planned/in-progress/placeholder/final), structure that promises visual/{sprites,furniture,tilesets,icons,effects}.md |
 | _templates/audio.md | Row + detailed-entry template for audio assets |
 | _templates/sprite.md | Row + detailed-entry template for sprites (ID prefixes SPR/FUR/TILE/ICO/FX, status, method, directions columns) |
 | _templates/video.md | Row + detailed-entry template for video |
 | audio/README.md | ACTIVE production-status catalog — per-category counts table (12 assets tracked), generation tools, bus architecture |
 | audio/palette.md | Sonic style bible |
 | audio/ambient.md / sfx.md / ui.md | Per-category audio asset tables |
 | visual/README.md | STUB — links 5 category files that do not exist, all counts 0 |
 | visual/palette.md | STUB — 4 bullet decisions (D-019/D-033/D-045, "functional warmth"), defers to art-direction workshop |
 | video/README.md / palette.md | Stubs |
 | downloads/ | Purchased character sources (Quaternius Universal Base Characters, Modular Character Outfits FBX) — characters only |
 # Per-asset-class matrix
 | Class | Written brief | Style guide | Naming/res conventions | Gen path | Produced assets | Tickets |
 |---|---|---|---|---|---|---|
 | Walls (station) | art-direction workshop-outcomes.md §3.3 (v0.1.1 set, pre-cascade) + docs/design/placeholder-art-spec-v01.md §2.1 | visual-grammar-v01.md §1/§4.3, workshop-outcomes §1.8 era palette (partly superseded by D-235), D-235 WallMaterial vocab | renderer/README.md:80-95 (`<model>_<dir>_<res>.png`, 1024→256→64 chain) | working — renderer/models/wall_structural.tscn rendered | wall_structural ×4 dirs in client/assets/sprites/; wall_bar_green model unrendered; 2 textures | T-961 (backlog), T-988 (token vocab, backlog) |
 | Walls (rural) | none found | D-235 tokens only (timber/rammed-earth etc.) | renderer/README.md (generic) | same path, unused | none | T-961 text mention only |
 | Doors | placeholder-art-spec-v01.md §2.1 (procedural rects, v0.1) + workshop-outcomes §3.3 `door_sliding` | none for generated world | none for open/closed state frames | same path, unused | none | T-979/T-998 (DoorSpec — logical only), T-961 |
 | Floors | placeholder-art-spec-v01.md §2.1 + workshop-outcomes §3.3 (4 floor sprites) | D-228 FloorMaterial axis (open vocab) | 64px visual tile (placeholder-art-spec §1) | same path, unused | none — client/assets/tilesets/ is .gitkeep only | T-959 (tile fill, logical), T-961 |
 | Styled buildings | D-232 + D-235 + wiki/economics/architecture_trait_catalog.toml (T-1005 done, ~35 visual bundles) | D-235 material/color-register grammar — tokens only, no rendered art direction | token→fallback-parent chain (catalog header :34-44) | no model/texture path defined for building exteriors | none | T-977, T-988, T-994, T-1004 (all backlog), T-1005 (done) |
 | Furniture | docs/assets/visual/furniture.md MISSING (linked); workshop-outcomes §3.3 lists 8-10 objects (pre-cascade) | era table §1.8 only | character-asset-organization.md explicitly excludes furniture ("follow Trellis spike conventions") | proven in spike | spikes/3dpipeline/models/furniture/ ×3 GLB (unpromoted); zero in client/assets | T-961, T-995 (ObjectTag vocab, backlog) |
 | Items | docs/design/icon-set-v01.md (HUD/inventory icons only) | D-086 icon constraints | SVG icon_ prefix (client/assets/icons/) | n/a for icons | 13 SVGs in client/assets/icons/ | none for in-world item sprites |
 | Cars/vehicles | none found | none | none | spike only | spikes/3dpipeline/models/props/vw_beetle.glb (spike) | none — Q-067 open ("driveable vs NPC traffic?") |
 | Lamp posts | none found (zero repo hits) | none | none | unused | none | T-961 text mention only |
 | Barns | none found (only T-750 epic text) | none | none | unused | none | none |
 | TVs | none found (only T-750 epic text) | none | none | unused | none | none |
 | Billboards | none found (only T-750 epic text; other hits = LOD impostors) | none | none | unused | none | none |
 | Characters (reference class — the working model) | docs/architecture/character-asset-organization.md + docs/design/character-visuals-spec.md | yes | full (manifest.json, seg_ naming, body-type keys) | proven end-to-end | 1002 files in client/assets/characters/ | done |
 # Generation path (verified files)
 image-gen (.claude/skills/image-gen/SKILL.md + scripts/image_connector.py, Gemini) → glb-gen (SKILL.md + tooling/db/trellis_connector.py + tooling/trellis-batch.sh + Blender postprocess scripts) → sprite-gen (SKILL.md + scripts/render.sh) → renderer/ (project.godot, render_scene.tscn, render_export.gd; D-019 camera, 3-light rig, 1024/256/64 chain) → client/assets/sprites/. renderer/output/ holds only .gdignore (intermediates gitignored).
 # Produced assets in client/assets (non-.import files)
 audio 16 · characters 1002 · fonts 3 · icons 13 · planets 7 · sprites 9 (npc_generic ×4, wall_structural ×4, .gitkeep) · theme 1 · tilesets 1 (.gitkeep only)
 ---
 Refinement (2026-06-16, /whats-next Si pass) — READY. This is a PLANNING task: break T-961 into per-class child stories.
 CLASS LIST (8 stories, parented under T-961, each gated on a one-page brief): station walls, rural walls, doors, floors, furniture/props, street furniture (lamp posts), signage/screens (TVs/billboards), rural structures (barns).
 - CARS (9th class named in T-750): DEFERRED — Q-067 needs a phase decision (Phase-4 deliverable vs Phase-5+) before any car story. File a deferral note, not a story.
 - Q-119 (generic-parent placeholder set as milestone-0): NOT a blocker — sequencing only. On each relevant story note that the generic parents (generic_wall/roof/facade/street + generic_door/floor/furniture; D-235 / wiki/economics/architecture_trait_catalog.toml:33-41) are milestone-0 IF Q-119 resolves "yes".
 BRIEF TEMPLATE: none exists. Create docs/assets/_templates/asset-brief.md (scope · visual reference · naming/resolution conventions · generation path · acceptance), or adopt docs/_templates/design.md. Each story states where its brief + catalog file live under docs/assets/visual/.
 READ before filing: docs/assets/README.md, docs/design/{placeholder-art-spec-v01,visual-grammar-v01}.md, docs/workshops/art-direction-mood-board/workshop-outcomes.md, D-235/D-228/D-232, docs/architecture/character-asset-organization.md (the reference model).
 NOTE: T-1050 (sibling) owns the catalog + naming conventions these stories reference — sequence T-1049 to consume T-1050''s conventions where they overlap.
 ---
 HELD 2026-06-16 (separated from the cascade batch; needs re-scoping before resuming).
 The asset pipeline this ticket assumes is sprite-centric (docs/assets/visual/sprites.md
 "Entities", _templates/sprite.md, SPR/FUR/TILE/ICO/FX prefixes, the /sprite-gen
 render-3D->2D-sprites step). Direction correction: we are NOT producing sprites — Trellis
 produces 3D objects (.glb) displayed in-world; the only flattened 2D content is TEXTURES.
 Re-scope this ticket around a 3D-model + texture catalog (not a sprite catalog) before
 activating. See the sprite-vs-3D root-cause note (this session).
 Refinement (batch-6 Si pass, 2026-07-25) — SUPERSEDES the 2026-06-16 READY block
 (sprite-era, invalidated by D-244) and the 2026-06-16 HELD note. T-1050 (catalog +
 conventions.md + manifest.json) and T-1052 (palette.md style bible) are both DONE
 (PR #211, merged 2026-07-25) — the conventions this ticket needed to consume now
 exist and are read, not hypothetical.
 SCOPE, 3D REALITY: T-961 breaks into per-asset-class stories, each gated on a
 one-page brief. A "written brief" is now a short doc that states, per class:
 (1) scope — which manifest category (furniture|props) and which D-235 ObjectTag
 axis if any; (2) visual reference — cites palette.md §3''s hue/grain/weathering
 register for the relevant token(s), or palette.md §2 toon/PBR default for classes
 with no D-235 token; (3) naming — conventions.md §1 model naming
 (<category>_<descriptor>.glb), §2 mask sidecar if tintable, §4 footprint_tiles if
 multi-tile; (4) generation path — /image-gen concept → /glb-gen (Trellis) →
 postprocess → promotion into client/assets/models/{furniture,props}/ +
 manifest.json entry (conventions.md, models.md); (5) acceptance — N assets at
 "planned" status registered in the manifest with a real manifest.json entry,
 status flips as production proceeds (models.md''s shared status vocab).
 BRIEF TEMPLATE (create docs/assets/_templates/asset-brief.md): the five headers
 above (scope / visual reference / naming / gen path / acceptance), one page,
 adopting docs/_templates/design.md frontmatter. Do NOT reuse _templates/sprite.md
 (SPR/FUR/TILE/ICO/FX prefixes) — explicitly retired by D-244/T-1050.
 CLASS LIST — station walls, rural walls, doors, floors, furniture/props already
 have partial prior art (station walls: wall_structural GLB exists unrendered per
 the 2026-06-12 audit matrix; doors: DoorSpec logical model exists, T-979/T-998;
 floors: D-228 FloorMaterial axis exists) — each gets a real story + brief, citing
 its relevant D-235 axis (wall/roof/facade/street) or, for doors/floors/furniture,
 the non-ObjectTag conventions (mask sidecars, footprint_tiles).
 STREET FURNITURE / SIGNAGE-SCREENS / RURAL STRUCTURES (lamp posts, barns,
 TVs, billboards — T-1051''s four classes): RECOMMEND FORMAL DEFERRAL, not real
 briefs, this batch. Rationale: Phase 4''s deliverable is "deterministic
 walkable-world data + per-layer Atlas maps + asset catalog" (cascade table) —
 outside-in generation (D-166) needs building EXTERIOR templates
 (D-235 wall/roof/facade/street axes) to make districts/blocks walkable and
 Atlas-legible. Lamp posts (street furniture), TVs/billboards (signage/screens
 placed as interior or exterior set-dressing), and barns (a rural building
 *type*, not a wall/roof/facade axis token) are furniture-tier detail objects
 that populate a walkable exterior once it exists — they read as Phase 6
 "detail coloring / room-level population" (cascade table) or, at best, late
 Phase 4 polish, not blocking asset classes for getting the world walkable.
 Zero repo hits exist for any of the four beyond epic text (T-1049''s own 2026-06-12
 audit matrix, unchanged) — there is no partial work to protect by deferring.
 Treat identically to Q-067''s car deferral: file a deferral note on the story
 (or fold into one placeholder story "T-1051 classes — deferred pending Phase 4
 exterior-generation completion"), do not write real briefs now. Distinguish from
 cars only in degree, not in kind — Q-067 is a still-open decision question
 (vehicle simulation is a bigger unresolved design surface); the T-1051 four are
 simpler set-dressing whose deferral needs no open question, just a scope note,
 because nothing about their eventual brief depends on an undecided mechanic.
 Q-119 (generic-parent placeholder set as milestone-0) is STILL OPEN (unresolved
 as of this pass) — unchanged from the prior block: not a blocker, sequencing-only.
 Each relevant story notes the generic_wall/roof/facade/street (+ generic
 door/floor/furniture) parents as milestone-0 IF Q-119 resolves yes.
 READ before filing: docs/assets/visual/{conventions,palette,models,textures,
 artwork}.md (T-1050/T-1052, merged), D-244, D-257, D-235 (ratified token list),
 D-243 (footprint unit), docs/architecture/character-asset-organization.md
 (mask/manifest precedent).
 VERDICT: READY (activation = batch 6; the T-1051 deferral recommendation above is
 ratified at activation, then T-1051 is updated to match the ruling this ticket records).', NULL, '2026-07-25 19:00:33', '2026-07-25 19:00:33.469', '2026-07-25 19:00:33.469', NULL, 'f74a4367cd0a4eb2664f4d575e7d31c2', 2) ON CONFLICT(hash) DO NOTHING;
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FBPTXPZ0ZTM1QQXD5K972CMM', 'description', '(description follows in first append)', '(description follows in first append)
 Refinement (batch-6 Si pass, 2026-07-25) — READY.
 WHAT''S WRONG WITH THE CURRENT BRIEFING: frozen at the 2026-03-13 v0.2 pivot.
 (1) Frames the role around "v0.2 Pivot — Key Changes" — v0.2 is formally dropped
 ("No scoping negotiations. Build the base systems fully"), so the framing section
 is obsolete, not just outdated. (2) Assumes the sprite-era pipeline throughout
 ("Phase 1-3 placeholder art pipeline," "sprite replacements" in Phase 4) —
 contradicted by D-244 (3D objects in-world, no per-object sprite layer) and D-149
 (live 3D characters). (3) Cites D-019 as simply "top-down art style" with no note
 that D-148 superseded it for gameplay (palette.md §1 had to correct exactly this
 conflation in its own stub — the briefing repeats the same error). (4) Zone visual
 grammar / tile palette framing predates D-235''s ObjectTag vocabulary and D-232''s
 trait-template catalog entirely. (5) Character customisation section (D-126,
 Q-WTF-039/040) predates the shipped D-159–D-164 body-type + slot-compositing
 system and the wardrobe/garment-token pipeline — describes a design question that
 has since been built. (6) No mention of the Atlas at all (D-255 stepped ladder,
 D-169/D-170 implant UI) despite araminta being a named co-designer of D-255''s
 named-feature encoding/payload schema/lakes work.
 WHAT THE REWRITE MUST CONTAIN:
 - Role framing: visual art direction for the Phase-4 asset pipeline + Atlas
  map-art rendering (not "generator spike prerequisites").
 - palette.md (T-1052, merged) as the SINGLE LIVE STYLE AUTHORITY — point to it,
  do not restate camera/color/material content (avoids a second drifting copy;
  visual-grammar-v01.md / workshop-outcomes.md supersession is palette.md §5''s
  to state, the briefing defers).
 - conventions.md (T-1050, merged) for naming/mask/manifest mechanics — note the
  category-first naming ruling (conventions.md §1) and the D-257 toon/PBR ruling
  with glazing carve-out as HER standing rulings she''ll be asked to extend.
 - D-235 ObjectTag vocabulary as the logical layer she renders a look on top of —
  co-maintained with Miri per D-235''s own text.
 - The character wardrobe pipeline (character-asset-organization.md) as the proven
  reference model conventions.md imitates (mask sidecars, manifest schema).
 - D-255 stepped Atlas as active/ongoing work she co-designed.
 - Current open items: T-1049 (per-class asset stories, re-scoped this batch), the
  D-257 glazing shader implementation (unassigned per palette.md §2.3), Q-119
  (generic-parent milestone question, hers to weigh in on per D-235
  co-maintenance).
 - Drop: v0.2 pivot section, Q-WTF-039/040, tycoon-bookmark/apartment-wakeup
  framing, and the v0.2-era decision list (D-114/D-117/D-119/D-122/D-128/D-131/
  D-135) — verify each against current decision status before dropping outright;
  a few (D-033 entity color, D-126 customisation dimensions) may still have live
  sub-parts even though the surrounding section is obsolete.
 ACCEPTANCE: rewritten briefing reflects the current Phase-4 landscape (D-235,
 D-244, D-257, D-255), points to palette.md/conventions.md as live authorities
 rather than duplicating content, drops v0.2-pivot and superseded-question
 framing, lists araminta''s actual current/open tickets.
 SEQUENCING: independent of T-1049 (different files, no shared input) — both can
 run in the same batch.', NULL, '2026-07-25 19:00:36', '2026-07-25 19:00:36.464', '2026-07-25 19:00:36.464', NULL, 'ebe504fa1213997aad9ae74f82212f6f', 2) ON CONFLICT(hash) DO NOTHING;
 INSERT INTO ticket_history (ticket_record_id, field, old_value, new_value, changed_by, changed_at, created_at, updated_at, deleted_at, hash, canonical_version) VALUES ('06FSM3MQ5GFSPKKFJF6GGZN8CM', 'description', 'Araminta''s PR #192 companion note, kept open by her PR #208 ruling (option b): the Quarter-vs-District relief-differentiation check was accepted for the T-1160 merge on a plains/river substitution (GJ380c world center -17723392,-10532864 — Quarter clearly shows finer higher-frequency texture than District), but plains texture and jagged tectonic relief are different visual regimes and the sharper case still needs its own capture. Task: identify the body pool with tectonics:high (or planet_class values mapping to TectonicClass::Volcanic/Active) and RECORD the pool in this ticket first — the PR #208 eyeball proved GJ380c/GJ820Bc are both tectonics:low via exhaustive two-Global color scans, don''t re-run that search blind — then capture quarter-relief vs district-relief at a true Alpine/MountainPass crop (MorphologyZone 13/14) at the same world center, matched crop, and have Araminta judge the differentiation. Reuse the pr208_driver fixed-center --step protocol (T-1157 harness inventory). Related: T-1160, PR #192, PR #208.', 'Araminta''s PR #192 companion note, kept open by her PR #208 ruling (option b): the Quarter-vs-District relief-differentiation check was accepted for the T-1160 merge on a plains/river substitution (GJ380c world center -17723392,-10532864 — Quarter clearly shows finer higher-frequency texture than District), but plains texture and jagged tectonic relief are different visual regimes and the sharper case still needs its own capture. Task: identify the body pool with tectonics:high (or planet_class values mapping to TectonicClass::Volcanic/Active) and RECORD the pool in this ticket first — the PR #208 eyeball proved GJ380c/GJ820Bc are both tectonics:low via exhaustive two-Global color scans, don''t re-run that search blind — then capture quarter-relief vs district-relief at a true Alpine/MountainPass crop (MorphologyZone 13/14) at the same world center, matched crop, and have Araminta judge the differentiation. Reuse the pr208_driver fixed-center --step protocol (T-1157 harness inventory). Related: T-1160, PR #192, PR #208.
 POOL RECORDED (2026-07-25, read-only survey per this ticket''s own first step —
 do not re-run the search blind).
 MAPPING MECHANISM: server/src/atlas/district_profile.rs:296-312
 derive_tectonic_class(): BodyParams::tectonic_activity authored override first
 ("volcanic"/"active"/"tidally_forced", NO column in current systems.db schema and
 no authored overrides in the wiki TOMLs — so today the override path is dead),
 else planet_class fallback: "volcanic" → Volcanic; "oceanic"/"ocean_world" →
 Active; everything else → Stable. Sanity anchors agree with the PR #208 eyeball:
 GJ380c (arid) and GJ820Bc (arid) both → Stable.
 THE FULL POOL (7 bodies):
 - Volcanic: GJ0c (Venus, no heightmap), GJ0f-1 (Io, no heightmap), GJ4056b
  (heightmap, airless, relief_energy_q=1.0), GJ660Ad (Seongho Prime, heightmap,
  thin atmosphere, relief_energy_q=1.0)
 - Active: GJ1002b (Caparica, heightmap, island world, standard atmosphere),
  GJ103c (Vlieland, heightmap, island world, standard atmosphere), GJ447c
  (Marevna, heightmap, ocean world) — all relief_energy_q=0.70 per D-239 §5.
 CAPTURE RECOMMENDATION (lead): for the true Alpine/MountainPass (MorphologyZone
 13/14) crop this ticket wants, the Active terrestrial island worlds GJ1002b and
 GJ103c are the primary candidates (coastal orogeny, standard climate — closest
 analogue to the GJ380c plains/river substitution being replaced). The Volcanic
 pair (GJ4056b, GJ660Ad) are alternates with maximum relief energy but
 airless/thin atmospheres — shield/lava-field morphology, likely NOT
 Alpine-classed; use only if the Active pair fails to produce MorphologyZone
 13/14 crops. GJ0c/GJ0f-1 lack terrain_reference and cannot open in the Atlas.', NULL, '2026-07-25 19:00:39', '2026-07-25 19:00:39.731', '2026-07-25 19:00:39.731', NULL, '194ab5efa1ff3b059c0296830334f801', 2) ON CONFLICT(hash) DO NOTHING;
@@ -7,6 +7,8 @@ Format based on [Keep a Changelog](https://keepachangelog.com/).
 ## [Unreleased]
 ### Added
 - **The visual asset pipeline has a production home** (D-244, T-1050) — the 3D-model/texture/artwork/icon/effect catalogs are instantiated with live status tables (the icon set's 13 as-built SVGs inventoried with their real wiring), `client/assets/models/` exists as the promotion target for glb-gen output with a schema-documented manifest, and a single conventions document fixes model naming (category-first), mask sidecars, and D-235-token texture naming — with door-state and TileSet conventions recorded as explicit Phase-5 deferrals rather than silent gaps
 - **The Phase-4 visual style bible** (D-257, T-1052) — palette.md is no longer a stub: gameplay vs offline-render cameras untangled, a rendered-look register for every ratified building material/roof/facade/street tag (hue range, grain, weathering), and the newly-recorded D-257 rendering treatment — environment props share the character toon look, with a minimal-PBR carve-out for glazing so glass reads truthfully about sightlines. Exactly one live style authority remains; the superseded v0.1 documents now say so themselves
 - **The Atlas can be driven by agents** (D-226, T-971) — a named-intent control channel (AtlasAgentInterface) lets automated QA agents navigate the real Atlas UI through the exact same handlers a click calls: open/select/descend, jump straight to any world coordinate at any zoom step, toggle overlays, and read back a summary of what's on screen — turning the hand-scripted verification drives into a first-class, repeatable QA capability. In-process consumers only for now; a shipped reference driver replaces the ad-hoc scripts
 ### Changed
@@ -15,6 +17,7 @@ Format based on [Keep a Changelog](https://keepachangelog.com/).
 - **Internal: legacy map-window field retired** (T-1159) — the obsolete duplicate zoom-granularity field was removed from the map wire protocol and caches; no player-visible change
 ### Fixed
 - **Map viewer unit tests no longer flake against a warm map cache** (T-1193 first slice) — the viewer test suites rode the machine-shared on-disk atlas cache, so a live play or capture session visiting the same body could flip "nothing has arrived yet" tests in any later test run; viewers under test now pin an isolated cache root through a proper injection seam
 - **One coastline at every zoom step** (T-1160) — the whole-body and Region views drew a structurally different coastline than the closer zoom steps (they skipped the deterministic coast-detail warp — about 5% of coastal cells disagreed, displaced up to ~19 km). All six zoom steps now draw the same coastline at their own density: zooming in refines the line instead of moving it. (Game version → 0.4.3: old on-disk map caches refresh automatically)
 - **Climate now respects latitude on both hemispheres** (T-1186) — a sign-convention bug made every northern hemisphere read as polar (+90°) and the southern hemisphere read as compressed northern tropics: temperate lakes at 76°S, ice only at the top of the world map. Temperature baselines, glaciation, and everything derived from them now band correctly north AND south of the equator — the same spot near the antarctic circle that used to derive as an 18 °C ice-free lake now comes out at 2 °C under light glaciation
 - **Lakes deepen from the shore** (T-1188) — lake shorelines no longer render as hard flat-blue step-edges: every lake cell now carries its settled-hydrology depth band, and the map shades lakes from pale at the water's edge to dark at the deepest point of the basin — the same visual grammar ocean coastlines already had — each lake shaded relative to its own deepest point, so shallow ponds and deep basins both read correctly. (Game version 0.4.0 → 0.4.2: old on-disk map caches refresh automatically)
@@ -0,0 +1,24 @@
 {
  "_comment": "Environment prop/furniture manifest — schema mirrors client/assets/characters/manifest.json's per-item entry pattern (see docs/architecture/character-asset-organization.md). Keyed by asset name (matches the .glb basename, no extension). One entry per asset, NOT per-body-type variant — environment props are not body-fitted (D-244). An asset entering this manifest AND client/assets/models/ IS the promotion act (D-241); nothing stays 'promoted' in spikes/. Naming/mask-sidecar/texture conventions: docs/assets/visual/conventions.md. CONTRACT: any top-level key prefixed with '_' (this key, '_fields', '_example_only_not_a_real_asset') is metadata/documentation, not an asset category — loaders MUST skip '_'-prefixed keys and must not iterate them as if they were real category/asset entries.",
  "_fields": {
    "name": "string — asset name, matches the manifest key and the .glb basename",
    "category": "string — 'furniture' | 'props' (matches the subdirectory)",
    "source_model": "string — path to the .glb under client/assets/models/, e.g. 'furniture/table_baroque.glb'",
    "subdirectory": "string — 'furniture' | 'props' (same as category; kept as its own field to match the character-manifest precedent of not deriving one field from another)",
    "status": "string — planned | in-progress | placeholder | final (docs/assets/README.md vocab — shared with the catalog tables so manifest and catalog never disagree)",
    "footprint_tiles": "[width, height] — multi-tile sim footprint in whole tiles (1 tile = 1m, D-243). Omit (or [1, 1]) for single-tile props.",
    "mask": "string, optional — path to the recolor mask sidecar, e.g. 'furniture/table_baroque_mask.png'. Omit if the asset has no tintable region."
  },
  "_example_only_not_a_real_asset": {
    "table_baroque": {
      "category": "furniture",
      "source_model": "furniture/table_baroque.glb",
      "subdirectory": "furniture",
      "status": "planned",
      "footprint_tiles": [2, 1],
      "mask": "furniture/table_baroque_mask.png"
    }
  },
  "furniture": {},
  "props": {}
 }
@@ -9,9 +9,20 @@ extends GdUnitTestSuite
 const LegendScript := preload("res://ui/implant/apps/atlas/step_canvas/step_canvas_legend.gd")
 const StepCanvasTransport := preload("res://ui/implant/apps/atlas/step_canvas/step_canvas_transport.gd")
 ## Same Tier-2/3 isolation as test_step_canvas_viewer.gd's _make_viewer()
 ## (T-1193 first slice): the default disk-cache root is machine-shared, and a
 ## warm cache delivers canvases into tests entered on real body ids.
 const TEST_DISK_CACHE_ROOT := "user://test_step_canvas_legend_cache/"
 func _make_viewer() -> StepCanvasViewer:
 	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
 	v.disk_cache_root_override = TEST_DISK_CACHE_ROOT
 	return v
 func test_legend_starts_hidden_before_refresh() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	var legend = LegendScript.new(v)
 	auto_free(legend)
@@ -19,7 +30,7 @@ func test_legend_starts_hidden_before_refresh() -> void:
 func test_refresh_makes_the_legend_visible_and_populates_content() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
 	var legend = LegendScript.new(v)
@@ -45,7 +56,7 @@ func test_refresh_without_a_viewer_does_not_crash() -> void:
 ## default, one section per active toggle" contract the retired
 ## atlas_window_legend.gd's own refresh() doc described.
 func test_refresh_reflects_the_active_overlay_toggle() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
 	var legend = LegendScript.new(v)
@@ -63,7 +74,7 @@ func test_refresh_reflects_the_active_overlay_toggle() -> void:
 func test_reposition_sets_a_fixed_panel_margin_position() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	var legend = LegendScript.new(v)
 	auto_free(legend)
@@ -81,7 +92,7 @@ func test_reposition_sets_a_fixed_panel_margin_position() -> void:
 ## column, with room to spare — never right up against it, and certainly
 ## never past it into where the canvas is centered from.
 func test_legend_actual_right_edge_stays_inside_the_reserved_column() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	var legend = LegendScript.new(v)
 	auto_free(legend)
@@ -16,16 +16,34 @@ const StepCanvasTransport := preload("res://ui/implant/apps/atlas/step_canvas/st
 ## .cache/screenshots/t1183-eyeball-run2/02-region.png.
 const GJ1C_GLOBAL_EXTENT := Vector2i(177, 88)
 ## Isolated Tier-2/3 disk-cache root for every directly-constructed viewer
 ## (T-1193 first slice). The default user://atlas_cache/ is SHARED across the
 ## whole machine — every worktree's gate run, live capture driver, and real
 ## play session writes the same directory — and T-1183's disk-cache lookup
 ## short-circuits BEFORE test_mode's silent-no-op IPC, so a warm shared cache
 ## delivers real canvases into tests written against "nothing ever arrives"
 ## (2026-07-25: a concurrent live GJ380c Global capture flipped the two
 ## before-any-canvas tests in another worktree's gate run). In test_mode the
 ## isolated root stays empty forever: no canvas ever arrives, so the cache
 ## never writes.
 const TEST_DISK_CACHE_ROOT := "user://test_step_canvas_viewer_cache/"
 func _make_viewer() -> StepCanvasViewer:
 	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
 	v.disk_cache_root_override = TEST_DISK_CACHE_ROOT
 	return v
 func test_enter_lands_on_the_global_opener() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
 	assert_str(v.get_held_rung()).is_equal(StepCanvasTransport.RUNG_GLOBAL)
 func test_get_body_id_reflects_the_entered_body() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	assert_str(v.get_body_id()).is_equal("")
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
@@ -35,7 +53,7 @@ func test_get_body_id_reflects_the_entered_body() -> void:
 ## Scrolling one notch descends the ladder — cursor-anchored, so a cursor
 ## position must be supplied; the rung index advances by exactly one.
 func test_scroll_rung_descends_one_notch() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
 	v._scroll_rung(1, Vector2(400.0, 300.0))
@@ -43,7 +61,7 @@ func test_scroll_rung_descends_one_notch() -> void:
 func test_scroll_rung_clamps_at_the_deepest_rung() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
 	for _i in range(10):
@@ -52,7 +70,7 @@ func test_scroll_rung_clamps_at_the_deepest_rung() -> void:
 func test_reset_to_global_returns_from_a_deep_rung() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
 	v._scroll_rung(1, Vector2(400.0, 300.0))
@@ -71,7 +89,7 @@ func test_reset_to_global_returns_from_a_deep_rung() -> void:
 ## regression (scroll_step() clamping at index 0 meant _scroll_rung()
 ## returned before ever reaching a reset call).
 func test_scroll_out_at_global_after_a_pan_resets_the_view() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
 	assert_str(v.get_held_rung()).is_equal(StepCanvasTransport.RUNG_GLOBAL)
@@ -94,7 +112,7 @@ func test_scroll_out_at_global_after_a_pan_resets_the_view() -> void:
 ## (un-drifted) Global frame must stay a no-op — the reset is edge-triggered
 ## on genuine drift, not a per-scroll unconditional reset.
 func test_scroll_out_at_undrifted_global_is_a_no_op() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
 	v._scroll_rung(-1, Vector2(400.0, 300.0))
@@ -103,21 +121,21 @@ func test_scroll_out_at_undrifted_global_is_a_no_op() -> void:
 func test_overlay_visibility_defaults_to_off_for_every_toggle() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	for def: Dictionary in v.get_overlay_defs():
 		assert_bool(v.is_overlay_visible(def["id"])).is_false()
 func test_set_overlay_visible_updates_state() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.set_overlay_visible("gen_dw_temp", true)
 	assert_bool(v.is_overlay_visible("gen_dw_temp")).is_true()
 func test_set_overlay_visible_unknown_id_is_a_no_op() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.set_overlay_visible("not_a_real_overlay", true)
 	assert_bool(v.is_overlay_visible("not_a_real_overlay")).is_false()
@@ -201,7 +219,7 @@ static func _synthetic_canvas(width: int, height: int) -> Dictionary:
 func test_maybe_refloat_is_inert_before_any_canvas_has_arrived() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
 	v._scroll_rung(1, Vector2(400.0, 300.0))  # District — footprint still ZERO, nothing arrived
@@ -214,7 +232,7 @@ func test_maybe_refloat_is_inert_before_any_canvas_has_arrived() -> void:
 ## the held canvas keeps drawing, no re-request (§4/§5's "only when a pan
 ## carries the view past the held window's edge").
 func test_maybe_refloat_does_not_refloat_on_a_small_pan() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(800.0, 600.0)  # a real viewport size — _maybe_refloat's
 	# drift math is relative to get_rect().size's own center; leaving this at
@@ -244,7 +262,7 @@ func test_maybe_refloat_does_not_refloat_on_a_small_pan() -> void:
 ## world_center, view_offset reset to ZERO (the canvas re-centers under the
 ## new request).
 func test_maybe_refloat_refloats_once_the_pan_crosses_the_edge_threshold() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
 	v._scroll_rung(1, Vector2(400.0, 300.0))  # District
@@ -266,7 +284,7 @@ func test_maybe_refloat_refloats_once_the_pan_crosses_the_edge_threshold() -> vo
 ## Global never re-floats on pan (D-255(a): its canvas is the whole body,
 ## no edge to cross) — even with a real texture held and a huge drift.
 func test_maybe_refloat_is_a_no_op_at_global_rung() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
 	v._terrain_layer.rebuild_from_canvas(_synthetic_canvas(200, 100), v.get_held_rung(), "")
@@ -284,7 +302,7 @@ func test_maybe_refloat_is_a_no_op_at_global_rung() -> void:
 func test_edge_scroll_suppressed_without_application_focus() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(800.0, 600.0)
 	v._app_has_focus = false
@@ -293,7 +311,7 @@ func test_edge_scroll_suppressed_without_application_focus() -> void:
 func test_edge_scroll_suppressed_when_cursor_has_never_moved_over_the_control() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(800.0, 600.0)
 	# _last_mouse_pos defaults to (-1, -1) — an impossible in-bounds position,
@@ -303,7 +321,7 @@ func test_edge_scroll_suppressed_when_cursor_has_never_moved_over_the_control()
 func test_edge_scroll_active_near_the_left_edge() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(800.0, 600.0)
 	v._app_has_focus = true
@@ -315,7 +333,7 @@ func test_edge_scroll_active_near_the_left_edge() -> void:
 func test_edge_scroll_active_near_the_right_edge() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(800.0, 600.0)
 	v._app_has_focus = true
@@ -325,7 +343,7 @@ func test_edge_scroll_active_near_the_right_edge() -> void:
 func test_edge_scroll_inactive_well_inside_the_viewport() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(800.0, 600.0)
 	v._app_has_focus = true
@@ -345,7 +363,7 @@ func test_edge_scroll_inactive_well_inside_the_viewport() -> void:
 ## ever cached under it, so the sweep is a true no-op read (no writes to the
 ## real user://atlas_cache/ directory this test could leak).
 func test_enter_runs_the_disk_cache_visit_sweep_without_crashing() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "T1183_sweep_smoke_test_body", "body_radius_km": 6238.4}, {})
 	# If the wiring is broken (e.g. calling a method that doesn't exist), the
@@ -358,7 +376,7 @@ func test_enter_runs_the_disk_cache_visit_sweep_without_crashing() -> void:
 ## Timer node, not a _process()-driven counter), autostarts, and is set to
 ## the documented coarse interval — never a sub-frame or per-frame value.
 func test_disk_sweep_timer_is_coarse_and_autostarts() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	var timer: Timer = v.get_node("DiskSweepTimer")
 	assert_object(timer).is_not_null()
@@ -383,7 +401,7 @@ func test_disk_sweep_timer_is_coarse_and_autostarts() -> void:
 ## non-blocking test speed) against an injected-root request so this test
 ## touches no real cache files.
 func test_disk_sweep_timeout_handler_runs_background_sweep_for_the_current_body() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "T1183_sweep_smoke_test_body", "body_radius_km": 6238.4}, {})
 	# No live server, no cached entries for this body — the assertion is
@@ -422,7 +440,7 @@ static func _land_global_canvas(v: StepCanvasViewer, width: int, height: int) ->
 func test_global_canvas_arrival_populates_the_body_extent_cap_source() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "T1189_extent_letterbox_test_body", "body_radius_km": 6371.0}, {})
 	TestStepCanvasViewer._land_global_canvas(v, GJ1C_GLOBAL_EXTENT.x, GJ1C_GLOBAL_EXTENT.y)
@@ -434,7 +452,7 @@ func test_global_canvas_arrival_populates_the_body_extent_cap_source() -> void:
 ## scrolling to Region and firing its request must produce a CAPPED extent
 ## — never the raw viewport-fit 384x216 that overran the body on both axes.
 func test_region_request_extent_is_capped_to_the_landed_global_extent() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(1920.0, 1080.0)  # the T-1183 eyeball's own viewport
 	v.enter({"body_id": "T1189_extent_letterbox_test_body", "body_radius_km": 6371.0}, {})
@@ -457,7 +475,7 @@ func test_region_request_extent_is_capped_to_the_landed_global_extent() -> void:
 ## ZERO — the Region request must go out UNCAPPED (server clamps
 ## independently) rather than silently collapsing to a zero-cell request.
 func test_region_request_extent_is_uncapped_before_the_global_echo_lands() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(1920.0, 1080.0)
 	v.enter({"body_id": "T1189_extent_letterbox_test_body", "body_radius_km": 6371.0}, {})
@@ -481,7 +499,7 @@ func test_region_request_extent_is_uncapped_before_the_global_echo_lands() -> vo
 ## same key, capped or not, never a key built from one extent and served
 ## under another.
 func test_capped_extent_matches_what_the_request_actually_sends() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(1920.0, 1080.0)
 	v.enter({"body_id": "T1189_extent_letterbox_test_body", "body_radius_km": 6371.0}, {})
@@ -515,7 +533,7 @@ func test_capped_extent_matches_what_the_request_actually_sends() -> void:
 ## proving the cap is spacing-keyed, not applied indiscriminately to every
 ## rung once a Global extent is known.
 func test_district_request_extent_is_never_capped_by_the_global_extent() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(1920.0, 1080.0)
 	v.enter({"body_id": "T1189_extent_letterbox_test_body", "body_radius_km": 6371.0}, {})
@@ -538,7 +556,7 @@ func test_district_request_extent_is_never_capped_by_the_global_extent() -> void
 ## region grid") — this proves it, through the real enter()/land/enter()
 ## sequence rather than asserting the field directly only.
 func test_global_body_extent_resets_on_a_different_body_and_does_not_leak() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(1920.0, 1080.0)
@@ -575,7 +593,7 @@ func test_global_body_extent_resets_on_a_different_body_and_does_not_leak() -> v
 ## have computed a non-zero centering offset (unless the canvas happens to
 ## exactly fill the viewport, not the case here).
 func test_global_canvas_arrival_centers_the_view_not_top_left() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(1920.0, 1080.0)
 	v.enter({"body_id": "T1189_extent_letterbox_test_body", "body_radius_km": 6371.0}, {})
@@ -596,7 +614,7 @@ func test_global_canvas_arrival_centers_the_view_not_top_left() -> void:
 ## legend-reserved 1920x1080 case (R=9, scale=1.8) AND the 4K case (R=20,
 ## scale=4.0) — the SAME formula, no separate coverage-threshold branch.
 func test_global_canvas_scale_yields_an_exact_integer_pixels_per_gridunit_ratio() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(1920.0, 1080.0)
 	v.enter({"body_id": "T1189_extent_letterbox_test_body", "body_radius_km": 6371.0}, {})
@@ -618,7 +636,7 @@ func test_global_canvas_scale_yields_an_exact_integer_pixels_per_gridunit_ratio(
 ## too (20/5 = 4.0) — confirms the 1080p case above isn't a coincidence of
 ## a small viewport, just the same formula at a different achievable ratio.
 func test_global_canvas_scale_at_4k_also_yields_an_exact_integer_ratio() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(3840.0, 2160.0)
 	v.enter({"body_id": "T1189_extent_letterbox_test_body", "body_radius_km": 6371.0}, {})
@@ -633,7 +651,7 @@ func test_global_canvas_scale_at_4k_also_yields_an_exact_integer_ratio() -> void
 ## stays 1.0 once the player has descended past Global, even though a
 ## Global canvas was landed earlier in the same session.
 func test_fixed_rung_canvas_scale_stays_one_after_descending_from_global() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(1920.0, 1080.0)
 	v.enter({"body_id": "T1189_extent_letterbox_test_body", "body_radius_km": 6371.0}, {})
@@ -654,7 +672,7 @@ func test_fixed_rung_canvas_scale_stays_one_after_descending_from_global() -> vo
 ## and centered, its drawn rect's LEFT edge must be at or past the legend's
 ## own right edge, never underneath it.
 func test_global_canvas_left_edge_never_overlaps_the_legend_column() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(1920.0, 1080.0)
 	v.enter({"body_id": "T1189_extent_letterbox_test_body", "body_radius_km": 6371.0}, {})
@@ -671,7 +689,7 @@ func test_global_canvas_left_edge_never_overlaps_the_legend_column() -> void:
 ## one — _notification(NOTIFICATION_RESIZED) wires _recompute_canvas_transform()
 ## for exactly this case.
 func test_resize_recenters_an_already_held_global_canvas() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(1920.0, 1080.0)
 	v.enter({"body_id": "T1189_extent_letterbox_test_body", "body_radius_km": 6371.0}, {})
@@ -695,7 +713,7 @@ func test_resize_recenters_an_already_held_global_canvas() -> void:
 ## under 1.0, end-to-end through the real viewer wiring, not just the pure
 ## transport function this mirrors.
 func test_global_canvas_scale_never_drops_below_native_on_a_narrow_viewport() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.size = Vector2(348.0 + StepCanvasTransport.LEGEND_COLUMN_PX, 1080.0)
 	v.enter({"body_id": "T1189_extent_letterbox_test_body", "body_radius_km": 6371.0}, {})
@@ -720,7 +738,7 @@ func test_global_canvas_scale_never_drops_below_native_on_a_narrow_viewport() ->
 ## (verified end-to-end, through act(), in test_atlas_agent_interface.gd;
 ## this is the narrower unit-level check directly against the viewer).
 func test_jump_to_sets_held_rung_and_world_center() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
@@ -733,7 +751,7 @@ func test_jump_to_sets_held_rung_and_world_center() -> void:
 ## Omitting `rung` keeps whatever rung is currently held — the "revisit
 ## within the same rung" common case shouldn't require repeating it.
 func test_jump_to_keeps_current_rung_when_omitted() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
 	v._scroll_rung(1, Vector2(400.0, 300.0))  # Region
@@ -746,7 +764,7 @@ func test_jump_to_keeps_current_rung_when_omitted() -> void:
 ## Jumping to Global always forces world_center to ZERO — Global has no
 ## panned-center concept (mirrors _scroll_rung()'s own Global-rung handling).
 func test_jump_to_global_forces_world_center_to_zero() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
 	v._scroll_rung(1, Vector2(400.0, 300.0))  # Region
@@ -757,7 +775,7 @@ func test_jump_to_global_forces_world_center_to_zero() -> void:
 func test_jump_to_unrecognized_rung_is_a_no_op() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
 	var rung_before: String = v.get_held_rung()
@@ -770,7 +788,7 @@ func test_jump_to_unrecognized_rung_is_a_no_op() -> void:
 func test_get_current_canvas_summary_before_any_canvas_arrives() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
@@ -786,7 +804,7 @@ func test_get_current_canvas_summary_before_any_canvas_arrives() -> void:
 ## StepCanvasAnnotationLayer._draw_settlements()'s own dedup discipline —
 ## covering the same cell id twice must not double-count).
 func test_get_current_canvas_summary_counts_match_a_fixture_canvas() -> void:
-	var v: StepCanvasViewer = auto_free(StepCanvasViewer.new())
+	var v: StepCanvasViewer = _make_viewer()
 	add_child(v)
 	v.enter({"body_id": "GJ380c", "body_radius_km": 6238.4}, {})
 	v._scroll_rung(1, Vector2(400.0, 300.0))  # Region
@@ -150,6 +150,14 @@ var _legend_panel = null
 var _request = null  # StepCanvasRequest
 var _disk_sweep_timer: Timer = null  # T-1183 coarse background sweep trigger
 ## Test-injection-only (T-1193 first slice): threads into StepCanvasRequest's
 ## disk_cache_root seam so tests can isolate the Tier-2/3 disk cache from the
 ## machine-shared user://atlas_cache/ (one directory for every worktree gate,
 ## live capture driver, and real play session — a warm shared cache delivers
 ## canvases into tests written against a cold one). Set BEFORE add_child();
 ## the request is constructed in _ready(). Production always leaves this "".
 var disk_cache_root_override: String = ""
 func _ready() -> void:
 	anchor_right = 1.0
@@ -176,7 +184,7 @@ func _ready() -> void:
 	_annotation_layer.name = "AnnotationLayer"
 	_canvas.add_child(_annotation_layer)
-	_request = StepCanvasRequest.new(self)
+	_request = StepCanvasRequest.new(self, disk_cache_root_override)
 	_request.name = "Request"
 	add_child(_request)
 	_request.canvas_ready.connect(_on_canvas_ready)
@@ -14,7 +14,7 @@ Centralized tracking for all production assets across pipelines. Each pipeline h
 | Pipeline | Index | Palette | Status |
 |----------|-------|---------|--------|
 | [Audio](audio/README.md) | [audio/README.md](audio/README.md) | [audio/palette.md](audio/palette.md) | Active — Sprint 7 |
-| [Visual](visual/README.md) | [visual/README.md](visual/README.md) | [visual/palette.md](visual/palette.md) | Stub |
+| [Visual](visual/README.md) | [visual/README.md](visual/README.md) | [visual/palette.md](visual/palette.md) | Active — catalog + style bible instantiated (T-1050/T-1052) |
 | [Video](video/README.md) | [video/README.md](video/README.md) | [video/palette.md](video/palette.md) | Stub |
 ## Structure
@@ -24,7 +24,7 @@ docs/assets/
  README.md              # This file — master index
  _templates/
    audio.md             # Row template + detailed entry template for audio
-    sprite.md            # LEGACY object-sprite row template (D-244 retired sprites; model/texture/artwork template TBD)
+    sprite.md            # LEGACY object-sprite row template (D-244 retired sprites; model/texture/artwork rows are inline in each visual/*.md file, T-1050)
    video.md             # Row template + detailed entry template for video
  audio/
    README.md            # Audio pipeline index — categories, status summary
@@ -40,6 +40,7 @@ docs/assets/
    artwork.md           # Flat 2D art — paintings, flags, billboards, signage (via /sprite-gen)
    icons.md             # UI icons, status indicators (UI layer)
    effects.md           # Shaders, particles, overlays
    conventions.md       # Model naming, mask sidecars, D-235-keyed texture naming, footprint metadata (T-1050)
  video/
    README.md            # Video pipeline index
    palette.md           # Motion/editing style guide
@@ -7,7 +7,8 @@ status: active
 # Visual Asset Pipeline
-Status: **Stub** — awaiting first visual sprint.
+Status: **Active** — catalog structure instantiated (T-1050); awaiting first
 model/texture/artwork/effects production sprint.
 Per **[D-244](../../../governance/decisions/architecture.md#d-244)**: the in-world view renders **3D objects directly**. The only flattened 2D content is **textures** (PBR maps on 3D surfaces) and **flat 2D artwork** (paintings, flags, billboards, signage) applied as textures/decals on flat surfaces *within* the 3D world. **There is no per-object sprite layer** — game objects are not pre-rendered to sprites.
@@ -19,12 +20,17 @@ Per **[D-244](../../../governance/decisions/architecture.md#d-244)**: the in-wor
 | Characters | — | — | 3D meshes composited at runtime via `CharacterVisualDescriptor` — see [character-asset-organization.md](../../architecture/character-asset-organization.md) |
 | Textures | [textures.md](textures.md) | 0 | PBR surface maps (BaseColor / Normal / Roughness) for 3D models |
 | Artwork | [artwork.md](artwork.md) | 0 | Flat 2D art — paintings, flags, billboards, signage, screen content — applied as textures/decals (via the `/sprite-gen` 2D-artwork generator) |
-| Icons | [icons.md](icons.md) | 0 | UI icons, status indicators (2D, UI layer only) |
+| Icons | [icons.md](icons.md) | 13 | UI icons, status indicators (2D, UI layer only) — as-built inventory of `client/assets/icons/` |
 | Effects | [effects.md](effects.md) | 0 | Shaders, particles, overlays |
 Naming, mask-sidecar, D-235-keyed texture, and multi-tile footprint
 conventions: [conventions.md](conventions.md).
 ## Palette
-See [palette.md](palette.md) for art direction, "the angle", and colour relationships.
+See [palette.md](palette.md) for art direction: camera (D-148 gameplay / D-019
 offline-render), the toon-vs-PBR rendering treatment for environment props, the
 D-235 ObjectTag color/material register, and entity colour relationships (D-033).
 ## Pipeline
@@ -0,0 +1,41 @@
 ---
 title: "Artwork — Visual Assets"
 description: "Flat 2D artwork asset catalog — paintings, flags, billboards, signage, screen content"
 type: design
 status: active
 ---
 # Artwork — Visual Assets
 Flat 2D art — paintings, flags, billboards, signage, screen content — that
 lives as a texture/decal on a flat surface *within* the 3D world. This is one
 of the two flattened-2D exceptions under D-244 (the other is
 [textures.md](textures.md)); it is not a sprite layer.
 ## Generation Approach
 Produced via the `/sprite-gen` skill's **repurposed 2D-artwork role**
 (`renderer/README.md`) — the offline Godot renderer originally built for the
 retired object-sprite pipeline (#541 spike), now serving flat-artwork
 rendering. Camera/lighting/resolution mechanics from that pipeline are
 reused; the "render every entity/object/wall to sprites" framing in
 `renderer/README.md` is legacy and does not apply here.
 Output is promoted to `client/assets/artwork/` (per `renderer/README.md`'s
 own note: flat-artwork output goes to `client/assets/textures` or
 `client/assets/artwork`, never `client/assets/sprites/`).
 ## Assets
 | ID | Filename | Status | Type | Notes | Sprint/Ticket |
 |----|----------|--------|------|-------|---------------|
 No artwork has entered production yet — this table is intentionally empty at
 initial creation (T-1050).
 ## Cross-References
 - [conventions.md](conventions.md)
 - `renderer/README.md` — repurposed 2D-artwork render pipeline
 - `.claude/skills/sprite-gen/SKILL.md`
 - [D-244](../../../governance/decisions/architecture.md#d-244-asset-rendering--3d-objects-in-world-2d-limited-to-textures--flat-artwork)
@@ -0,0 +1,164 @@
 ---
 title: "Visual Asset Conventions"
 description: "Model naming, mask sidecars, D-235-keyed texture naming, and multi-tile footprint metadata for the 3D asset pipeline"
 type: design
 status: active
 ---
 # Visual Asset Conventions
 Single home for 3D-model and texture naming conventions (D-244 — the in-world
 view renders 3D objects directly; there is no per-object sprite layer). This
 file is the cross-link target from `.claude/skills/glb-gen/SKILL.md` and the
 catalog files in `docs/assets/visual/`.
 ---
 ## 1. Model naming
 Format: `<category>_<descriptor>.glb`
 Examples: `table_baroque.glb`, `chair_modernist.glb`, `door_sliding.glb`.
 **Ruling (delegated to Araminta, art seat, batch-5 activation — T-1052):
 category-first, not descriptor-first.** Rationale: category-first matches
 the shipped, in-active-use character-asset clothing convention
 (`<category>_<descriptor>` — `jacket_modern.glb`, `boots_modern.glb`,
 `pants_formal.glb`, all live under `client/assets/characters/clothing/`) and
 gives every future asset browser/manifest a stable alphabetical grouping by
 type (all tables sort together, all chairs sort together). The 3dpipeline
 spike's descriptor-first naming (`baroque_table.glb`, `modernist_chair.glb`,
 `spikes/3dpipeline/README.md`) was pre-convention scratch work, not a
 precedent to preserve — D-241 already treats spike naming as disposable at
 promotion time. **Surfaced to Jeroen for cheap veto** (lead delegation note,
 T-1050).
 `category` is the subdirectory-agnostic object type (`table`, `chair`,
 `door`, `lamp`, `statue`); `descriptor` distinguishes style/variant
 (`baroque`, `modernist`, `sliding`). Category is not required to match the
 `furniture`/`props` subdirectory name — a `statue` and a `table` can both be
 `furniture` or both be `props` depending on where they land; the manifest's
 `category` field, not the filename, is authoritative for that split.
 ---
 ## 2. Mask sidecars
 Convention: `<model>_mask.png`, single-channel (R-only greyscale), sitting
 next to the GLB — per the character-asset convention (Section 5,
 `docs/architecture/character-asset-organization.md`). White = tintable
 region, black = preserve original texture (`hint_default_black`: no mask
 loaded = no tinting; assets without a mask render with their original
 Trellis/authored texture).
 Example: `table_baroque.glb` + `table_baroque_mask.png` (this is the existing
 Trellis convention already proven in the 3dpipeline spike, just re-ordered to
 category-first per Section 1's ruling).
 **Multi-region tinting.** If a single-channel mask is insufficient (an item
 needs more than one independently-tintable region), follow the RGBA
 channel-routed convention from `character-asset-organization.md` Section 11
 (T-1089) — R/G/B/A → four independent tints, cross-fading correctly under
 bilinear filtering — rather than reinventing a greyscale-band scheme (that
 approach was tried for clothing and rejected; see that doc's Section 5
 correction). This is the same mechanism, not a parallel one: environment
 props reuse the clothing pipeline's multi-region shader family rather than
 inventing a props-specific one.
 ---
 ## 3. Texture naming (D-235 tokens)
 PBR surface maps for building-exterior and environment surfaces are named by
 the ratified **ObjectTag vocabulary** (D-235 amendment, T-995,
 `wiki/economics/object_tag_vocabulary.toml`), **not** the retired
 `renderer/README.md` `<model>_<direction>_<resolution>.png` scheme. That
 scheme is direction-indexed for the retired 4-direction sprite render
 (`renderer/README.md`'s now-legacy object-sprite mode) and does not apply to
 PBR texture maps at all — there is no "direction" for a texture, only **map
 type** (basecolor / normal / roughness).
 Format: `<object_tag>_<map_type>.png`
 Map types: `basecolor`, `normal`, `roughness`. (`metallic` and `ao` may be
 added if a material needs them; keep the same `<object_tag>_<map_type>`
 shape.)
 Examples, drawn from the D-235 ratified tags:
 - `concrete_wall_basecolor.png`, `concrete_wall_normal.png`, `concrete_wall_roughness.png`
 - `pitched_roof_normal.png`
 - `stone_wall_basecolor.png`
 - `generic_wall_basecolor.png` (fallback-terminal placeholder — D-235's
  fallback hierarchy: every specific tag degrades to a `generic_*` parent
  until the specific texture ships, then upgrades in place)
 The full ratified axis lists (wall/roof/facade/street + the four `generic_*`
 fallback terminals) live in `wiki/economics/object_tag_vocabulary.toml` —
 this doc does not duplicate that list; it names the file-naming convention
 that keys off it.
 ---
 ## 4. Multi-tile footprint metadata
 A table or couch spans multiple sim tiles regardless of whether it's
 rendered as a sprite or a 3D model — this requirement survived the D-244
 pivot unchanged. It is **not** expressed as a filename convention (a
 sprite-era assumption); it is a `manifest.json` field:
 ```json
 "footprint_tiles": [2, 1]
 ```
 `[width, height]` in whole tiles (1 tile = 1m, D-243's voxel unit). Single
 Sunday-tile props omit the field or set it to `[1, 1]`. See
 `client/assets/models/manifest.json`'s `_fields` block for the authoritative
 field definition alongside the rest of the per-asset schema.
 ---
 ## 5. Deferred — door state handling (NOT authored here)
 **Do not port a state-frame suffix convention** (e.g. `door_open.glb` /
 `door_closed.glb`). An earlier draft of this ticket (2026-06-16, since
 superseded) proposed exactly that — baked open/closed GLB variants — which is
 sprite-era thinking (baked state = baked frame) carried over from the retired
 object-sprite model.
 For a 3D object, the door's open/closed/locked/sealed state (T-979/T-998's
 `DoorSpec`: `Open` / `Closed` / `TemporalWindow` / `Locked` / `Sealed`) is a
 **runtime transform/animation on one door `.glb`**, not separate baked
 models. The actual mechanism — `AnimationPlayer` swing vs. procedural hinge
 rotation vs. a door-model convention with a named "open" bone — is a
 **Phase-5 rendering decision** (player control + in-world rendering, per the
 cascade), not a Phase-4 asset-catalog concern.
 This doc records only the negative constraint above (no state-frame suffix
 files). The positive mechanism is deferred to whichever Phase-5 ticket
 handles door rendering — do not invent a runtime convention here.
 ---
 ## 6. Deferred — Godot TileSet resource organization (OUT OF SCOPE)
 A `<tileset>/<tile_id>.tres` convention is **out of scope for this doc and
 this phase**, removed from the ticket that spawned this file. Reason:
 `client/scripts/rendering/tile_renderer.gd`'s `TileSet` is an
 explicitly-labeled placeholder ("programmatic TileSet with placeholder
 colored rectangles," D-014), and the cascade rule (`CLAUDE.md`) states the
 current in-world rendering code is "left as-is until Phase 5 — do not build
 on it or rip it out before then." Authoring a `.tres` resource-organization
 convention now would mean building on Phase-5-owned code from Phase 4. This
 section is intentionally not written; a future Phase-5 ticket owns it.
 ---
 ## Cross-References
 - [D-244](../../../governance/decisions/architecture.md#d-244-asset-rendering--3d-objects-in-world-2d-limited-to-textures--flat-artwork) — 3D objects in-world; 2D limited to textures + flat artwork.
 - [D-241](../../../governance/decisions/process.md#d-241-spike-binary-artifacts-are-local-only--gitignored-never-tracked) — spike binaries local-only; promotion = repo-canonical.
 - [D-235](../../../governance/decisions/architecture.md#d-235-building-exterior-visual-grammar-and-material-vocabulary) — ObjectTag vocabulary, ratified in `wiki/economics/object_tag_vocabulary.toml` (T-995).
 - [D-243](../../../governance/decisions/architecture.md#d-243) — spatial scale ladder (1m voxel/tile unit for `footprint_tiles`).
 - `docs/architecture/character-asset-organization.md` — Sections 3, 5, 11 (mask sidecar precedent, multi-region RGBA convention).
 - `client/assets/models/manifest.json` — the schema this doc's naming/metadata conventions feed.
 - `.claude/skills/glb-gen/SKILL.md` — the production pipeline that writes to `client/assets/models/`.
@@ -0,0 +1,41 @@
 ---
 title: "Effects — Visual Assets"
 description: "Shader, particle, and overlay asset catalog"
 type: design
 status: active
 ---
 # Effects — Visual Assets
 Shaders, particles, and overlays — the visual-effects layer distinct from
 models/textures/artwork/icons. Includes both 3D-scene shaders (toon,
 recolor-mask, outline) and 2D `CanvasLayer` overlays (fog, cognitive-delay
 cues).
 ## Generation Approach
 Hand-authored GDShader files, proven in the 3dpipeline and quaternius-
 aesthetic spikes (`toon.gdshader`, `toon_masked.gdshader`,
 `outline.gdshader`, `toon_garment.gdshader` — see
 `docs/architecture/character-asset-organization.md` Sections 2 and 11 for
 the shader family already in production use under
 `client/assets/characters/shaders/`). New effects (particles, screen-space
 overlays) are authored per-need; there is no generation pipeline analogous to
 `/glb-gen`/`/sprite-gen` for this category.
 ## Assets
 | ID | Filename | Status | Type | Notes | Sprint/Ticket |
 |----|----------|--------|------|-------|---------------|
 This table intentionally excludes the character shader family already
 tracked and documented in `character-asset-organization.md` (`toon.gdshader`,
 `toon_masked.gdshader`, `outline.gdshader`, `toon_garment.gdshader`) — that
 doc is their canonical home. This file catalogs effects/particles/overlays
 that don't already have a home elsewhere; none exist yet (T-1050 initial
 creation).
 ## Cross-References
 - `docs/architecture/character-asset-organization.md` — character shader family (canonical home, not duplicated here)
 - [D-244](../../../governance/decisions/architecture.md#d-244-asset-rendering--3d-objects-in-world-2d-limited-to-textures--flat-artwork)
@@ -0,0 +1,50 @@
 ---
 title: "Icons — Visual Assets"
 description: "UI icon asset catalog — as-built inventory of client/assets/icons/"
 type: design
 status: active
 ---
 # Icons — Visual Assets
 UI icons and status indicators. 2D, UI layer only — not part of the D-244 3D
 in-world pivot (icons render on `CanvasLayer`, not in the 3D scene).
 ## Generation Approach
 Hand-authored SVGs, tinted at runtime via `res://shaders/icon_tint.gdshader`
 (see `client/ui/stance_indicator.gd` for the tint-shader integration
 pattern).
 ## Assets
 13 SVGs already exist in `client/assets/icons/` — this is the as-built
 inventory, not a zero-start table.
 | ID | Filename | Status | Category | Notes | Sprint/Ticket |
 |----|----------|--------|----------|-------|---------------|
 | ICO-001 | `icon_stance_walk.svg` | final | Stance | Wired into `client/ui/stance_indicator.gd` (`StanceIndicator` icon map), tinted via `icon_tint.gdshader`. | — |
 | ICO-002 | `icon_stance_sprint.svg` | final | Stance | Wired into `stance_indicator.gd`. | — |
 | ICO-003 | `icon_stance_careful.svg` | final | Stance | Wired into `stance_indicator.gd`. | — |
 | ICO-004 | `icon_stance_crouch.svg` | final | Stance | Wired into `stance_indicator.gd`. | — |
 | ICO-005 | `icon_action_examine.svg` | placeholder | Action | SVG exists; not yet wired into a script/scene. | — |
 | ICO-006 | `icon_action_follow.svg` | placeholder | Action | SVG exists; not yet wired into a script/scene. | — |
 | ICO-007 | `icon_action_observe.svg` | placeholder | Action | SVG exists; not yet wired into a script/scene. | — |
 | ICO-008 | `icon_action_talk.svg` | placeholder | Action | SVG exists; not yet wired into a script/scene. | — |
 | ICO-009 | `icon_item_access_token.svg` | placeholder | Item | SVG exists; not yet wired into a script/scene. | — |
 | ICO-010 | `icon_item_comm_log.svg` | placeholder | Item | SVG exists; not yet wired into a script/scene. | — |
 | ICO-011 | `icon_item_manifest.svg` | placeholder | Item | SVG exists; not yet wired into a script/scene. | — |
 | ICO-012 | `icon_status_health.svg` | placeholder | Status | SVG exists; not yet wired into a script/scene. | — |
 | ICO-013 | `icon_status_perception.svg` | placeholder | Status | SVG exists; not yet wired into a script/scene. | — |
 `Status` here follows the shared vocab (`docs/assets/README.md`):
 `final` = shipped and wired into a consuming script/scene; `placeholder` =
 asset exists on disk but has no confirmed runtime consumer yet. If a
 teammate wires one of the placeholder icons into a script, flip its row to
 `final` and note the consumer, matching the `icon_stance_*` rows above.
 ## Cross-References
 - `client/assets/icons/` — the SVG files this table catalogs
 - `client/ui/stance_indicator.gd` — the one confirmed consumer (stance icons)
 - `res://shaders/icon_tint.gdshader` — runtime tint shader
@@ -0,0 +1,48 @@
 ---
 title: "Models — Visual Assets"
 description: "3D object/furniture/prop asset catalog — status table for the Trellis glb-gen pipeline"
 type: design
 status: active
 ---
 # Models — Visual Assets
 3D objects and furniture/props rendered directly in-world as `.glb` meshes
 (D-244 — no per-object sprite layer). Produced via the Trellis image→3D
 pipeline (`/glb-gen`), promoted per D-241 into `client/assets/models/`.
 ## Generation Approach
 - **Concept image:** `/image-gen` produces a concept PNG (square, plain dark
  background, single object centered, 3/4 front view — see
  `.claude/skills/glb-gen/SKILL.md` "Input Requirements").
 - **Image → 3D:** Trellis (`tower-of-joy:11510`) converts the concept image
  to a raw `.glb` (`/glb-gen`).
 - **Post-process:** Blender normalizes scale, centers the mesh, generates the
  recolor mask sidecar, and sets up materials for toon rendering
  (`.claude/skills/glb-gen/scripts/postprocess_glb.py`).
 - **Promotion:** the finished `.glb` (+ mask, if tintable) lands in
  `client/assets/models/{furniture,props}/` and is registered in
  `client/assets/models/manifest.json` — that registration + placement act
  **is** the promotion (D-241); it does not stay a "promoted" spike artifact.
 Naming, mask-sidecar, and multi-tile footprint conventions:
 [conventions.md](conventions.md).
 ## Assets
 | ID | Filename | Status | Category | Footprint | Notes | Sprint/Ticket |
 |----|----------|--------|----------|-----------|-------|---------------|
 No models have entered production yet — this table is intentionally empty at
 initial creation (T-1050). Rows are added as assets are promoted into
 `client/assets/models/manifest.json`; keep this table's `Status` column in
 sync with the manifest's `status` field (shared vocab: planned /
 in-progress / placeholder / final, `docs/assets/README.md`).
 ## Cross-References
 - [conventions.md](conventions.md) — naming, mask sidecars, footprint metadata
 - `client/assets/models/manifest.json` — the machine-readable per-asset registry
 - `.claude/skills/glb-gen/SKILL.md` — production pipeline
 - [D-244](../../../governance/decisions/architecture.md#d-244-asset-rendering--3d-objects-in-world-2d-limited-to-textures--flat-artwork), [D-241](../../../governance/decisions/process.md#d-241-spike-binary-artifacts-are-local-only--gitignored-never-tracked)
@@ -1,23 +1,296 @@
 ---
 title: "Visual Palette"
-description: "Art direction guide covering camera angle, entity colours, environmental neutrality, and functional warmth"
+description: "Phase-4 visual style bible — camera, entity colours, environmental neutrality, D-235 material register, toon/PBR rendering treatment"
 type: design
 status: active
 decision_refs: [D-019, D-033, D-045, D-148, D-149, D-217, D-232, D-235, D-244, D-257]
 author: "Araminta"
 updated: 2026-07-25
 ---
 # Visual Palette
-Status: **Stub** — reference art direction workshop output for current guidance.
+Status: **Active** — Phase-4 style bible (T-1052). This document is the current
 source of truth for rendered-look art direction. It supersedes the framing (not
 every individual data point) of the two documents listed in
 [Superseded documents](#superseded-documents) below.
-See `docs/workshops/art-direction-mood-board/` for the established visual identity.
+---
-## Key Decisions
+## 1. Camera
- **Camera angle:** 15-20deg from vertical ("the angle"), rendered via orthographic Camera3D at -72.5deg from horizontal (D-019)
+Two camera records exist and apply to **different surfaces** — this was
- **Entity colors:** Relationship-based per D-033 (green = known/friendly, amber = neutral, red = hostile, etc.)
+previously conflated in this document; corrected here.
 - **Environmental neutrality:** Spaces don't visually shift with narrative state (D-045)
 - **Functional warmth:** Industrial infrastructure that people made livable — not military, not luxury
-## Style Guide
+| Camera | Angle | Where it applies | Status |
 |--------|-------|-------------------|--------|
 | **D-148** | 30° low-angle `Camera3D` tilt (−30° from horizontal in the confirmed preset convention — see D-148's 2026-07-06 editorial note), 45° static map rotation | **Gameplay** — the default in-world camera, Phase 5+ | Active, confirmed default |
 | **D-019** | −72.5° from horizontal (midpoint of the old "15–20° from vertical" sprite-tilt range) | **Offline renderer only** — the retired 2D-sprite production pipeline (`renderer/`, now repurposed per D-244 as the flat-2D-artwork generator; see `renderer/README.md`) | Superseded for gameplay by D-148; survives only as an offline-render setting |
-To be populated from art direction workshop synthesis and first visual sprint.
+**Why both survive:** D-148 replaced D-019 as the *gameplay* camera when the
 project pivoted to live 3D rendering (D-149) and then to fully 3D objects
 (D-244) — there are no more pre-rendered sprites to view "at the angle" during
 play. D-019's −72.5° Camera3D setting doesn't disappear, though: it is still
 the correct setting for the **offline** render rig that produces flat 2D
 artwork (paintings, flags, billboards, signage — the D-244(b) content class),
 because that pipeline still renders a 3D scene down to a flat texture and
 needs a camera angle to do it. Two different cameras, two different jobs:
 one drives what the player sees, the other bakes a flat image asset.
 Do not describe D-019's angle as "the gameplay camera" anywhere — that
 sentence is what this document is correcting.
 ---
 ## 2. Rendering Treatment — Toon vs. PBR
 **Authority: [D-257](../../../governance/decisions/architecture.md#d-257)**
 (environment props share the character toon shading treatment; minimal-PBR
 carve-out for glazing). This section is the practical guidance derived from
 that record — the binding decision and its full rationale live in D-257, not
 here.
 **Environment props and furniture share the character toon treatment.** One
 shader family, one look, across the whole in-world scene — with a single,
 narrow, explicitly-named exception for glass and polished/mirror-finish metal,
 which get a distinct minimal-PBR material rather than being forced through
 the flat toon response.
 ### 2.1 What "sharing the toon treatment" means concretely
 - Environment `.glb` models use the same shader family as characters:
  `toon.gdshader` (flat lit/shadow-band ALBEDO, no PBR lighting response) or
  `toon_masked.gdshader` (recolor-mask variant, for props that need
  runtime tint — e.g. a favorite-color object, a faction-liveried crate)
  where a tintable variant is needed. `outline.gdshader` (inverted-hull) applies
  to entities and to any prop the visual hierarchy calls out for emphasis (see
  §4 tie-in below) — not universally to every static prop, matching the
  existing entity > object > structure outline-weight hierarchy.
 - glb-gen's postprocess convention of forcing `roughness=1.0` / `specular=0.0`
  "for toon compatibility" (`postprocess_glb.py::setup_materials()`) is
  **correct and should stay the default** for furniture, crates, signage,
  street furniture, wall/roof/facade instances — i.e. the D-235 ObjectTag
  vocabulary's bulk. These are flat-lit like characters; PBR specular
  highlights on a stucco wall or a wooden table would read as wet/glossy and
  contradict the toon language players already learn from characters.
 - **Exception — glass and polished/mirror metal:** the ratified D-235 glazing
  tokens — `glass_curtain_wall` (wall axis) and `industrial_glazing` (facade
  axis) — and any deliberately reflective/polished metal prop (chrome
  fixture, mirror, glossy display case) get a **separate minimal-PBR
  material**: `roughness` in the 0.05–0.2
  range and non-zero `specular`/`metallic` as appropriate, layered so the toon
  shadow-band lighting model still governs the base surface response but a
  specular highlight and partial reflection are allowed through. This is not
  a full PBR material (no environment-map reflections, no fresnel-driven
  rim lighting stack) — it is "toon base + a controlled specular term,"
  the smallest change that makes glass/chrome read as glass/chrome instead
  of matte plastic.
 ### 2.2 Rationale
 - **Visual grammar principle (mine, standing):** consistency compounds. A
  station built from one shader family reads as one coherent object, the way
  a real Rimworld or Hades scene does. Mixing a toon character against a
  PBR-lit room breaks the "one world" read the moment the player looks at
  their own hands next to a table.
 - **The postprocess default already assumes this.** `setup_materials()`'s
  comment ("for toon compatibility") was written for the character pipeline
  but Trellis is the *same* generator for environment `.glb`s — there was
  never a second postprocess path. Diverging environment props onto full PBR
  would mean either forking the postprocess script or hand-overriding every
  environment asset's material after generation. Sharing the treatment means
  the existing pipeline is already correct and needs no new work.
 - **The exception is functionally forced, not a taste call.** A flat-lit
  `roughness=1.0` glass pane doesn't read as glass — it reads as a frosted
  grey panel, which actively misleads the player about sightlines (can I see
  through this?) in a game whose core mechanic is occlusion-based perception.
  Where "can the player see through/into this" is gameplay-load-bearing
  (`glass_curtain_wall`, `industrial_glazing`, shop windows, interior
  partitions), the material needs to look transparent/reflective enough to answer that
  question at a glance. This is a narrow, named carve-out — it does not
  reopen "should props be PBR," it closes the one case where toon-only
  actively lies to the player.
 - **Scale check:** this holds at boxes-with-labels (a labelled box has no
  material response either way, so the ruling is inert today) and at full
  art fidelity (the shader assignment is a material-slot choice per asset,
  not a structural decision that needs revisiting later).
 ### 2.3 What this does NOT change
 - D-235's material/token vocabulary (§3) is unaffected — this ruling is about
  shader/material *response*, not which WallMaterial/RoofForm tokens exist.
 - The character shader files (`toon.gdshader`, `toon_masked.gdshader`,
  `toon_garment.gdshader`, `outline.gdshader`) are not modified by this
  ruling — environment props reuse them as-is or via a near-identical
  environment-scoped shader variant if slot/uniform needs differ
  (implementation detail for whoever writes the first environment shader,
  not an art-direction fork).
 ---
 ## 3. Color / Material Register — D-235 ObjectTag Vocabulary
 D-235 (amended 2026-07-07, T-995) ratified the **logical** token vocabulary —
 `wiki/economics/object_tag_vocabulary.toml`, 4 axes (wall/roof/facade/street),
 32 specific tags + 4 generic fallback parents. D-235's own Implementation note
 explicitly parks the **rendered look** of those tokens for "Phase 5+ and
 post-launch" art production. This section is that pre-specification: not new
 tokens, but the register each shipped tag should render into once art exists,
 so asset generation has a target from day one instead of drifting per-artist.
 **How to read this table:** "Hue range" is a rendered-look target for the
 BaseColor map's dominant tone, expressed relative to D-235's existing rule
 that *color is template-bounded* (a cultural palette cue, seed-selected within
 range, "always within the template's register") — this section does not
 override that; it gives the physically-plausible hue family each material
 naturally sits in, which the template's seed then picks within. "Weathering
 baseline" describes how the D-217 condition bands (Intact/Worn/Cracked/Broken,
 driven by `prosperity_score`) should visually read for that material — i.e.
 what "Worn" looks like on brick vs. on corrugated metal.
 ### 3.1 Wall materials
 | Tag | Hue range | Grain / texture character | Weathering baseline (D-217 bands) |
 |-----|-----------|---------------------------|-----------------------------------|
 | `concrete_wall` | Neutral grey, warm-leaning (`#8a8680`–`#6e6a64`) — never blue-grey/cool-grey | Fine aggregate speckle, subtle formwork seams | Intact: even matte. Worn: rain-streak staining, hairline cracks. Cracked: spalling at edges, exposed aggregate. Broken: structural cracking, rebar-stain bleed. |
 | `steel_frame` | Dark warm charcoal (`#3a3632`) frame, infill panel per adjacent tag | Visible bolted/riveted joints, slight surface oxidation tint even when "new" | Intact: factory-primer flat. Worn: surface rust bloom at joints/edges only. Cracked: running rust streaks below joints. Broken: structural rust-through, buckling. |
 | `brick_wall` | Warm terracotta-to-ochre (`#9c5a3c`–`#c48a52`), never cool/pink brick | Visible coursing + mortar joint shadow, slight color variance per brick (batch variation) | Intact: crisp mortar lines. Worn: mortar erosion, efflorescence (white mineral bloom). Cracked: spalled brick faces, missing mortar. Broken: collapsed coursing, exposed cavity. |
 | `rendered_wall` | Whatever the template's color_range picks — this tag is a **finish**, not a hue driver; render surface is smooth and matte regardless of picked hue | Smooth, minimal grain, occasional hairline render-crack pattern at Worn+ | Intact: clean flat. Worn: surface staining/streaking under sills and joints. Cracked: render cracking, patches of exposed substrate. Broken: large render loss, substrate fully exposed. |
 | `stone_wall` | Warm grey-tan (`#a89e8c`–`#8c8172`) — **explicitly not cool grey**; this is the D-235 fallback-graph's flagship "gets it wrong if literal" case | Visible coursing, natural stone color variance tile-to-tile, tool-marked or rough-cut surface depending on template | Intact: clean-cut edges. Worn: lichen/mineral staining in shadowed recesses. Cracked: mortar loss between courses, chipped arrises. Broken: displaced/missing stone units. |
 | `timber_wall` | Warm mid-brown (`#7a5c3e`–`#9c7a52`), never grey-driftwood unless template explicitly calls for weathered/coastal register | Visible grain direction, board/plank joint lines | Intact: sealed/painted finish. Worn: graying at exposed edges, finish wear. Cracked: splitting along grain, missing sealant. Broken: rot, structural failure at joints. |
 | `stucco_wall` | Warm off-white to sand (`#d8cfc0`–`#c4b89e`) — whitewashed per the tag description, not pure white | Smooth, slight hand-applied texture variance | Intact: clean whitewash. Worn: staining streaks below any protrusion. Cracked: hairline cracking, patch-color mismatch. Broken: large loss exposing substrate. |
 | `glass_curtain_wall` | Neutral-to-cool tint on the glass itself (`#c8d4dc` @ low opacity) — see §2.1 PBR exception, this tag needs the specular/transparency carve-out to read correctly | Minimal — reflection/transparency IS the surface character | Intact: clean, full reflectivity. Worn: dust film, reduced reflectivity. Cracked: visible cracked panes, some opacity/frosting. Broken: missing panes, boarded sections. |
 | `composite_panel` | Whatever the template picks — this is a **prefab finish** tag like `rendered_wall`, hue is template-driven, not material-driven | Flat, uniform, visible panel-seam grid (regular rhythm, unlike masonry coursing) | Intact: crisp seams, clean finish. Worn: seam staining, minor panel discoloration. Cracked: panel warping, seam gaps. Broken: missing/detached panels. |
 | `rammed_earth_wall` | Warm ochre-to-umber banding (`#8a6a48`–`#6e5236`) | Strong horizontal layer banding (the construction method's signature texture) | Intact: crisp banding. Worn: surface erosion softening band edges. Cracked: vertical cracking across bands. Broken: section collapse/erosion loss. |
 ### 3.2 Roof forms
 | Tag | Hue range | Grain / texture character | Weathering baseline |
 |-----|-----------|---------------------------|----------------------|
 | `flat_roof` | Neutral dark grey (`#4a4844`) membrane | Flat, minimal detail, visible seam lines on membrane roofing | Intact→Broken: ponding stains, membrane cracking, vegetation growth at Broken. |
 | `pitched_roof` | Warm dark grey-brown (`#4e463e`) generic shingle/slate register unless a specific tag (below) narrows it | Directional shingle/slate coursing | Moss/lichen accumulation in shadowed pitches as condition drops; missing units at Cracked+. |
 | `corrugated_roof` | Weathers to **rust-orange** (`#a65c34` at Worn+), starts galvanized silver-grey (`#a8aaa8`) when Intact — **explicitly does not stay silver**, this is D-235's own worked example | Strong linear corrugation shadow pattern | Intact: clean galvanized sheen. Worn: rust bleeding from fastener points. Cracked: broad rust staining, sheet lifting at edges. Broken: rust-through holes, missing sheets. |
 | `clay_tile_roof` | Warm terracotta (`#b56a3e`–`#c8825a`) | Individual tile coursing, slight per-tile hue variance | Moss/lichen in valleys, cracked/slipped/missing tiles as condition drops. |
 | `terraced_roof` | Match the deck surface (paved/planted per template), not a distinct roof hue | Flat deck texture, visible rail/parapet line | Surface wear on deck material, planter/rail deterioration at lower bands. |
 | `vaulted_roof` | Match wall material register (masonry vaults read as an extension of the wall) | Strong directional shadow from the vault curvature | Standard masonry weathering (see stone/brick wall rows). |
 | `green_roof` | Vegetated green, seasonally variable — the one roof tag where hue is intentionally NOT template-locked | Organic, irregular, canopy-like from above | Condition reads as vegetation health/coverage, not surface damage — sparse/patchy at low prosperity, not "broken" in the structural sense. |
 ### 3.3 Facade rhythms
 Facade tags are about **pattern**, not hue — they inherit the wall tag's
 color register. Grain/weathering guidance:
 | Tag | Grain / texture character | Weathering note |
 |-----|---------------------------|-------------------|
 | `regular_facade` | Even punched-window grid, no added ornament | Weathers uniformly with the wall tag beneath it. |
 | `ornamental_facade` | Raised/carved detail catches shadow distinctly from the flat wall plane | Ornament erodes/chips *before* the flat wall does — detail loss is the earliest visible sign of Worn. |
 | `industrial_glazing` | Regular steel-mullion grid, factory-sash proportions (wide, short panes) | Mullion rust (if `steel_frame`) precedes glazing damage. |
 | `arcade_facade` | Strong repeating vertical shadow rhythm from the colonnade | Ground-level wear concentrates at column bases (foot traffic), not evenly. |
 | `shuttered_facade` | Shutters as a distinct, often higher-saturation accent color against the wall's muted register — this is the one facade tag allowed a saturation bump (bounded by §5's object-tier ceiling below) | Shutter paint fades/peels faster than the wall behind it — a secondary, faster weathering clock on the same building. |
 | `screen_facade` | Fine repeating perforation/louvre pattern, strong dappled shadow | Screen material (usually metal or timber lattice) weathers per its own material, independent of the wall plane behind. |
 | `colonnade` | Freestanding column rhythm, deep shadow gaps between wall plane and columns | Column bases weather fastest (ground contact, foot traffic) — same principle as arcade_facade. |
 | `lattice_screen` | Fine timber/metal lattice, mashrabiya-register — highest facade information-density texture in the set | Individual lattice elements can go missing at Cracked+ without the whole screen reading as "broken." |
 ### 3.4 Street surfaces
 | Tag | Hue range | Grain / texture character | Weathering baseline |
 |-----|-----------|---------------------------|----------------------|
 | `paved` | Neutral grey (`#66645e`) | Fine asphalt/poured texture, minimal joint pattern | Cracking, pothole pattern, patch-color mismatch (repairs never quite match original pour age). |
 | `cobble` | Warm grey-tan, matches `stone_wall`'s register | Strong individual-unit texture, visible mortar/sand joints | Displaced/missing setts at low condition, not surface cracking (it's a unit system, not a monolithic pour). |
 | `packed_earth` | Warm ochre-brown (`#7a6248`) | Irregular, rutted, no joint pattern | Erosion channels, puddling, vegetation encroachment at edges as condition drops (or rises — an unmaintained packed-earth street can look "worse" toward Intact-adjacent if it's simply less used, this tag is condition-light). |
 | `canal_way` | Water surface — hue driven by reflection/sediment, not a fixed material hue | N/A — water plane, not a solid surface | Water clarity/debris accumulation stands in for the condition ladder here. |
 | `elevated_walkway` | Match structural material used (steel_frame/timber/composite per template) | Grated or planked, visible support structure below | Structural weathering per the material tag, plus grate/plank gap wear from foot traffic. |
 | `heavy_haul` | Neutral dark grey (`#4e4c48`), reinforced-surface register | Deep rut/track wear pattern, wider joint spacing than `paved` | Track-line wear concentrates on wheel paths, not the full surface — an asymmetric wear pattern unlike `paved`'s even cracking. |
 | `boardwalk` | Warm grey-brown weathered timber (`#8a7a64`), coastal/wetland register — assume some grey weathering even at Intact, unlike `timber_wall` | Plank coursing, visible gap joints | Individual plank replacement creates natural color patchwork even at high condition — this is the one street tag where slight inconsistency reads as *authentic*, not neglected. |
 ### 3.5 Generic fallback placeholders
 `generic_wall` / `generic_roof` / `generic_facade` / `generic_street` render
 as a **flat, neutral, low-saturation grey** (`#6e6c68`, roughly the mean of
 the specific tags' hue ranges above) with **no grain and no weathering
 variation** — condition bands still apply structurally (a generic-fallback
 wall can still be "Broken" in silhouette/damage terms) but carry no material
 texture, so the player reads "this is a wall, its specific material hasn't
 loaded yet" rather than mistaking it for a deliberate concrete/stone choice.
 This is the visual expression of D-235's fallback chain — a placeholder must
 look *unfinished*, not like a legitimate flat concrete building, or its
 eventual upgrade-in-place would read as an inconsistent hue reskin instead of
 "detail arriving."
 ---
 ## 4. What Survives from Prior Documents
 Per D-244 (2026-06-16) and this ticket's acceptance criteria, the following
 principles from the pre-cascade documents are **inherited**, restated here in
 current (3D, D-235-token) terms rather than the old (2D sprite, era-ladder)
 terms they were originally written in:
 | Inherited principle | Original source | Current expression |
 |----------------------|------------------|----------------------|
 | **D-033 entity color = relationship to player** | visual-grammar-v01.md §3 | Unchanged — still the live decision, still the single most important color system in the game. Applies to entities regardless of render fidelity (rectangle, sprite, or 3D mesh). |
 | **"Functional warmth"** — industrial infrastructure that people made livable, not military/not luxury | workshop-outcomes.md visual identity statements + Gore's naming | Standing mood target for the whole D-235 register above: hue ranges lean warm (terracotta, ochre, warm grey) rather than cool/clinical by default; the *cool* end of the palette is reserved for deliberately institutional template registers, not the default. |
 | **Environmental neutrality — strict zero shift (D-045)** | workshop-outcomes.md §1.10 | Unchanged, still active. The D-235 material register above is diegetic/static — condition bands (D-217) shift with simulated `prosperity_score`, never with narrative/investigation state. This document adds no new exception to D-045. |
 | **Entity always wins visual ties** (visual hierarchy: entity > object > structure) | visual-grammar-v01.md §1.4 / workshop-outcomes.md §1.4 | Survives as a rendering-order and saturation-hierarchy principle. In 3D terms: entity materials/outlines render at the highest effective saturation and are never occluded in a way that reads ambiguous against a similarly-toned prop. The saturation ceiling table (below) replaces the old sprite-outline-weight table. |
 | **Object favorite-color saturation ceiling** | visual-grammar-v01.md §3.6 | Restated: entity colors (D-033) remain the highest-saturation elements in any scene; object/prop colors (including the `shuttered_facade` accent exception in §3.3) stay moderate; structure/material register (§3) stays lowest. Same three-tier ceiling, now anchored to D-235 tags instead of "Era 1/2/3" object palettes. |
 | **Readability over beauty** | workshop-outcomes.md Principle 1 | Standing design principle, unchanged, applies to 3D rendering exactly as it applied to sprites. |
 | **Lighting-driven atmosphere, not baked mood** | workshop-outcomes.md Principle 2 / visual-grammar-v01.md's "sprites are shape templates" | Re-expressed in §2 above: toon shading gives shape + local shadow-band response; scene lighting (not baked texture darkening) still carries mood. The specific mechanism (Light2D CanvasModulate) is retired with the sprite pipeline, but the principle — materials stay neutral, the lighting rig does the emotional work — carries forward to whatever 3D lighting setup Phase 5 builds. |
 ---
 ## 5. Superseded Documents
 Two earlier documents overlap this one. Both are **explicitly marked
 superseded** here (and should carry a pointer back to this section) so no two
 contradictory style documents stay live at the same time:
 - **`docs/design/visual-grammar-v01.md`** — already carries its own
  superseded banner (2026-06-12, pointing at D-235/D-228 +
  character-visuals-spec.md). That banner is correct and unchanged by this
  ticket. What's **specifically superseded**: the entire sprite-era
  production model (§2 entity sprite dimensions, §4 z-level stack as sprite
  layers, §5 typography-for-sprite-canvas sizing, §6 two-tier animation
  frame-count tables, §9 resolution chain / Nano Banana workflow, and the
  **three-construction-era material ladder** (§1.2–§1.4's Era 1/2/3 zone
  palettes) — the era ladder is the exact thing D-235 retired by name
  ("Era is NOT a material filter here"). What's **not superseded, just
  relocated**: D-033 entity color values and D-045 environmental neutrality,
  both restated current in §4 above.
 - **`docs/workshops/art-direction-mood-board/workshop-outcomes.md`** — this
  is a workshop transcript (type: workshop, status: archived) and correctly
  stays archived as a historical record; it is **not** live style guidance
  and should not be read as such. Its **candidate decisions** (D-042 through
  D-051 as originally proposed in §7) were formally recorded as the
  perception.md decisions cited throughout this document (D-033, D-043,
  D-044, D-045, D-046 etc.) — those decision records are the living
  artifact, not the workshop prose. Its **three-construction-era material
  system** (§1.8) is superseded by D-235 for the same reason as above. Its
  **camera angle content** (§1.2, §9.3) is superseded by D-148 per §1 above.
  Everything else in it (lighting three-reference model, fog cone shape,
  two-tier animation, insert overlay bloom spec, z-stack layer *concept* —
  not its sprite-specific implementation detail) remains valid design
  intent, inherited into current decisions rather than into this document
  directly.
 **Going forward:** this document (`palette.md`) is the single live style
 bible for Phase 4+ rendered look. New style guidance should land here, not in
 a new parallel document, unless it's genuinely a different concern (e.g. the
 model-naming/manifest conventions living in `docs/assets/visual/conventions.md`,
 authored separately under T-1050).
 ---
 ## Appendix — Decision Cross-References
 | Decision | Relevance |
 |----------|-----------|
 | D-019 | Original top-down camera + offline-renderer −72.5° angle. Superseded for gameplay by D-148; survives for the offline 2D-artwork render rig. |
 | D-033 | Entity color = relationship to player. Unchanged, still the primary information-bearing color system. |
 | D-045 | Environmental neutrality — strict zero shift. Governs how the D-235 material register (§3) may and may not respond to game state. |
 | D-148 | 30° low-angle gameplay `Camera3D`, supersedes D-019 for gameplay. |
 | D-149 | 3D characters rendered live in scene, not pre-rendered sprites. Backing precedent for §1's camera reasoning (no more sprites to render "at the angle" during play) and for §2's shared-shader-family argument (characters and environment props come off the same live-3D pipeline). |
 | D-217 | Tile condition thresholds (prosperity_score bands) — the weathering-baseline mechanism referenced throughout §3. |
 | D-232 | Trait-template catalog — era reframed as maintenance/wear + past-vogue holdover, not a material-tech ladder. Backing rationale for retiring the Era 1/2/3 ladder in §5. |
 | D-235 | Building exterior visual grammar + ratified ObjectTag vocabulary (`object_tag_vocabulary.toml`). §3 is this document's rendered-look layer on top of D-235's logical tokens; §2's glazing carve-out names its ratified `glass_curtain_wall`/`industrial_glazing` tokens. |
 | D-244 | 3D objects in-world; 2D limited to textures + flat artwork. Establishes the toon-shader-family context for §2 and retires the sprite catalog framing this document used to have. |
 | D-257 | Environment props share the character toon shading treatment; minimal-PBR carve-out for glazing. The binding authority for §2 — this document states the practical guidance, D-257 is the decision record. |
@@ -0,0 +1,42 @@
 ---
 title: "Textures — Visual Assets"
 description: "PBR surface map asset catalog for 3D model and building-exterior surfaces"
 type: design
 status: active
 ---
 # Textures — Visual Assets
 PBR surface maps (BaseColor / Normal / Roughness) applied to 3D model and
 building-exterior surfaces. The only flattened 2D content besides flat
 artwork (D-244) — a texture has no "direction," only map type.
 ## Generation Approach
 - **Building-exterior surfaces:** named by the ratified ObjectTag vocabulary
  (D-235 amendment, T-995, `wiki/economics/object_tag_vocabulary.toml`) — see
  [conventions.md](conventions.md) §3 for the naming convention and the
  fallback-hierarchy note (every specific tag degrades to a `generic_*`
  parent until its specific texture ships).
 - **Model surfaces:** Trellis bakes a texture atlas directly onto the mesh
  during `/glb-gen` (no separate texture-authoring step for most
  furniture/props) — see [models.md](models.md). This file tracks
  **standalone** PBR maps: building-exterior materials and any surface asset
  authored or generated independent of a specific model.
 ## Assets
 | ID | Filename | Status | Map Type | ObjectTag | Notes | Sprint/Ticket |
 |----|----------|--------|----------|-----------|-------|---------------|
 No textures have entered production yet — this table is intentionally empty
 at initial creation (T-1050). Per D-235, the texture library itself is
 Phase 5+/post-launch work (behind the fallback chain); this file exists now
 so the catalog structure is in place ahead of that work landing.
 ## Cross-References
 - [conventions.md](conventions.md) — D-235-keyed texture naming
 - [models.md](models.md) — model-baked textures (Trellis atlas bakes)
 - [D-235](../../../governance/decisions/architecture.md#d-235-building-exterior-visual-grammar-and-material-vocabulary) — ObjectTag vocabulary + fallback hierarchy
 - `wiki/economics/object_tag_vocabulary.toml` — the ratified tag registry
@@ -8,9 +8,19 @@ decision_refs: [D-033, D-043, D-044, D-045, D-046, D-047, D-048, D-049, D-052, D
 author: "Araminta"
 created: 2026-02-19
 updated: 2026-02-20
-superseded_by: "D-235/D-228"
+superseded_by: "D-235/D-228, docs/assets/visual/palette.md"
 ---
 > **SUPERSEDED (2026-06-12, cascade-refocus sweep):** superseded by D-235/D-228 (generated-world visual grammar) + docs/design/character-visuals-spec.md (character rules). Kept as historical record; do not build against it.
 >
 > **Style-bible pointer (2026-07-25, T-1052):** the live Phase-4 style bible is
 > `docs/assets/visual/palette.md`. Its §4/§5 spell out exactly what survives from
 > this document (D-033 colors, "functional warmth," the visual-hierarchy/saturation
 > principles) versus what's superseded (the sprite-era production model: entity
 > sprite sizing, the 8-layer sprite z-stack, sprite typography sizing, animation
 > frame counts, the resolution chain, and — specifically — the **three-construction-era
 > material ladder** in §1.2–§1.4, which D-235 retired by name: "Era is NOT a material
 > filter here"). Read palette.md first; come back here only for historical
 > attribution.
 # The Settled Reach — v0.1 Visual Grammar
@@ -18,6 +18,17 @@ decision_refs: [D-015, D-019, D-024, D-031, D-033, D-035, D-041, D-042, D-043, D
 **Rounds:** 3 + Closing + Post-workshop technical session (Tyre, Araminta, Stig)
 **Status:** Complete — all items resolved, ready for formal decision recording. Pipeline test in progress.
 > **Archived transcript, not live style guidance (2026-07-25, T-1052):** this is a
 > workshop record — read it for attribution and reasoning history, not as current
 > art direction. Its candidate decisions (§7) were formally recorded as the
 > perception.md/architecture.md decisions cited throughout (D-033, D-043–D-051); those
 > D-records are the living artifact. Its camera-angle content (§1.2, §9.3) is
 > superseded by [D-148](../../../governance/decisions/architecture.md#d-148-30-low-angle-camera-with-45-map-rotation--supersedes-d-019).
 > Its three-construction-era material system (§1.8) is superseded by
 > [D-235](../../../governance/decisions/architecture.md#d-235) ("Era is NOT a material
 > filter here"). The current Phase-4 style bible is
 > [`docs/assets/visual/palette.md`](../../assets/visual/palette.md) — start there.
 ---
 ## Visual Identity Statements
@@ -79,7 +90,7 @@ Each participant's one-sentence answer to "What is this game's visual identity?"
 ### 1.3 World Composition — Tile-Based
-**Dual-scale grid: 0.5m sim tiles, 1m visual tiles (2x retina factor, [D-066](../../decisions/architecture.md#d-066-dual-scale-grid--05m-simulation-1m-visual-2x-retina-factor)). Visual tiles are 64x64px.**
+**Dual-scale grid: 0.5m sim tiles, 1m visual tiles (2x retina factor, [D-066](../../../governance/decisions/architecture.md#d-066-dual-scale-grid--05m-simulation-1m-visual-2x-retina-factor)). Visual tiles are 64x64px.**
 - Structural tiles (walls, floors, doors): 64x64px per visual tile (= 2x2 sim tiles), muted, minimal outlines, zone palette + era differentiation
 - Object tiles (furniture, equipment, containers): 64x64px per visual tile, medium outline (1px), Rimworld object detail as target. All world geometry is 2x2 sim tile minimum.
@@ -108,7 +119,7 @@ Each participant's one-sentence answer to "What is this game's visual identity?"
 ### 1.5 Entity System
-**24x32 pixel footprint within 64x64px visual tiles. Entities occupy 1x1 sim tiles (0.5m) but render across a 2x2 sim tile sprite footprint per [D-066](../../decisions/architecture.md#d-066-dual-scale-grid--05m-simulation-1m-visual-2x-retina-factor).**
+**24x32 pixel footprint within 64x64px visual tiles. Entities occupy 1x1 sim tiles (0.5m) but render across a 2x2 sim tile sprite footprint per [D-066](../../../governance/decisions/architecture.md#d-066-dual-scale-grid--05m-simulation-1m-visual-2x-retina-factor).**
 - Entity smaller than visual tile = clear figure-ground relationship
 - D-033 color as primary information signal
@@ -322,6 +322,7 @@ line in place — keep the Q-record for the audit trail rather than deleting it.
 - [D-254: Standalone Atlas companion app — `make atlas`, dual-connection reader](decisions/architecture.md#d-254-standalone-atlas-companion-app--make-atlas-dual-connection-reader) — _architecture_
 - [D-255: Body Map Viewer — stepped Atlas render architecture (supersedes the T-1143 continuous-ladder mechanism)](decisions/architecture.md#d-255-body-map-viewer--stepped-atlas-render-architecture-supersedes-the-t-1143-continuous-ladder-mechanism) — _architecture_
 - [D-256: Canonical sampling convention — one absolute-metre derive core; the batch layer is a survey raster](decisions/architecture.md#d-256-canonical-sampling-convention--one-absolute-metre-derive-core-the-batch-layer-is-a-survey-raster) — _architecture_
 - [D-257: Environment props share the character toon shading treatment; minimal-PBR carve-out for glazing](decisions/architecture.md#d-257-environment-props-share-the-character-toon-shading-treatment-minimal-pbr-carve-out-for-glazing) — _architecture_
 ## Open questions
@@ -2313,4 +2313,18 @@ Technical foundation decisions that constrain implementation: engine, client-ser
 ---
 ### D-257: Environment props share the character toon shading treatment; minimal-PBR carve-out for glazing
 - **Date:** 2026-07-25
 - **Decision:** Environment props and furniture (the D-235 ObjectTag vocabulary's rendered assets — walls, roofs, facades, street furniture, and freestanding furniture/prop `.glb`s produced via `/glb-gen`) render with the **same toon shader family already used for characters** (`toon.gdshader` / `toon_masked.gdshader`, per `client/assets/characters/shaders/`), not a distinct PBR material response. The glb-gen postprocess convention of forcing `roughness=1.0`/`specular=0.0` ("for toon compatibility," `postprocess_glb.py::setup_materials()`) is confirmed as the correct default for this vocabulary's bulk.
  - **Named carve-out — glass and polished/mirror metal.** The ratified glazing tokens — **`glass_curtain_wall`** (D-235 wall axis) and **`industrial_glazing`** (D-235 facade axis) — plus any deliberately reflective/polished-metal prop (chrome fixture, mirror, glossy display case), get a **minimal-PBR material** layered over the toon base: `roughness` in the 0.05–0.2 range with non-zero `specular`/`metallic`, not a full PBR stack (no environment-map reflections, no fresnel-driven rim lighting). The toon shadow-band lighting model still governs the base surface response; only a controlled specular/transparency term is added.
  - This binds all future environment-shader and glb-gen-postprocess work; it is not a per-asset style suggestion.
 - **Rationale:** Consistency compounds — a station built from one shader family reads as one coherent object; mixing a toon character against a PBR-lit room breaks the "one world" read the instant the player looks at their own hands next to a table (D-043/D-044 visual-hierarchy principles: entities and objects share one legible material language, distinguished by outline weight and saturation tier, not by rendering technology). The postprocess default already assumes this — `setup_materials()`'s "for toon compatibility" comment was written against the one Trellis pipeline that produces both character and environment `.glb`s, so sharing the treatment requires no new postprocess fork. The glazing carve-out is functionally forced, not a taste call: a flat-lit `roughness=1.0` glass pane reads as a frosted grey panel, which actively misleads the player about sightlines in a game whose core mechanic is occlusion-based perception (D-033's asymmetric-information model depends on the player correctly judging what they can and cannot see through). The carve-out is narrow and named — it closes only the case where toon-only actively lies to the player about transparency, and does not reopen "should props be PBR" generally.
 - **Raised by:** Araminta (T-1052, delegated art-direction ruling), reviewed by Tyre (PR #211 — flagged the original carve-out's use of retired pre-amendment example tokens `precision_glass`/`smart_facade`, corrected here to the ratified D-235 tokens).
 - **Implementation:** `docs/assets/visual/palette.md` §2 (practical guidance, cites this record as authority). Applies to all environment-prop material setup from Phase 4 asset production onward; no existing shipped assets to migrate as of this record's date.
 - **Cross-reference:** [D-235](#d-235) (ratified ObjectTag vocabulary — `glass_curtain_wall`/`industrial_glazing` are its tokens, not this record's invention), [D-244](#d-244) (3D objects in-world; establishes the shared Trellis/glb-gen pipeline this record's postprocess argument depends on), [D-043/D-044](perception.md#d-043-art-direction--visual-style-functional-warmth) (visual-hierarchy/"functional warmth" rationale — entity/object/structure share one material language), [D-033](perception.md#d-033-entity-color--relationship-to-player) (the occlusion/asymmetric-information rationale for the glazing carve-out).
 - **Dissent:** None.
 ---
 *110 decisions (D-001 through D-256, excluding gaps). Last updated: 2026-07-24 (D-256 — canonical sampling convention: one absolute-metre derive core (`derive_at_metres` family), batch pseudo-grid re-scoped as `SurveyCellPos` survey raster (real newtype), thin-wrapper unification fixing the three-latitude inconsistency and the region-(0,0) climate collapse, exact-position feature judgments, step-canvas fencing; discharges the T-1174 blocker on D-255's T-1181 envelope).*