Draw tfrag per-vertex colour, unlit, and record why the frame is dark

The tfrag RGBA array now modulates the texture. It is indexed by
*position*, not by vertex-info: several vertex-info entries can share a
position, and the array is sized vert_count, which is a position count.

Colour follows the same rule as the vertex itself. Morphing blends the
position only -- the game keeps the high-detail colour -- while collapse
replaces the vertex outright, colour included. Getting that the other way
round would tint distant terrain.

The material is unlit, and that is a decision rather than a shortcut.
Level 0's colours really are dark (patch 0 runs (25,26,26) to
(108,144,141)) and the game's lighting is what brightens them, so what is
on screen is the true albedo.

Worth recording: the obvious move -- add a directional light, or crank
ambient brightness -- does nothing at all. With no normals the PBR path
is degenerate, so the brightness change has no effect, and a host light
would have hidden the gap rather than shown it. The colour data was
always reaching the mesh; what is missing is lighting, not a material
flag.

Written up in plan/level-assets.md, including what LightTfrags actually
needs: the light bank from the level's gameplay extent, the 256-entry
normal table at BOOT_NORMAL_TABLE_VADDR (0x165500) which every level
overlay carries, and the already-parsed azimuth/elevation light records.

crates/gc-engine/src/level.rs

@@ -66,6 +66,7 @@ pub struct PatchBounds {
 pub struct TriangleBatch {
     pub positions: Vec<[f32; 3]>,
     pub uvs: Vec<[f32; 2]>,
+    pub colors: Vec<[f32; 4]>,
     /// Three vertex indices per triangle.
     pub indices: Vec<u32>,
     /// The ad-gif in force per triangle, parallel to `indices`.
@@ -170,6 +171,31 @@ pub fn uv(g: &TfragGeometry, vinfo: usize) -> [f32; 2] {
 pub struct ResolvedVertex {
     pub position: [f32; 3],
     pub uv: [f32; 2],
+    /// The patch's stored per-vertex colour.
+    ///
+    /// Indexed by *position*, not by vertex-info: several vertex-info entries
+    /// can share a position, and the colour array is sized `vert_count`, which
+    /// is a position count. This is the game's base colour before lighting
+    /// multiplies into it, so on its own it is the unlit albedo.
+    pub color: [f32; 4],
+}
+
+/// The stored colour of a vertex's position, or opaque white if absent.
+pub fn vertex_color(g: &TfragGeometry, vinfo: usize) -> [f32; 4] {
+    let i = g.position_index(vinfo);
+    match g.rgba.get(i) {
+        // The disc's 0x80-is-1.0 alpha convention, same as the texture CLUT.
+        Some(c) => {
+            let a = if c.a < 0x80 { c.a as f32 / 128.0 } else { 1.0 };
+            [
+                c.r as f32 / 255.0,
+                c.g as f32 / 255.0,
+                c.b as f32 / 255.0,
+                a,
+            ]
+        }
+        None => [1.0, 1.0, 1.0, 1.0],
+    }
 }
 
 /// Resolve one vertex-info entry at a given view depth.
@@ -190,6 +216,7 @@ pub fn resolve_vertex(
         return ResolvedVertex {
             position: world(g, vinfo),
             uv: uv(g, vinfo),
+            color: vertex_color(g, vinfo),
         };
     };
     let limit = link.tier.collapse_distance(lod_distances);
@@ -197,12 +224,15 @@ pub fn resolve_vertex(
         return ResolvedVertex {
             position: world(g, link.parent1_vinfo),
             uv: uv(g, link.parent1_vinfo),
+            // Collapse replaces the vertex outright, colour included.
+            color: vertex_color(g, link.parent1_vinfo),
         };
     }
     if !link.morphs {
         return ResolvedVertex {
             position: world(g, vinfo),
             uv: uv(g, vinfo),
+            color: vertex_color(g, vinfo),
         };
     }
     let u = link.tier.weight(depth, lod_distances);
@@ -210,6 +240,7 @@ pub fn resolve_vertex(
         return ResolvedVertex {
             position: world(g, vinfo),
             uv: uv(g, vinfo),
+            color: vertex_color(g, vinfo),
         };
     }
     let p1 = raw_position(g, link.parent1_vinfo);
@@ -225,10 +256,12 @@ pub fn resolve_vertex(
         here[1] + (mid[1] - here[1]) * u,
         here[2] + (mid[2] - here[2]) * u,
     ];
-    // UVs are not morphed: the game keeps the high-detail texture coordinate.
+    // UVs and colours are not morphed: the blend is positional only, and the
+    // game keeps the high-detail texture coordinate and colour.
     ResolvedVertex {
         position: to_world(blended),
         uv: uv(g, vinfo),
+        color: vertex_color(g, vinfo),
     }
 }
 
@@ -316,6 +349,7 @@ pub fn build_level(src: &LevelSource<'_>, camera: [f32; 3]) -> anyhow::Result<Le
                 .or_insert_with(|| TriangleBatch {
                     positions: Vec::new(),
                     uvs: Vec::new(),
+                    colors: Vec::new(),
                     indices: Vec::new(),
                     ad_gifs: Vec::new(),
                 });
@@ -324,6 +358,7 @@ pub fn build_level(src: &LevelSource<'_>, camera: [f32; 3]) -> anyhow::Result<Le
                 let r = resolve_vertex(&g, v, depth, lod_distances, collapse);
                 batch.positions.push(r.position);
                 batch.uvs.push(r.uv);
+                batch.colors.push(r.color);
             }
             batch.indices.extend_from_slice(&[base, base + 1, base + 2]);
             batch.ad_gifs.push(*ad);

crates/gc-engine/src/main.rs

@@ -388,6 +388,7 @@ fn mesh_of(b: &TriangleBatch) -> Option<Mesh> {
     );
     mesh.insert_attribute(Mesh::ATTRIBUTE_POSITION, b.positions.clone());
     mesh.insert_attribute(Mesh::ATTRIBUTE_UV_0, b.uvs.clone());
+    mesh.insert_attribute(Mesh::ATTRIBUTE_COLOR, b.colors.clone());
     // No normals and no colours: lighting is the game's own per-vertex light
     // array, which is parsed but not yet reproduced. Inventing normals would
     // make unlit terrain look lit and hide that gap.

docs/plan/level-assets.md

@@ -516,12 +516,40 @@ Lesson for the rest of the project: a parser can be entirely self-consistent
 and still describe the wrong thing. Where a value is absolute rather than
 relative, check it against an independent source -- here, the bounding spheres.
 
+### Why the first frame is dark
+
+The renderer draws unlit: texture times the tfrag's own per-vertex colour.
+That product is genuinely dark -- level 0 patch 0's colours are `(25, 26, 26)`
+and `(92, 121, 119)` in 0..255 -- and the game's lighting is what brightens it.
+
+Trying a host directional light or cranking ambient brightness was the wrong
+instinct and it showed: with no normals the PBR path is degenerate, so
+brightness changes did nothing at all. Faking it would also have hidden the
+gap. The colour attribute is picked up from the mesh automatically, so it is
+the *colour* data that is missing lighting, not a flag that was left off.
+
+The real model is `LightTfrags`, which needs three things this project has not
+parsed yet:
+
+* the **light bank** from the level's `gameplay` extent (directional and point
+  light sets, and the per-vertex light `select` field);
+* the **normal table**, 256 pairs of cos/sin read from an ELF at
+  `BOOT_NORMAL_TABLE_VADDR` (`0x165500`) -- every level overlay carries one, so
+  it does not need the boot ELF;
+* the **light records** themselves, which are already parsed: azimuth and
+  elevation at pi/128 radians per unit, and the base colour.
+
+Given the table, a normal is `(cos(az)cos(el), sin(az)cos(el), sin(el))`,
+negated, exactly as `decode_normal` builds it. The lighting pass itself is
+`base + A.rgb * max(dA, dA*wA) + B.rgb * max(dB, dB*wB)` per vertex, and it
+must be reproduced in the PS2's float model rather than in Rust `f32`, per
+the fidelity rule.
+
 ### Still open
 
-The GS register values the level load builds from the texture entry (TBP0/TBP1
-and so on), which a renderer needs if it emulates the GS rather than resolving
-texture addresses itself, and the guard-band clip test that sends patches
-crossing the guard band to the clipping program.
+The lighting above, the GS register values the level load builds from the
+texture entry (TBP0/TBP1 and so on), and the guard-band clip test that sends
+patches crossing the guard band to the clipping program.
 
 A tfrag can now be turned into positioned, UV-mapped, textured triangles,
 told which LOD to draw, and morphed toward its parents at the right rate.

docs/roadmap.md

@@ -31,8 +31,16 @@ early, while they are still cheap to retire.
       passed: the vertex count matched, the triangle count matched, only the
       *place* was wrong. Fixed by taking the STROW immediately preceding the
       first position unpack rather than the last one.
-- [ ] Next: the per-vertex light records, which is why the first frame is flat
-      and dark. Then the GS register set and the guard-band clip test.
+- [x] **Per-vertex colour.** The tfrag's RGBA array now modulates the texture,
+      drawn unlit so the frame shows real albedo rather than a faked lighting
+      model. Verified against the stored values (level 0 patch 0 runs
+      `(25,26,26)` to `(108,144,141)`), which are genuinely dark.
+- [ ] **tfrag lighting.** Needs the level light bank from `gameplay`, the
+      256-entry normal table from the overlay ELF at `0x165500`, and the
+      already-parsed azimuth/elevation light records. The pass itself is
+      `base + A*max(dA, dA*wA) + B*max(dB, dB*wB)` and has to run in the PS2
+      float model. Detail in `plan/level-assets.md`.
+- [ ] Then the GS register set and the guard-band clip test.
 
 The disc being the exact build the reference projects pin is the single
 biggest favourable fact so far: no disc conversion, no regional timing work,