fix(sky): drive wrap mode from mesh UV range — fixes Bug B (stars-as-square)

Bug B in docs/research/2026-04-26-sky-investigation-handoff.md: stars
rendered as a small square in one corner of the sky instead of stretching
across the dome.

Root cause: the wrap-mode heuristic at SkyRenderer.cs:234-237 was
"GL_CLAMP_TO_EDGE unless TexVelocity != 0". That heuristic was tuned to
fix a separate symptom (the outer dome 0x010015EE/F0/F1/F2 shows
wall-seam bleed under GL_REPEAT because of bilinear-filter sampling at
texel boundaries). But it misclassified any *static* sky object whose
mesh UVs are deliberately authored outside [0,1] to tile the texture
across the geometry.

The smoking gun: GfxObj 0x010015EF is OI-1 in EVERY DayGroup (always
loaded), has TexVelocity = 0 (no scrolling), and authors UVs in
[0.398, 4.602] (texture tiles ~4× across each face). Under
CLAMP_TO_EDGE the bulk of the inner dome sampled the texture's edge
texels; only the small region where UVs happened to fall in [0,1]
showed actual texture content. Hence "a square in one corner".

Fix:

* GfxObjMesh.Build() now scans the resulting per-vertex UVs and sets
  GfxObjSubMesh.NeedsUvRepeat true when any component lies outside
  [0,1]. Mesh-time scan, not draw-time guess.
* SubMeshGpu carries the flag through to draw time.
* SkyRenderer uses `sub.NeedsUvRepeat || obj.TexVelocity != 0` to
  decide REPEAT vs CLAMP_TO_EDGE. The dome (UVs in [0,1]) keeps
  CLAMP — no seam regression. The inner star/sky layer 0x010015EF
  (UVs outside [0,1]) gets REPEAT — texture tiles across the dome.
  Cloud meshes (UVs outside [0,1] AND non-zero TexVelocity) keep
  REPEAT via either branch.

Probe-driven: tools/StarsProbe (committed in 991fb9a) dumps every
SkyObject's geometry + UVs and flags meshes whose UV range exceeds
[0,1]. Run `dotnet run --project tools/StarsProbe -c Release` to
re-derive.

Verified visually by user against the live ACE server in Holtburg —
stars now stretch across the night sky instead of appearing as a
square in one corner. Build green, dotnet test 1222 pass.

Note: this is functionally retail-equivalent for the reported bug but
not the exact retail mechanism. Retail's GameSky::Draw at 0x00506ff0
relies on D3D's global default D3DTADDRESS_WRAP (i.e. REPEAT
everywhere). True retail-faithfulness would require investigating why
our pipeline shows seams on the dome under REPEAT (likely a bilinear
filter / non-seamless texture detail). The data-driven approach taken
here preserves working dome behavior while fixing the broken star
behavior.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

src/AcDream.App/Rendering/Sky/SkyRenderer.cs

@@ -222,17 +222,31 @@ public sealed unsafe class SkyRenderer : IDisposable
                 _gl.ActiveTexture(TextureUnit.Texture0);
                 _gl.BindTexture(TextureTarget.Texture2D, tex);
 
-                // Sky meshes need per-object wrap mode. The dome is 5 flat
-                // walls meeting at edges — under GL_REPEAT any UV drift
-                // past [0,1] wraps to the opposite edge of the texture,
-                // drawing a visible line along each wall seam. Static
-                // sky GfxObjs (dome, sun, moon, stars) should use
-                // CLAMP_TO_EDGE to avoid that bleed. Scrolling cloud
-                // layers (TexVelocity != 0) still need REPEAT so the
-                // animated UV offset wraps correctly. Detection heuristic:
-                // non-zero TexVelocity on either axis ⇒ scrolling layer.
-                bool scrolling = obj.TexVelocityX != 0f || obj.TexVelocityY != 0f;
-                int wrapMode = scrolling
+                // Sky meshes need per-object wrap mode driven by the
+                // mesh's authored UV range, not by TexVelocity:
+                //   * The outer dome (0x010015EE/F0/F1/F2) authors UVs
+                //     strictly in [0,1]. Under GL_REPEAT the bilinear
+                //     filter at wall-seam edges would average a texel
+                //     near the right edge with one near the left edge of
+                //     the texture, drawing a visible "bleed line" along
+                //     every dome seam. CLAMP_TO_EDGE avoids that.
+                //   * The inner sky/star layer (0x010015EF) and the
+                //     cloud meshes (0x010015B6, 0x01004C36 etc) author
+                //     UVs that deliberately exceed [0,1] (~0.4..4.6) so
+                //     the texture tiles across the geometry. CLAMP_TO_EDGE
+                //     would clamp ~99% of the surface to a single edge
+                //     texel, leaving only a small "square" where UVs
+                //     happen to fall in [0,1] (Bug B in
+                //     docs/research/2026-04-26-sky-investigation-handoff.md).
+                // The mesh builder pre-computes NeedsUvRepeat from the
+                // actual UV range so the right answer is data-driven.
+                // Scrolling clouds are also forced to REPEAT (the running
+                // UV offset can drift outside [0,1] regardless of authored
+                // range, and they'd show their own seam bleed otherwise).
+                bool needsRepeat = sub.NeedsUvRepeat
+                    || obj.TexVelocityX != 0f
+                    || obj.TexVelocityY != 0f;
+                int wrapMode = needsRepeat
                     ? (int)TextureWrapMode.Repeat
                     : (int)TextureWrapMode.ClampToEdge;
                 _gl.TexParameter(TextureTarget.Texture2D, TextureParameterName.TextureWrapS, wrapMode);
@@ -423,6 +437,7 @@ public sealed unsafe class SkyRenderer : IDisposable
             SurfaceId = sm.SurfaceId,
             IsAdditive = isAdditive,
             SurfLuminosity = sm.Luminosity,
+            NeedsUvRepeat = sm.NeedsUvRepeat,
         };
     }
 
@@ -462,5 +477,15 @@ public sealed unsafe class SkyRenderer : IDisposable
         /// <c>docs/research/2026-04-23-sky-retail-verbatim.md</c> §6.
         /// </summary>
         public float SurfLuminosity;
+        /// <summary>
+        /// True when the source mesh's authored UVs exceed [0,1] (e.g.
+        /// the inner sky/star layer 0x010015EF and the cloud meshes —
+        /// they tile their texture across the geometry). The renderer
+        /// must use <c>GL_REPEAT</c> for these or only the small region
+        /// where UVs fall in [0,1] samples the actual texture; the rest
+        /// clamps to the edge texel ("square in one corner" symptom).
+        /// Computed once at mesh build from the actual UV range.
+        /// </summary>
+        public bool NeedsUvRepeat;
     }
 }

src/AcDream.Core/Meshing/GfxObjMesh.cs

@@ -210,6 +210,23 @@ public static class GfxObjMesh
                 }
             }
 
+            // Authored UV range determines the wrap-mode choice in the
+            // sky pass. A mesh whose UVs are strictly in [0,1] (e.g. the
+            // outer dome 0x010015EE) wants CLAMP_TO_EDGE to avoid
+            // bilinear-filter bleed at the wall-seam edges; a mesh whose
+            // UVs deliberately tile (e.g. 0x010015EF, ~0.4..4.6) wants
+            // REPEAT so the texture tiles across the geometry. We make
+            // the call data-driven here rather than guessing from
+            // TexVelocity at draw time. See
+            // docs/research/2026-04-26-sky-investigation-handoff.md (Bug B).
+            bool needsUvRepeat = false;
+            foreach (var v in kvp.Value.Vertices)
+            {
+                if (v.TexCoord.X < 0f || v.TexCoord.X > 1f
+                    || v.TexCoord.Y < 0f || v.TexCoord.Y > 1f)
+                { needsUvRepeat = true; break; }
+            }
+
             result.Add(new GfxObjSubMesh(
                 SurfaceId: surfaceId,
                 Vertices: kvp.Value.Vertices.ToArray(),
@@ -217,6 +234,7 @@ public static class GfxObjMesh
             {
                 Translucency = translucency,
                 Luminosity = luminosity,
+                NeedsUvRepeat = needsUvRepeat,
             });
         }
         return result;

src/AcDream.Core/Meshing/GfxObjSubMesh.cs

@@ -39,4 +39,18 @@ public sealed record GfxObjSubMesh(
     /// normal lighting path without change.
     /// </summary>
     public float Luminosity { get; init; } = 0f;
+
+    /// <summary>
+    /// True when at least one vertex's UV component lies outside the
+    /// <c>[0, 1]</c> range, meaning the mesh was authored to have its
+    /// texture tile across the geometry (i.e. it expects
+    /// <c>GL_REPEAT</c>/<c>D3DTADDRESS_WRAP</c>). The sky renderer reads
+    /// this to decide between <c>GL_REPEAT</c> (this flag set, or any
+    /// scrolling layer) and <c>GL_CLAMP_TO_EDGE</c> (all UVs strictly
+    /// in <c>[0,1]</c>), which avoids wall-seam bleed on the dome
+    /// (UVs in <c>[0,1]</c>) while still tiling the inner star/cloud
+    /// layers (UVs in <c>[~0.4, ~4.6]</c>) correctly.
+    /// Defaults to false so non-sky consumers get the previous behavior.
+    /// </summary>
+    public bool NeedsUvRepeat { get; init; } = false;
 }