Skip to content

Commit e2d7974

Browse files
TMHSDigitalclaude
andcommitted
docs(showcase): regenerate the hay-bale gallery page
The page embeds the piece README, whose texel figure was corrected after the last build. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Signed-off-by: TMHSDigital <154358121+TMHSDigital@users.noreply.github.com>
1 parent bf9a3a6 commit e2d7974

1 file changed

Lines changed: 2 additions & 2 deletions

File tree

‎docs/gallery/hay-bale/index.html‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -289,7 +289,7 @@ <h2>How the belts are built</h2>
289289
<h2>Budgets</h2>
290290
<p>Declared as named constants; every gate <strong>recomputes</strong> from the mesh, materials, UVs, evaluated LOD, collider, or export file. Measurements below are identical on 4.5.11, 5.1.2 and 5.2.1 — construction uses no RNG and no version-dependent operator.</p>
291291
<div class="table-wrap"><table><thead><tr><th scope="col">Axis</th><th scope="col">Declared</th><th scope="col">Measured (4.5.11 / 5.1.2 / 5.2.1)</th></tr></thead><tbody><tr><td>Base triangles</td><td>1700–2400</td><td>2044 on all three</td></tr><tr><td>LOD1 ratio</td><td>0.32–0.62 of base</td><td>0.5000</td></tr><tr><td>LOD2 ratio</td><td>0.10–0.35 of base</td><td>0.2192</td></tr><tr><td>Materials</td><td>exactly 2 distinct, ≥24 hay, ≥24 twine</td><td>2 slots, 1436 hay, 304 twine</td></tr><tr><td>UVs</td><td>in <code>0..1</code>, AABB overlap ≤ 1e-5</td><td>in range, overlap 0</td></tr><tr><td>Outer AABB</td><td>(0.921, 0.508, 0.417) m ± 0.01</td><td>(0.9208, 0.5075, 0.4165)</td></tr><tr><td>Grounded zmin</td><td>within 1e-4 of 0</td><td>0.000000</td></tr><tr><td>Hygiene</td><td>loose/nonman/zero-area/doubles/n-gons/<strong>z-fight</strong> = 0</td><td>all 0</td></tr><tr><td>Shell count</td><td>exactly 5 (loaf, 2 belts, 2 loops)</td><td>5</td></tr><tr><td><strong>Belt wrap zmin</strong></td><td>each belt shell ≤ 0.006 m</td><td>0.001460 both</td></tr><tr><td><strong>Belt seat depth</strong></td><td>every station in [0.002, 0.020] m</td><td>0.00400..0.01114 over 20/20 stations, both belts</td></tr><tr><td><strong>Mirror symmetry</strong></td><td>belts and loops matched within 5e-4 m</td><td>0.000000</td></tr><tr><td><strong>Loaf mirror</strong></td><td>every loaf vertex has an X-mirror partner within 2e-4 m</td><td>0.000093</td></tr><tr><td><strong>Cinch depth</strong></td><td>mid-span minus waist in [0.006, 0.020] m</td><td>0.01053</td></tr><tr><td>Loaf X × Z</td><td>(0.921, 0.393) m ± 0.02</td><td>(0.9208, 0.3931)</td></tr><tr><td>Hay–twine gap</td><td>BVH surface &lt; 0.008 m</td><td>0.00009</td></tr><tr><td>Collider tris</td><td>≤ 400</td><td>140</td></tr><tr><td>Bake texels</td><td>smallest UV cell ≥ 12 px at the baked resolution</td><td>22.43 px at 1024 px</td></tr><tr><td>Export</td><td>written, size &gt; 0</td><td>203508 / 203508 / 203496 bytes</td></tr></tbody></table></div>
292-
<p>DECIMATE COLLAPSE triangle counts are <strong>not</strong> guaranteed identical across series — the gate is a ratio band, not an exact count. Bake pixels are stochastic; the gate is <code>has_data</code> plus operator <code>FINISHED</code>, not byte-identity, plus the texel floor. The bake was 256 px, which left about 8 texels per UV cell and smeared neighbouring cells&#x27; normals into streaks across the straw; at 1024 px with a 0.02 m cage (was 0.04, which reached the twine standing proud of the loaf) every cell gets 22. Export byte counts differ by 12 B on 5.2.1 (glTF serializer), which is not a gated axis; every geometric figure above is identical on all three binaries.</p>
292+
<p>DECIMATE COLLAPSE triangle counts are <strong>not</strong> guaranteed identical across series — the gate is a ratio band, not an exact count. Bake pixels are stochastic; the gate is <code>has_data</code> plus operator <code>FINISHED</code>, not byte-identity, plus the texel floor. The bake was 256 px, which left 5.6 texels per UV cell and smeared neighbouring cells&#x27; normals into streaks across the straw; at 1024 px with a 0.02 m cage (was 0.04, which reached the twine standing proud of the loaf) every cell gets 22. Export byte counts differ by 12 B on 5.2.1 (glTF serializer), which is not a gated axis; every geometric figure above is identical on all three binaries.</p>
293293
<p>The export is deleted once its size is measured. Blender sets <code>TMPDIR</code> from its own preference, which resolves to the working directory on a stock portable build, so <code>tempfile.gettempdir()</code> is the repo root under CI and every run used to leave a <code>.glb</code> behind.</p>
294294
<h3>Why these budgets and not an AABB</h3>
295295
<p><code>zmin</code> alone cannot see a belt that stops at the bottom edge: the loaf still grounds the box. A single global hay-twine gap cannot see a wrap that hugs one side and gaps the other. A bounding box cannot see two belts 5.4 mm out of step, or a flank with no waist in it. Each of the five budgets in bold above exists because a defect got past everything else and had to be found by looking at a render.</p>
@@ -431,7 +431,7 @@ <h2>Source</h2>
431431
UV_EPS = <span class="n">1e-4</span>
432432
UV_OVERLAP_MAX = <span class="n">1e-5</span>
433433
COLLIDER_TRIS_MAX = <span class="n">400</span>
434-
<span class="c"># One UV cell per face: at 256 px each cell got about 8 texels, and the</span>
434+
<span class="c"># One UV cell per face: at 256 px each cell got 5.6 texels, and the</span>
435435
<span class="c"># bilinear lookup smeared neighbouring cells&#x27; normals into streaks across</span>
436436
<span class="c"># the straw (tavern-stool&#x27;s defect). 1024 px, with a floor asserted.</span>
437437
BAKE_RES = <span class="n">1024</span>

0 commit comments

Comments
 (0)