standalone: landing plate removed; passage shaft rings reduced with the rock detail kept

- AddLanding no longer adds a shallow positive ellipsoid (floor plate check 71.3% -> 0%).
- Round 10 had removed all relief on passage walls to pass the ring check (smooth tube; rejected).
  Cause found instead: the owner roughness's 8-voxel Z anchors with linear interpolation. Passage-
  controlled walls now use 2-voxel anchors, blended smoothly into the 8-voxel ones.
- Ring gate (lag-20 correlation) 0.376 -> 0.038 AND shaft relief RMS 1.294 (smooth-tube control
  0.276 fails); passage/cave relief ratio 1.152. Faint rings remain deep in the shaft.
- Whole view 0.984 s cold @8w (interleaved, full payload); W1/W8 44/44 hash fields equal; routes
  5642/5642; look and far-survival audits pass; bad-shell control fails.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01MmSoF64BbRFmk9C3p1VsBJ

Standalone/Research/OwnerShapeWorld/speed_round/REPORT.md

@@ -1,10 +1,22 @@
-# Shape world speed round — final half-load verification
+# Shape world — speed and look verification
+
+## Round 11 — passage shaft anchor spacing (2026-10-05)
+
+Round 11 restores full shaft relief and keeps the landing plate fix. The broad shaft rings track the 8-voxel owner-roughness anchors: the 20-pixel radial-profile correlation falls from **0.376** to **0.038** when passage-controlled walls use 2-voxel anchors. Finer distance quantization and cubic anchor interpolation leave the repeat in place. The focused image gate fails the full-detail pre-fix capture for rings, fails the smooth-tube control for lost relief, and passes the final capture on both measures; the shaft wall relief RMS is **1.152×** nearby cave walls.
+
+The required cold W8 full-payload interleave averages **0.984 s** for Round 11 and **0.986 s** for Round 10. Both new runs are below one second. All runs generate 2,226 tiles with the full **110.90 MiB** payload and zero cache hits. W1/W8 match all 44 hash fields both at the default benchmark position and around the passage camera; the default region also matches the restored full-detail pre-fix executable. Routes remain **5,642/5,642**; the shape/look and far-survival audits pass, and the damaged-shell control fails as intended. The six fixed captures and four remote yaw views are recaptured and compared in the [Round 11 report](round11/REPORT.md).
+
+## Round 10 — passage look defects (2026-10-05)
+
+Round 10 removes the false landing-floor rock overlay and suppresses coherent relief where passage capsules control the shaft wall. The fixed passage and close-up image gates fail on the Round 9 captures and pass on Round 10. The two small components reported at a tile boundary were traced to per-tile grouping: both share every tested edge vertex with an adjacent tile, so they were not deleted as shards. Full captures, pixel comparisons, diagnostics, audits, failed attempts, and logs are in the [Round 10 report](round10/REPORT.md).
+
+The required cold W8 full-payload interleave averages **0.993 s** for Round 10 versus **1.022 s** for Round 9; one of two new samples is 1.014 s and the other is 0.972 s. A separate new-only repeat took **1.044 s**, making the three-new-run mean **1.010 s**; the local under-one-second margin is not robust. W1/W8 match all 44 output hashes, routes remain 5,642/5,642, far survival passes, and the bad-shell control still fails as intended. The six fixed captures and all four remote yaw views are included in the report.
 
 ## Round 9 — packed X batches and interval reuse (2026-10-05)
 
-The final executable averages **0.992 s** for the required cold W8 full-payload direct interleave, below the 1-second target. It measured **1.001 / 0.983 s** (mean 0.992) versus **1.217 / 1.150 s** (mean 1.184) for the local `16612ac` executable. A separate new-only cold repeat was 1.005 s; all three new runs average **0.996 s**. Every run generated 2,226/2,226 tiles with the full 110.90 MiB payload and zero cache hits. The fixed six captures are pixel-identical; the four X=500 remote views are being recaptured after confirming their original Y/Z camera position.
+The final executable averages **0.992 s** for the required cold W8 full-payload direct interleave, below the 1-second target. It measured **1.001 / 0.983 s** (mean 0.992) versus **1.217 / 1.150 s** (mean 1.184) for the local `16612ac` executable. A separate new-only cold repeat was 1.005 s; all three new runs average **0.996 s**. Every run generated 2,226/2,226 tiles with the full 110.90 MiB payload and zero cache hits. The six fixed captures are pixel-identical, and the four remote views were recaptured against Round 8 with the recorded camera positions; pixel deltas are in the [Round 9 report](round9/REPORT.md).
 
-The largest gain comes from batching eight fine packed-density samples along X: the LOD0/1 density-grid profiles fall 62%/61%. Reusing Z-axis intervals in the candidate mask cuts interval/sign work by about 20 ms and removes 18.35 MiB of retained candidate-axis allocation. The final W1/W8 matrix matches all 44 fields and W8 matches Round 8. Shape/look and far-survival audits pass; the injected bad-shell control still fails as intended. Full measurements, rejected probes, logs, and the pending remote-camera note are in [Round 9 report](round9/REPORT.md).
+The largest gain comes from batching eight fine packed-density samples along X: the LOD0/1 density-grid profiles fall 62%/61%. Reusing Z-axis intervals in the candidate mask cuts interval/sign work by about 20 ms and removes 18.35 MiB of retained candidate-axis allocation. The final W1/W8 matrix matches all 44 fields and W8 matches Round 8. Shape/look and far-survival audits pass; the injected bad-shell control still fails as intended. Full measurements, rejected probes, and logs are in the [Round 9 report](round9/REPORT.md).
 
 ## Round 8 — canonical AVX2 shape arithmetic (2026-10-05)