wip(assembler): M3 round 3 -- separated islands + Unreal island route (owner build 18:46)

Islands read as separate (91.89% open sky at mid-height), underside cavity closed,
rim teeth were capture shading. Island manifest import + BuildHost proof passed on the
pre-cache tree; M1 SurfaceWorld and M2 V1/V2 gates passed.

Known open: final XY-column cache compiled in the owner build but is unverified;
island LOD0 p95 ~155 ms misses the 133 ms sprint deadline (legacy ~43 ms); the island
manifest only fits the BuildHost fixture layout (owner's world: Core init fails,
empty world) -> round 4 builds plans inside Unreal.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01MmSoF64BbRFmk9C3p1VsBJ
This commit is contained in:
Fr0zkaandClaude Opus 5.5 committed 2026-10-10 20:53:03 +02:00
1 parent 4afea7c802
commit 48fa028086
37 files changed
+3788 -107

No files matched your search

@@ -1063,3 +1063,8 @@ match; the M2 mesher hash matrix was not replayed. No editor/viewer, project bui
New source file this round: `RenderFloatingIslandRound2.py`.
— Luna
## World assembler M3 round 3 — 2026-10-10
The owner's scale is settled at 12.5 cm per legacy voxel and 25 cm per Core voxel (owner-to-Core length ratio 0.5). The Round 3 shape review reads as four separated masses; sampled mid-height open sky is 91.89% of the XY bounding box. The underside center is solid, and the crust rim no longer shows the earlier sawtooth appearance in the matched offline captures.
The manifest import and Core route passed the recorded BuildHost checks, along with SurfaceWorld M1 and HybridV1/V2 M2. Island/span collision contact, anchors/height/strate, Blueprint carve/fill, Core-off legacy behavior, W1/W8/W16 determinism, and tile hash parity passed on the verified source state. The recorded LOD0 island and mixed-tile sprint p95s still exceeded the 133.333 ms deadline by about 24 ms and 20 ms. A follow-up XY-column cache was added to target meshing/density cost but could not be compiled after the Windows command runner began returning `0xC0000142`; therefore I am not calling the current tree verified. The detailed measurements, hand-off, and run ledger are in `Standalone/Research/WorldAssemblerM3/Round3/REPORT.md`.