wip(assembler): M3 round 2 -- islands get real mass: deep crust, separate tapered roots, arched spans

Pinch cause: SmoothStep taper drove the underside radius to zero at one point. Version 2: depth 0.75x
diameter (thin-pad control at 0.30 fails), 3-4 seeded roots, bounded contour warp + roughness, one
off-route side island, 3 m rock spans. Bounds/route/W1-W8-W16 pass. Open: legacy probe metadata says
25 cm but production legacy is 12.5 cm -- round 3 must fix the import scale contract.

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

Standalone/Research/WorldAssemblerM3/Round2/REPORT.md

@@ -0,0 +1,172 @@
+# M3 Round 2 — FloatingIslands look and standalone gates
+
+**Result:** geometry version 2 gives the islands a deep, rough underside with separated tapered
+roots, variable-height placement, and broad arched connectors. Fresh matched CPU captures include
+overview, underside, edge/cliff, player-eye and route views. The measurable bounds, thickness,
+capsule-route and determinism gates pass. The native sample remains much denser than the owner
+field, so this is a deliberate routeable recipe rather than a visual-parity claim.
+
+## Owner recipe and scale
+
+The legacy settings do not request a fixed island count. `IslandDensity=0.5` is a per-hashed-XY-site
+existence chance. The owner parameters are spacing 112 legacy voxels, radius fields 24–48 legacy
+voxels, thickness ratio 0.6, vertical jitter/top flatten 0.6, roughness 3, blend radius 6 and boundary
+seal 4. The production `Source` constant and plugin-root `CLAUDE.md` say a legacy voxel is 12.5 cm;
+Core uses 25 cm. Round 2 therefore queries the legacy field at twice each Core coordinate and maps
+owner lengths to Core lengths at 0.5 scale:
+
+| Setting | Owner | Round 2 Core | Physical intent at production scale |
+|---|---:|---:|---:|
+| Site spacing | 112 voxels | 56 voxels | 14 m |
+| Radius | 24–48 voxels | 12–24 voxels | 3–6 m radius |
+| Roughness | 3 voxels | 1.5 voxels | 0.375 m amplitude bound |
+| Blend radius | 6 voxels | 3 voxels | 0.75 m |
+| Boundary seal | 4 voxels | 2 voxels | 0.5 m |
+
+There is a scale conflict to resolve before importing this recipe. The frozen owner probe metadata
+and `Standalone/Source` mirror say 25 cm, and the radius comment describes 12–24 m diameters at that
+scale. Production `Source` and `CLAUDE.md` say 12.5 cm, which makes the same radius fields 6–12 m
+diameters. The audit uses the production 12.5 cm field, with legacy anchor `(48, 16, -1904.035)`
+mapped to Core `(24, 8, -952.016)`. Round 3 must fix the snapshot producer/metadata and publish one
+import scale contract; it must not silently choose between these interpretations.
+
+The native planner requires three route-connected primary islands, then uses the 0.5 density roll
+for optional off-route sites; this seed made one, for four islands and three connectors. That is a
+stated routeability constraint, not an owner fixed-count setting. Geometry version 2 adds a crust
+depth bonus of 0.15 radius and a root depth of 0.75 radius to the owner 0.6-ratio body. Total measured
+depth is 0.75 of island diameter. The deeper roots and guaranteed route account for the local density
+increase below.
+
+## Visual review
+
+The round-1 reading was substantially right. Its low thickness ratio and `SmoothStep` taper drove the
+underside radius to zero at the bottom plane. The vanishing cross-section made the underside gradient
+collapse at a shared point and produced the radial crease/pinch. The old profile also had no
+independent roots, and its main route lay on a nearly straight chain.
+
+Version 2 gives the crust a broad, gently bowed bottom surface, then smooth-unions three or four
+seeded ellipsoid roots per island. Their lengths taper independently, so they do not converge at one
+apex. The bounded 6.3-Core-voxel contour warp breaks the silhouettes; 1.5-voxel surface roughness is
+faded to zero across a three-voxel surface band. The certified boxes include the contour warp,
+roughness and blend reach. The underside and edge captures show the additional mass and root lobes.
+
+The three primary islands form a bent, vertically fitted traversal. This seed also produces one
+higher side island that is not on the route. There are still three connectors because the route has
+to cross the three primary sites; they are 3 m wide, 1 m thick at the ends, and their underside sags
+by up to 2 m at midspan. Each is marked sharp-tag-ready. The captures show a rock-span profile rather
+than the old narrow straight rods, although the route-connected backbone remains visually legible as
+a chain. The route-island tops vary by about 0.67 m; the off-route island is about 1.46 m above the
+middle route island.
+
+The underside and edge captures are close views of the middle island so the larger cluster does not
+set their scale. The player-eye view is an offline perspective render from 1.8 m above the route
+floor, using the 25 cm Core voxel scale and the approximately 1.8 m player height. These are
+field-mesh review images, not Unreal camera, traversal, or collision evidence.
+
+| View | Capture |
+|---|---|
+| All five matched views | [Contact sheet](results/captures/contact_sheet.jpg) |
+| Cluster overview | [overview_matched.jpg](results/captures/overview_matched.jpg) |
+| Underside and roots | [underside_matched.jpg](results/captures/underside_matched.jpg) |
+| Thick crust and edge/cliff | [edge_cliff_matched.jpg](results/captures/edge_cliff_matched.jpg) |
+| Player-eye toward next island | [player_eye_matched.jpg](results/captures/player_eye_matched.jpg) |
+| Planned route | [route_matched.jpg](results/captures/route_matched.jpg) |
+
+The CPU renderer lowered its process priority. All images are 640×258 pixels individually; the
+contact sheet is 640×1,290 pixels and 71,668 bytes. Every JPEG is below 300 KB. Legacy/native pairs
+use the same target, camera and scale. The legacy side remains much smaller because it is the exact
+stochastic owner field at the same physical anchor, not a hand-expanded parity target.
+
+## Standalone gates
+
+The final `Target All` Core build and the audit wrapper passed. The wrapper samples the same 32,768
+physical query locations for legacy and native timing.
+
+| Gate | Final result |
+|---|---:|
+| Recipe / plan / manifest hashes | `5b8f27274bad89ba` / `ac33c9460b16337a` / `0db30684eaefa75e` |
+| Bounds | 515 boxes, 13,851 point samples, 74 unknown, **0 missed**; shrunken-box control detected |
+| Thickness | 4 islands; minimum depth/diameter 0.75 vs 0.70 required; all pass |
+| Thin-pad control | Ratio 0.2999; all 4 islands fail, control detected |
+| Route | 28 segments, 254 capsule centers, 69,088 capsule samples, 1,430 support probes; 0 blocked, 0 unsupported, 0 grade failures |
+| Minimum sampled density clearance | 78 subvoxels (0.305 Core voxel) |
+| Raised-route control (+8 voxels) | Detected: 254 unsupported centers, 0 blocked |
+| Field hash at W1 / W8 / W16 | `566b6115f1aaeb71` for all three; repeated plan stable |
+| Query timing, 32,768 locations | Legacy 64.64 ms; native 38.06 ms (0.589× legacy time, about 1.70× faster) |
+| Local sign / occupancy difference | 4.437% sign mismatch; 130 legacy-solid vs 1,380 native-solid samples |
+
+The sample contains 10.6× as many native-solid points as legacy (1,352 native-only and 102
+legacy-only). This is a local sparse-field sample, not a global volume estimate. It confirms the
+native density difference visible in the matched images. The extra route backbone, thick crust and
+deep roots are the stated reasons; the native field is not being presented as owner occupancy parity.
+
+The audit reparsed the unchanged existing M1 round-1 manifest hash `c87ae279b0e26f53` and
+SurfaceWorld round-4 manifest hash `8933693d9560ba9f`. No existing Core world evaluator,
+SurfaceWorld recipe or mesher implementation was edited. The complete M2 mesher hash matrix was not
+replayed in this round; the M2 report records the prior fixture mesh hash `9d58e0603f1bf9c7` and
+cliff tile hash `fc7bd7a3cf997992`.
+
+Full audit data is in [audit.json](results/audit.json); the successful wrapper ledger is
+[run_ledger.txt](results/run_ledger.txt). Mesh extracts used for the captures and audit are in the
+same `results` directory.
+
+## Round 3 — Unreal delivery plan
+
+1. Reconcile the 12.5 cm production field with the 25 cm frozen-probe metadata at the snapshot
+   producer. Add one explicit owner-to-Core length scale and reject payloads whose declared scale
+   conflicts with it.
+2. Add the recipe and seeded plan to the existing optional `world_assembly` import path. Parse via
+   `ResolvedWorldSnapshot::FromJson`; validate schema/geometry versions, recipe/plan/manifest hashes,
+   stable IDs, bounds and scale before generation. Preserve round-trip JSON and stable IDs.
+3. Prove imported generation in a null-RHI BuildHost commandlet. Compare canonical plan/field hashes
+   at W1/W8/W16 and exercise malformed-hash, unknown-version and mismatched-scale controls.
+4. Build the islands and sharp-tag-ready spans through the selected mesher path. Check chunk borders,
+   LOD transitions, route visibility and finite bounds in Unreal. Keep the Core field available to
+   server queries without requiring render meshes.
+5. Cook collision and sweep the actual player capsule over the imported route. Check floor support,
+   clearance, grade and connector landings against cooked collision, then hand the owner the imported
+   build, capture set, collision evidence, resolved scale note and exact source/hash manifest.
+
+Round 2 did not add Unreal import, cooked collision, BuildHost proof or owner hand-off evidence.
+
+## Run ledger and scope
+
+The final build used four compile workers at BelowNormal priority and completed in about 51 seconds.
+The final audit completed in 1.65 seconds and peaked at 58.8 MiB private memory; the CPU renderer
+completed in about 2 seconds. No run approached the 15-minute limit, no process crash occurred, and
+there was no `0xC0000142` retry. The build emitted the existing French `vswhere.exe`-not-found
+warning and `VoxelStrateManager.cpp` double-to-float conversion warnings before succeeding.
+
+Every failed iteration before that final pass:
+
+| Failed run | Result and correction |
+|---|---|
+| Core `Target All` compile | Failed on one unqualified fixed-scale `Q`; qualified it and rebuilt successfully. |
+| Audit 1 | One unsupported route center at segment 36. |
+| Audit 2 | Diagnostic repeat showed support just beyond the 2.5-voxel search depth (solid by 3); extended/adjusted support fitting. |
+| Audit 3 | Used a stale Core library after source changes and repeated the route failure; rebuilt Core before continuing. |
+| Audit 4 | Raising the connector deck fixed support but caused 17 blocked capsule samples; reduced the route margin. |
+| Audit 5 | No blocked samples, but one unsupported center at segment 49 near a connector end; tapered endpoint detail. |
+| Audit 6 | After converting owner dimensions to production physical scale, the old 0.62-radius landing produced 17 blocked and 9 unsupported samples. |
+| Audit 7 | Top-detail attenuation alone left those same counts. |
+| Audit 8 | Diagnostic repeat located the route near a root-lobe edge; moving the landing to 0.38 radius and using a 1-voxel deck lift fixed it. |
+| Renderer run | The first renderer pass stopped after four views when a raster row reached `HEIGHT`; clamped the row to `HEIGHT - 1`, then regenerated all five captures successfully. |
+
+The repeated pre-final audit processes peaked at 44.6–58.8 MiB. No editor, viewer or GPU was
+launched. No owner project file, `.uasset` or `Source/` file was opened for writing or changed; no
+project build was run. No commit, push or stash was made.
+
+### Source inventory
+
+New source file in this round:
+
+- `RenderFloatingIslandRound2.py` — CPU-only matched capture renderer.
+
+Modified source files:
+
+- `Standalone/Core/include/VoxelForgeCore/VoxelForgeCore.h`
+- `Standalone/Core/src/FloatingIslandsInternal.h`
+- `Standalone/Research/WorldAssemblerM3/Round1/FloatingIslandsAudit.cpp` (extended for round-2 physical-scale and visual audits)
+
+Documentation and dialogue changes are in `Standalone/Core/API.md`, `Standalone/DESIGN_WORLD_ASSEMBLER.md`,
+`CODEMAP.md` and `Standalone/Research/OwnerShapeWorld/gpu_field_port/LUNA_DIALOGUE.md`.

Standalone/Research/WorldAssemblerM3/Round2/RenderFloatingIslandRound2.py

@@ -0,0 +1,286 @@
+#!/usr/bin/env python3
+"""Small CPU-only matched views of the legacy and native FloatingIslands fields."""
+from __future__ import annotations
+
+import ctypes
+import json
+import math
+import struct
+from pathlib import Path
+
+from PIL import Image, ImageDraw
+
+
+ROOT = Path(__file__).resolve().parent
+RESULTS = ROOT / "results"
+CAPTURES = RESULTS / "captures"
+CAPTURES.mkdir(parents=True, exist_ok=True)
+WIDTH, HEIGHT = 320, 226
+BACKGROUND = (236, 240, 242)
+LEGACY_COLOR = (133, 127, 105)
+NATIVE_COLOR = (78, 135, 151)
+ROUTE_COLOR = (223, 72, 45)
+TILE_COLOR = (178, 60, 52)
+LOD_COLOR = (211, 136, 45)
+Q = 256.0
+
+
+def lower_priority() -> None:
+    if hasattr(ctypes, "windll"):
+        ctypes.windll.kernel32.SetPriorityClass(ctypes.windll.kernel32.GetCurrentProcess(), 0x00004000)
+
+
+def read_mesh(path: Path):
+    raw = path.read_bytes()
+    if len(raw) < 8 or raw[:4] != b"VFMI":
+        raise RuntimeError(f"invalid field mesh: {path}")
+    count = struct.unpack_from("<I", raw, 4)[0]
+    stride = struct.calcsize("<12f")
+    if len(raw) != 8 + count * stride:
+        raise RuntimeError(f"truncated field mesh: {path}")
+    triangles = [struct.unpack_from("<12f", raw, 8 + i * stride) for i in range(count)]
+    vertices = [tri[0:3] for tri in triangles]
+    vertices.extend(tri[3:6] for tri in triangles)
+    vertices.extend(tri[6:9] for tri in triangles)
+    return triangles, vertices
+
+
+def normalize(vector):
+    size = math.sqrt(sum(value * value for value in vector))
+    return tuple(value / size for value in vector) if size > 1.0e-12 else (0.0, 0.0, 1.0)
+
+
+def dot(a, b):
+    return sum(x * y for x, y in zip(a, b))
+
+
+def cross(a, b):
+    return (a[1] * b[2] - a[2] * b[1], a[2] * b[0] - a[0] * b[2], a[0] * b[1] - a[1] * b[0])
+
+
+def project(points, target, view_dir, scale=None):
+    direction = normalize(view_dir)
+    forward = tuple(-value for value in direction)
+    reference_up = (0.0, 1.0, 0.0) if abs(forward[2]) > 0.98 else (0.0, 0.0, 1.0)
+    right = normalize(cross(forward, reference_up))
+    up = normalize(cross(right, forward))
+    camera_points = []
+    for point in points:
+        delta = tuple(point[i] - target[i] for i in range(3))
+        camera_points.append((dot(delta, right), dot(delta, up), dot(delta, forward)))
+    if scale is None:
+        span_x = max((p[0] for p in camera_points), default=1.0) - min((p[0] for p in camera_points), default=0.0)
+        span_y = max((p[1] for p in camera_points), default=1.0) - min((p[1] for p in camera_points), default=0.0)
+        scale = 0.86 * min((WIDTH - 30) / max(span_x, 1.0e-6), (HEIGHT - 30) / max(span_y, 1.0e-6))
+    pixels = [
+        (WIDTH * 0.5 + point[0] * scale, 31 + (HEIGHT - 31) * 0.5 - point[1] * scale, point[2])
+        for point in camera_points
+    ]
+    return pixels, scale, right, up, forward
+
+
+def common_scale(mesh_pair, target, view_dir):
+    vertices = mesh_pair[0][1] + mesh_pair[1][1]
+    _, scale, _, _, _ = project(vertices, target, view_dir)
+    return scale
+
+
+def projected_line(a, b, target, basis, scale):
+    right, up, forward = basis
+    points = []
+    for point in (a, b):
+        delta = tuple(point[i] - target[i] for i in range(3))
+        points.append((WIDTH * 0.5 + dot(delta, right) * scale,
+                       31 + (HEIGHT - 31) * 0.5 - dot(delta, up) * scale))
+    return points
+
+
+def render(mesh, mesh_pair, method, target, view_dir, title, subtitle, *, route=None,
+           guide=None, camera_eye=None, framing_radius=None):
+    triangles, vertices = mesh
+    perspective = camera_eye is not None
+    if perspective:
+        forward = normalize(tuple(target[i] - camera_eye[i] for i in range(3)))
+        reference_up = (0.0, 1.0, 0.0) if abs(forward[2]) > 0.98 else (0.0, 0.0, 1.0)
+        right = normalize(cross(forward, reference_up))
+        up = normalize(cross(right, forward))
+        scale = WIDTH * 0.91
+        origin = camera_eye
+    else:
+        if framing_radius is None:
+            scale = common_scale(mesh_pair, target, view_dir)
+        else:
+            scale = 0.86 * min((WIDTH - 30) / (2.0 * framing_radius),
+                               (HEIGHT - 30) / (2.0 * framing_radius))
+        _, _, right, up, forward = project((), target, view_dir, scale)
+        origin = target
+
+    def screen(point):
+        delta = tuple(point[i] - origin[i] for i in range(3))
+        cx, cy, depth = dot(delta, right), dot(delta, up), dot(delta, forward)
+        if perspective:
+            if depth <= 1.0:
+                return None
+            return (WIDTH * 0.5 + cx * scale / depth,
+                    31 + (HEIGHT - 31) * 0.5 - cy * scale / depth, depth)
+        return (WIDTH * 0.5 + cx * scale,
+                31 + (HEIGHT - 31) * 0.5 - cy * scale, depth)
+
+    background = Image.new("RGB", (WIDTH, HEIGHT + 32), BACKGROUND)
+    draw = ImageDraw.Draw(background)
+    draw.rectangle((0, 0, WIDTH - 1, 31), fill=(34, 46, 54))
+    draw.text((8, 4), title, fill=(250, 252, 253))
+    draw.text((8, 17), subtitle, fill=(190, 207, 215))
+    pixels = background.load()
+    zbuffer = [float("inf")] * (WIDTH * (HEIGHT - 31))
+    camera_fill = forward if perspective else normalize(view_dir)
+    light = normalize((-0.25 + 1.2 * camera_fill[0],
+                       -0.35 + 1.2 * camera_fill[1],
+                       0.45 + 1.2 * camera_fill[2]))
+    base = LEGACY_COLOR if method == "legacy" else NATIVE_COLOR
+    for tri in triangles:
+        points = [screen(tri[offset:offset + 3]) for offset in (0, 3, 6)]
+        if any(point is None for point in points):
+            continue
+        x0 = max(0, int(math.floor(min(p[0] for p in points))))
+        x1 = min(WIDTH - 1, int(math.ceil(max(p[0] for p in points))))
+        y0 = max(31, int(math.floor(min(p[1] for p in points))))
+        y1 = min(HEIGHT - 1, int(math.ceil(max(p[1] for p in points))))
+        if x0 > x1 or y0 > y1:
+            continue
+        (ax, ay, az), (bx, by, bz), (cx, cy, cz) = points
+        denominator = (by - cy) * (ax - cx) + (cx - bx) * (ay - cy)
+        if abs(denominator) < 1.0e-10:
+            continue
+        normal = normalize(tri[9:12])
+        shade = 0.50 + 0.50 * max(0.0, dot(normal, light))
+        color = tuple(max(0, min(255, int(channel * shade))) for channel in base)
+        for y in range(y0, y1 + 1):
+            row = (y - 31) * WIDTH
+            for x in range(x0, x1 + 1):
+                sx, sy = x + 0.5, y + 0.5
+                u = ((by - cy) * (sx - cx) + (cx - bx) * (sy - cy)) / denominator
+                v = ((cy - ay) * (sx - cx) + (ax - cx) * (sy - cy)) / denominator
+                w = 1.0 - u - v
+                if u < -1e-8 or v < -1e-8 or w < -1e-8:
+                    continue
+                if perspective:
+                    depth = 1.0 / (u / az + v / bz + w / cz)
+                else:
+                    depth = u * az + v * bz + w * cz
+                index = row + x
+                if depth < zbuffer[index]:
+                    zbuffer[index] = depth
+                    pixels[x, y] = color
+
+    basis = (right, up, forward)
+    if guide:
+        origin = guide["origin_voxels"]
+        top = guide["top_voxels"]
+        for x_delta, color, label in ((32.0, TILE_COLOR, "32v tile"), (64.0, LOD_COLOR, "64v LOD1")):
+            line = projected_line((origin[0] + x_delta, origin[1] - 190.0, top - 65.0),
+                                  (origin[0] + x_delta, origin[1] + 190.0, top + 14.0),
+                                  target, basis, scale)
+            draw.line(line, fill=color, width=2)
+            draw.text((max(2, min(WIDTH - 55, int(line[1][0]) + 2)),
+                       max(33, min(HEIGHT - 11, int(line[1][1]) - 12))), label, fill=color)
+    if route:
+        points = [screen(point) for point in route]
+        points = [point[:2] for point in points if point is not None]
+        if len(points) >= 2:
+            draw.line(points, fill=(250, 248, 239), width=5, joint="curve")
+            draw.line(points, fill=ROUTE_COLOR, width=3, joint="curve")
+            for index in (0, len(points) - 1):
+                x, y = points[index]
+                draw.ellipse((x - 4, y - 4, x + 4, y + 4), fill=(250, 248, 239), outline=ROUTE_COLOR, width=2)
+    return background
+
+
+def titled_pair(left, right, category):
+    pair = Image.new("RGB", (WIDTH * 2, HEIGHT + 32), (220, 226, 229))
+    pair.paste(left, (0, 0))
+    pair.paste(right, (WIDTH, 0))
+    draw = ImageDraw.Draw(pair)
+    draw.rectangle((0, 0, WIDTH * 2 - 1, HEIGHT + 31), outline=(72, 84, 91), width=1)
+    draw.text((8, HEIGHT + 14), f"{category} | same target, camera and scale", fill=(45, 55, 61))
+    return pair
+
+
+def main():
+    lower_priority()
+    data = json.loads((RESULTS / "audit.json").read_text(encoding="utf-8"))
+    anchor = data["comparison_anchor_core_voxels"]
+    route = data["route_points_core_voxels"]
+    profiles = data["island_profiles"]
+    middle_island = profiles[1]
+    cluster = {
+        "legacy": read_mesh(RESULTS / "cluster_legacy.vfmesh"),
+        "native": read_mesh(RESULTS / "cluster_native.vfmesh"),
+    }
+    detail = {
+        "legacy": read_mesh(RESULTS / "detail_legacy.vfmesh"),
+        "native": read_mesh(RESULTS / "detail_native.vfmesh"),
+    }
+    a = anchor
+    scenes = (
+        ("overview", "Cluster overview", cluster, (a[0], a[1], a[2] - 28.0), (8.0, -10.0, 7.0), None, None),
+        ("underside", "Cluster underside and lobed roots", detail,
+            (middle_island["center_voxels"][0], middle_island["center_voxels"][1],
+             middle_island["bottom_voxels"] - 6.0), (0.12, -0.08, -1.0), None, None),
+        ("edge_cliff", "Thick crust and bounded rock detail", detail,
+            (middle_island["center_voxels"][0], middle_island["center_voxels"][1],
+             middle_island["top_voxels"] - middle_island["radius_voxels"] * 0.65),
+            (1.0, -0.25, 0.18), None, None),
+        ("route", "Supported route across the cluster", cluster,
+            (middle_island["center_voxels"][0], middle_island["center_voxels"][1],
+             middle_island["top_voxels"] - 18.0), (8.0, -10.0, 7.0), route, None),
+    )
+    pairs = []
+    for key, title, meshes, target, camera, path, grid in scenes:
+        framing_radius = 28.0 if key in ("underside", "edge_cliff") else None
+        left = render(meshes["legacy"], (meshes["legacy"], meshes["native"]), "legacy",
+                      target, camera, title, "Legacy bridge | exact owner field", route=None,
+                      guide=grid, framing_radius=framing_radius)
+        route_caption = "Core native | planner route" if path else "Core native | matched field view"
+        right = render(meshes["native"], (meshes["legacy"], meshes["native"]), "native",
+                       target, camera, title, route_caption, route=path, guide=grid,
+                       framing_radius=framing_radius)
+        pair = titled_pair(left, right, title)
+        out = CAPTURES / f"{key}_matched.jpg"
+        pair.save(out, quality=82, optimize=True)
+        pairs.append(pair)
+        print(f"capture={out.name} bytes={out.stat().st_size} size={pair.size[0]}x{pair.size[1]}")
+
+    eye_index = 5
+    eye_route = route[eye_index]
+    eye = (eye_route[0], eye_route[1], eye_route[2] + 3.68)
+    next_island = profiles[1]
+    look = (next_island["center_voxels"][0], next_island["center_voxels"][1],
+            next_island["top_voxels"] + 5.0)
+    perspective_pair = []
+    for method, meshes in (("legacy", cluster), ("native", cluster)):
+        perspective_pair.append(render(meshes[method], (meshes["legacy"], meshes["native"]), method,
+            look, (0.0, 0.0, -1.0), "Island-edge player view", "Eye at 1.8 m | route span to next island",
+            route=None, camera_eye=eye))
+    eye_pair = titled_pair(perspective_pair[0], perspective_pair[1], "Island-edge player view")
+    eye_path = CAPTURES / "player_eye_matched.jpg"
+    eye_pair.save(eye_path, quality=80, optimize=True)
+    pairs.insert(3, eye_pair)
+    print(f"capture={eye_path.name} bytes={eye_path.stat().st_size} size={eye_pair.size[0]}x{eye_pair.size[1]}")
+
+    sheet = Image.new("RGB", (WIDTH * 2, (HEIGHT + 32) * len(pairs)), (218, 225, 228))
+    for row, pair in enumerate(pairs):
+        sheet.paste(pair, (0, row * (HEIGHT + 32)))
+    sheet_path = CAPTURES / "contact_sheet.jpg"
+    sheet.save(sheet_path, quality=76, optimize=True)
+    oversized = [(path.name, path.stat().st_size) for path in sorted(CAPTURES.glob("*.jpg"))
+                 if path.stat().st_size >= 300_000]
+    for path in sorted(CAPTURES.glob("*.jpg")):
+        print(f"capture_size={path.name} bytes={path.stat().st_size}")
+    if oversized:
+        raise RuntimeError(f"captures exceed the 300 KB limit: {oversized}")
+    print(f"contact_sheet={sheet_path.name} bytes={sheet_path.stat().st_size} views=5 matched_pairs=5")
+
+
+if __name__ == "__main__":
+    main()

Standalone/Research/WorldAssemblerM3/Round2/results/audit.json

@@ -0,0 +1,29 @@
+{
+  "recipe_hash":"5b8f27274bad89ba",
+  "plan_hash":"ac33c9460b16337a",
+  "manifest_hash":"0db30684eaefa75e",
+  "unchanged_existing_hashes":{"m1_round1_manifest":"c87ae279b0e26f53","surfaceworld_round4_manifest":"8933693d9560ba9f"},
+  "geometry_version":2,
+  "owner_voxel_size_cm":12.5,
+  "core_voxel_size_cm":25,
+  "owner_to_core_length_scale":0.5,
+  "plan_islands":4,
+  "plan_connectors":3,
+  "route_points":29,
+  "legacy_anchor_owner_voxels":[48,16,-1904.03515625],
+  "comparison_anchor_core_voxels":[24,8,-952.015625],
+  "legacy_anchor_plateau_samples":25,
+  "recipe_parameters":{"seed":0,"primary_island_count":3,"density_q16":32768,"spacing_voxels":56,"radius_voxels":[12,24],"thickness_ratio_q16":39322,"vertical_jitter_q16":39322,"top_flatten_q16":39322,"contour_warp_voxels":6.30078125,"crust_depth_bonus_q16":9830,"root_depth_ratio_q16":49152,"roughness_voxels":1.5,"blend_voxels":3,"bridge_half_width_voxels":6,"bridge_thickness_voxels":4},
+  "stable_island_ids":["e6e461e18b3a9c87","2bc1820baba047f6","adc259cbe2a395e9","7855fb022cc561c9"],
+  "stable_connector_ids":["eaa5dd310e846520","034023599b466c51","4962fa885545fb26"],
+  "island_profiles":[{"id":"e6e461e18b3a9c87","ordinal":0,"center_voxels":[-15.19921875,25.2734375,-957.5078125],"radius_voxels":14.08203125,"top_voxels":-954.69140625,"bottom_voxels":-975.81640625,"depth_to_diameter":0.750069348128},{"id":"2bc1820baba047f6","ordinal":1,"center_voxels":[24,-12.2265625,-954.8984375],"radius_voxels":14.42578125,"top_voxels":-952.015625,"bottom_voxels":-973.65625,"depth_to_diameter":0.75006769564},{"id":"adc259cbe2a395e9","ordinal":2,"center_voxels":[63.19921875,26.28515625,-958.57421875],"radius_voxels":20.23828125,"top_voxels":-954.52734375,"bottom_voxels":-984.88671875,"depth_to_diameter":0.750048253233},{"id":"7855fb022cc561c9","ordinal":3,"center_voxels":[-1.19921875,-68.2265625,-950.75390625],"radius_voxels":22.8125,"top_voxels":-946.19140625,"bottom_voxels":-980.41015625,"depth_to_diameter":0.75}],
+  "bounds":{"passed":true,"negative_control_detected":true,"boxes":515,"samples":13851,"unknown":74,"missed":0,"island_boxes":4,"connector_boxes":3,"void_boxes_proved_air":1},
+  "thickness":{"passed":true,"islands_tested":4,"islands_below_minimum":0,"minimum_depth_to_diameter_ratio":0.75,"required_depth_to_diameter_ratio":0.7,"minimum_island_id":"7855fb022cc561c9"},
+  "thin_pad_negative_control":{"detected":true,"failed_islands":4,"minimum_depth_to_diameter_ratio":0.299891686975},
+  "route":{"passed":true,"segments":28,"centers":254,"capsule_samples":69088,"blocked":0,"support_samples":1430,"unsupported":0,"grade_failures":0,"minimum_clearance_subvoxels":78},
+  "raised_route_negative_control":{"detected":true,"passed":false,"unsupported":254,"blocked":0},
+  "determinism":{"W1":"566b6115f1aaeb71","W8":"566b6115f1aaeb71","W16":"566b6115f1aaeb71","plan_repeat":true,"hash_samples":32768},
+  "timing":{"same_queries":32768,"legacy_ms":64.6405,"native_ms":38.0581,"native_over_legacy":0.588765557197,"sign_mismatches":1454,"sign_mismatch_percent":4.43725585938,"legacy_solid_samples":130,"native_solid_samples":1380,"legacy_only_solid_samples":102,"native_only_solid_samples":1352},
+  "mesh":[{"name":"cluster_legacy","grid_samples":36613,"triangles":1532,"bytes":73544},{"name":"cluster_native","grid_samples":36613,"triangles":8764,"bytes":420680},{"name":"detail_legacy","grid_samples":207025,"triangles":4124,"bytes":197960},{"name":"detail_native","grid_samples":207025,"triangles":27794,"bytes":1334120}],
+  "route_points_core_voxels":[[-15.19921875,25.2734375,-950.4375],[-12.3984375,22.59375,-949.54296875],[-9.59765625,19.91796875,-949.453125],[-6.80078125,17.23828125,-949.3125],[-4,14.55859375,-949.23046875],[-1.19921875,11.87890625,-949.0234375],[1.6015625,9.203125,-948.734375],[4.40234375,6.5234375,-948.53515625],[7.19921875,3.84375,-948.328125],[10,1.16796875,-947.99609375],[12.80078125,-1.51171875,-947.6953125],[15.6015625,-4.19140625,-947.5390625],[18.3984375,-6.87109375,-947.33984375],[21.19921875,-9.546875,-947.93359375],[24,-12.2265625,-947.87109375],[26.80078125,-9.4765625,-946.87109375],[29.6015625,-6.7265625,-946.93359375],[32.3984375,-3.97265625,-947.31640625],[35.19921875,-1.22265625,-947.72265625],[38,1.52734375,-948.17578125],[40.80078125,4.27734375,-948.34375],[43.6015625,7.03125,-948.48046875],[46.3984375,9.78125,-948.72265625],[49.19921875,12.53125,-949.046875],[52,15.28125,-949.2109375],[54.80078125,18.03125,-949.265625],[57.59765625,20.78515625,-949.52734375],[60.3984375,23.53515625,-950.3984375],[63.19921875,26.28515625,-950.265625]]
+}