pixal3d_mrp_mlx/scripts/image_to_mesh.py
m3ultra bf8b1a35d5 Close all four open items: MoGe camera, manifold remesh, winding, UV bake
THE DECIMATION FLOOR WAS MISDIAGNOSED. I attributed it to ~180k boundary edges. It is
non-manifold edges. Measured on the shipped 500k mesh:

  boundary edges     32,370
  NON-MANIFOLD       81,112     <- the actual blocker, 2.5x more

Quadric decimation cannot collapse an edge shared by more than two faces. Upstream's
own fix (fill_holes, via CUDA-only cumesh) targets boundaries and caps at
max_hole_perimeter=3e-2, so it was never going to help: trimesh's equivalent moved
boundaries 32,370 -> 30,990 and the floor only 214k -> 210k. That falsified it.

--manifold: voxelise -> fill -> marching cubes. Removes BOTH classes at once and so
closes three of the four items in one change:

  as shipped   499,984 faces  bnd 32,370  nonmani 81,112  watertight=F  IoU 0.969
  remeshed   1,178,142 faces  bnd      0  nonmani      0  watertight=T  IoU 0.949
  -> 20k        19,998 faces  bnd      0  winding consistent            IoU 0.956

25x smaller, fully manifold, consistent winding, for 1.3% silhouette IoU. Lossy by
design - it gives up the dual grid's open-surface representation - so it is opt-in.

UV BAKE is unblocked by the same change: its cost is driven by face count, not by
remesh. 5.0s at 20k faces against >20min at 214k. No longer offline-only when paired
with manifold.

THE SCALING TRAP, worth knowing: marching_cubes returns vertices in VOXEL INDEX space.
Translating without apply_scale(pitch) leaves the mesh ~292x too large. It still
exports and renders as a plausible object; it silhouettes at IoU 0.08. That is how it
was caught.

MoGe-2 CAMERA is now wired and is the default, matching upstream; --fixed-fov keeps
the old constant. It runs once per image in torch/MPS, ~0.4s after load.

Reporting this one straight: it did NOT improve the samples. On 1_img, fixed 49.1 deg
scored 0.893 and MoGe's 29.7 deg scored 0.883. Two caveats keep it as the default
anyway - the silhouette metric projects with the SAME FOV used to generate, so a
wrong-but-consistent camera can still score well and the metric cannot fully arbitrate
camera correctness; and the bundled samples are synthetic renders, not the photographs
MoGe reads. Real photos are the intended input here, and upstream estimates too. But
the constant is one flag away and the measurement is on record rather than assumed.

Operator gains manifold, divisions, fixed_fov. README and PROFILE.md corrected where
they repeated the boundary-edge claim.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-08-03 20:39:45 +10:00

221 lines
9.9 KiB
Python

"""Image -> GLB through the full Pixal3D cascade, with a silhouette check.
Usage: python scripts/image_to_mesh.py IMAGE [-o OUT.glb] [--fov RAD] [--seed N]
The silhouette IoU is the acceptance test: re-project the mesh through the same camera
and compare against the input matte. Pixal3D's entire claim is pixel alignment, so a
run that completes with a poor IoU has failed even though nothing raised.
"""
import argparse
import json
import sys
import time
from pathlib import Path
import mlx.core as mx
import numpy as np
import trimesh
REPO = Path(__file__).resolve().parents[1]
sys.path.insert(0, str(REPO))
from pixal3d_mlx.mesh import PBR_ATTR_LAYOUT, to_camera_frame # noqa: E402
from pixal3d_mlx.models import load_all, normalization # noqa: E402
from pixal3d_mlx.pipeline import DEFAULT_FOV, image_to_mesh # noqa: E402
DEFAULT_IMAGE = REPO / "upstream" / "Pixal3D" / "assets" / "images" / "0_img.png"
def silhouette_iou(mesh_or_vertices, image_path, fov, res=512, samples=3_000_000):
"""Re-project the mesh through the generating camera; IoU against the input matte.
Points are SAMPLED UNIFORMLY OVER THE SURFACE, not taken from the vertex list.
Projecting vertices makes the score depend on tessellation: the same shape scored
0.965 at 227k vertices and 0.790 at 134k, purely because a sparser point cloud
leaves holes inside its own silhouette. That would fail good assets for the crime
of being decimated. Fixed-count surface sampling makes density a constant of the
metric instead of a property of the mesh.
Points MUST be rotated into the camera frame — o_voxel returns geometry in the
voxel-grid frame while ProjGrid rotates its lattice before projecting.
"""
from PIL import Image
from scipy.ndimage import binary_dilation
from pixal3d_mlx.cond import preprocess_image
from pixal3d_mlx.proj import _FRONT_VIEW, distance_from_fov, project_points
img = Image.open(image_path)
if img.mode != "RGBA":
return None
matte = np.asarray(preprocess_image(img).convert("L").resize((res, res))) > 8
if isinstance(mesh_or_vertices, trimesh.Trimesh):
pts3, _ = trimesh.sample.sample_surface(mesh_or_vertices, samples)
else:
pts3 = np.asarray(mesh_or_vertices) # bare vertices: caller accepts the bias
tm = _FRONT_VIEW.copy()
tm[1, 3] = -distance_from_fov(fov, 1.0, res)
pts = to_camera_frame(pts3).astype(np.float32)[None]
px, _, _ = project_points(mx.array(pts), mx.array(tm[None]), fov, res)
px = np.asarray(px)[0].astype(int)
keep = (px[:, 0] >= 0) & (px[:, 0] < res) & (px[:, 1] >= 0) & (px[:, 1] < res)
proj = np.zeros((res, res), bool)
proj[px[keep, 1], px[keep, 0]] = True
proj = binary_dilation(proj, np.ones((3, 3), bool))
return (proj & matte).sum() / (proj | matte).sum(), proj, matte
def gate(mesh, a, info):
"""Silhouette check + non-zero exit. Shared by the textured and geometry paths."""
if a.min_iou <= 0:
return
got = silhouette_iou(mesh, a.image, a.fov)
if got is None:
print("(input has no alpha matte — skipping silhouette check)")
return
iou = float(got[0])
info["silhouette_iou"] = round(iou, 4)
print(f"silhouette IoU {iou:.3f}")
if iou < a.min_iou:
# completing is not succeeding — a run can finish cleanly and still have
# produced a blob that does not match the input at all
print(f"FAIL: IoU {iou:.3f} < {a.min_iou} — not tracking the input")
if a.json:
Path(a.json).write_text(json.dumps(info, indent=2))
sys.exit(1)
def main():
ap = argparse.ArgumentParser()
ap.add_argument("image", nargs="?", default=str(DEFAULT_IMAGE))
ap.add_argument("-o", "--output", default=str(REPO / "output.glb"))
ap.add_argument("--fov", type=float, default=None,
help="camera FOV in radians; omitted = estimate per image with "
"MoGe-2 (upstream's behaviour)")
ap.add_argument("--fixed-fov", action="store_true",
help=f"skip MoGe and use the constant {DEFAULT_FOV:.4f} rad")
ap.add_argument("--manifold", action="store_true",
help="voxel-remesh to a watertight manifold surface — the only way "
"past the ~214k non-manifold decimation floor; lossy, opt-in")
ap.add_argument("--divisions", type=int, default=256,
help="voxel resolution for --manifold")
ap.add_argument("--seed", type=int, default=0)
ap.add_argument("--target-faces", type=int, default=100_000,
help="face budget after cleanup; 0 disables decimation")
ap.add_argument("--raw", action="store_true",
help="skip cleanup entirely and export the decoder output as-is")
ap.add_argument("--min-iou", type=float, default=0.85,
help="fail the run below this silhouette IoU; 0 disables the gate")
ap.add_argument("--json", help="write run metadata here")
ap.add_argument("--texture", action="store_true",
help="run the texture stage and bake PBR maps through o_voxel")
ap.add_argument("--texture-size", type=int, default=2048)
ap.add_argument("--baker", choices=("vertex", "uv"), default="vertex",
help="vertex = seconds, base colour only (default); "
"uv = o_voxel unwrap + full PBR maps, but >20min CPU here")
a = ap.parse_args()
t = time.time()
models = load_all(with_texture=a.texture)
print(f"loaded models ({time.time() - t:.1f}s, lazy — weights fault in on first use)")
# Camera FIRST — everything downstream is placed by it. A wrong FOV reconstructs
# at the wrong depth scale and fails silently, so estimation is the default.
if a.fov is None:
if a.fixed_fov:
a.fov = DEFAULT_FOV
print(f" camera fixed {a.fov:.4f} rad ({np.degrees(a.fov):.1f} deg)")
else:
from pixal3d_mlx.camera import camera_for
t = time.time()
cam = camera_for(a.image)
a.fov = cam["camera_angle_x"]
print(f" camera MoGe-2 {a.fov:.4f} rad ({np.degrees(a.fov):.1f} deg), "
f"distance {cam['distance']:.3f} {time.time() - t:5.1f}s")
info_fov = a.fov
v, f, info = image_to_mesh(a.image, models, camera_angle_x=a.fov, seed=a.seed,
normalization=normalization("shape"),
tex_normalization=normalization("tex") if a.texture else None,
texture=a.texture)
info["fov"] = round(float(info_fov), 6)
info["fov_source"] = "fixed" if a.fixed_fov else "moge2"
subs = info.pop("subs", None)
info.pop("hr_slat", None)
tex_voxels = info.pop("tex_voxels", None)
if a.texture and tex_voxels is not None:
# CLEAN BEFORE BAKING. o_voxel's remesh+unwrap on the raw ~8M-face mesh hangs
# (killed at 20min); the trellis2 lane hit the same wall and its operator note
# says the uncapped bake peaks at 75GB. Welding and stripping floaters first
# makes it tractable, and the baker samples the attribute VOLUME at mesh
# positions, so a decimated mesh still gets correct colours.
from pixal3d_mlx.cleanup import clean
from pixal3d_mlx.mesh import bake_vertex_colors, to_glb
t = time.time()
pre, _ = clean(v.cpu().numpy(), f.cpu().numpy(),
target_faces=a.target_faces or 500_000,
manifold=a.manifold, divisions=a.divisions)
print(f" pre-bake {len(pre.faces):,} faces {time.time() - t:6.1f}s")
t = time.time()
if a.baker == "vertex":
mesh, bstat = bake_vertex_colors(pre, tex_voxels, info["output_resolution"])
info["bake"] = bstat
print(f" bake vertex colours, {bstat['hit_rate']:.1%} of vertices hit"
f" {time.time() - t:6.1f}s")
else:
import torch
scene = to_glb(torch.from_numpy(np.asarray(pre.vertices, np.float32)),
torch.from_numpy(np.asarray(pre.faces, np.int32)),
tex_voxels, PBR_ATTR_LAYOUT, info["output_resolution"],
texture_size=a.texture_size,
decimation_target=a.target_faces or 500_000)
mesh = scene if isinstance(scene, trimesh.Trimesh) else scene.dump(concatenate=True)
print(f" bake UV {len(mesh.faces):,} faces, {a.texture_size}px "
f"{time.time() - t:6.1f}s")
info["baker"] = a.baker
mesh.export(a.output)
info["faces"], info["vertices"] = len(mesh.faces), len(mesh.vertices)
print(f"\nTOTAL {info['seconds']}s peak {info['peak_gb']} GB -> {a.output}")
# The gate applies to TEXTURED runs too. Skipping it here would mean the
# textured path — the one most likely to be used for real assets — is the only
# one that can ship a blob silently.
gate(mesh, a, info)
if a.json:
Path(a.json).write_text(json.dumps(info, indent=2))
return
if a.raw:
mesh = trimesh.Trimesh(v.cpu().numpy(), f.cpu().numpy(), process=False)
else:
from pixal3d_mlx.cleanup import clean
t = time.time()
mesh, stats = clean(v.cpu().numpy(), f.cpu().numpy(),
target_faces=a.target_faces or None,
manifold=a.manifold, divisions=a.divisions)
print(f" cleanup {time.time() - t:6.1f}s")
info["cleanup"] = stats
mesh.export(a.output)
info["faces"] = len(mesh.faces)
info["vertices"] = len(mesh.vertices)
print(f"\nTOTAL {info['seconds']}s peak {info['peak_gb']} GB "
f"{len(mesh.faces):,} faces -> {a.output}")
gate(mesh, a, info)
if a.json:
Path(a.json).write_text(json.dumps(info, indent=2))
if __name__ == "__main__":
main()