Spaces:
Runtime error
Runtime error
File size: 2,022 Bytes
cc7edd6 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 | """Diagnostic replay of saved tensors; capture or substitute pre-simplification remesh."""
from pathlib import Path
import json
import numpy as np
def replay(raw_path, layout_path, output, filtered_path=None):
import torch
import cumesh
import o_voxel
output = Path(output)
output.mkdir(parents=True, exist_ok=True)
with np.load(raw_path, allow_pickle=False) as data:
arrays = {key: data[key].copy() for key in data.files}
layout = {key: slice(*value) for key, value in json.loads(Path(layout_path).read_text()).items()}
original = cumesh.remeshing.remesh_narrow_band_dc
class Captured(Exception):
pass
def interception(*args, **kwargs):
if filtered_path:
with np.load(filtered_path, allow_pickle=False) as data:
return (torch.from_numpy(data['vertices'].astype(np.float32)).cuda(),
torch.from_numpy(data['faces'].astype(np.int32)).cuda())
vertices, faces = original(*args, **kwargs)
np.savez_compressed(output / 'pre-simplify.npz', vertices=vertices.cpu().numpy(), faces=faces.cpu().numpy())
raise Captured()
cumesh.remeshing.remesh_narrow_band_dc = interception
try:
result = o_voxel.postprocess.to_glb(
vertices=torch.from_numpy(arrays['vertices']).cuda(), faces=torch.from_numpy(arrays['faces']).cuda(),
attr_volume=torch.from_numpy(arrays['attrs']).cuda(), coords=torch.from_numpy(arrays['coords']).cuda(),
attr_layout=layout, grid_size=int(arrays['grid_size']) if arrays['grid_size'].ndim == 0
else arrays['grid_size'].tolist(), aabb=[[-.5, -.5, -.5], [.5, .5, .5]],
decimation_target=50000, texture_size=1024, remesh=True, remesh_band=1, remesh_project=0)
result.export(output / 'filtered-remesh.glb')
except Captured:
return str(output / 'pre-simplify.npz')
finally:
cumesh.remeshing.remesh_narrow_band_dc = original
return str(output / 'filtered-remesh.glb')
|