# tag=r-robot-1 start=2026-11-03T18:15:14+01:00 cmd=<home>/supratex-0311/venv/bin/python scripts/run_example.py --config <home>/supratex-0311/ready/models.json --input <home>/supratex-0311/in-rgb/robot-gpu.png --output <home>/supratex-0311/out/r-robot-1.glb
[    0.61] [Supra-Tex inference] attention=sdpa; sparse_attention=sdpa; low_vram=False
[    0.61] [1/2] Foreground, MoGe-2, geometry, material maps -> <home>/supratex-0311/out/r-robot-1_work/input
[    0.61] [Supra-Tex inference] attention=sdpa; sparse_attention=sdpa; low_vram=False
[   71.01] [2/2] Surface observations, PBR completion, GLB export -> <home>/supratex-0311/out/r-robot-1.glb
[   71.01] [Supra-Tex inference] attention=sdpa; sparse_attention=sdpa; low_vram=False
[   71.37] [SPARSE] Conv backend: flex_gemm; Attention backend: sdpa; Batch broadcast: direct
[   71.37] Floaters: removed 0 of 136 components (0 of 25819718 faces)
[   92.63] Observation: silhouette IoU 0.980 (geometry inside image 0.995, image covered 0.985); admitted 368450 of 386264 trusted pixels (missed geometry 428, depth edge 2750, image-depth mismatch 14636)
[   92.63] Image depth gate: applied; 401124 paired depth pixels, 23726 mismatch pixels
[   92.65] Observation image depth: excluded 3101 additional voxel samples crossing the mismatch mask; 324124 observed voxels
[   93.68] [ATTENTION] Using backend: sdpa
[   93.68] Observation cache: enabled
[  126.41] 
Extracting GLB:   0%|          | 0/6 [00:00<?, ?it/s]
Building BVH:  17%|█▋        | 1/6 [00:00<00:00, 40.28it/s]
Building BVH:  33%|███▎      | 2/6 [00:06<00:13,  3.40s/it]
Cleaning mesh:  33%|███▎      | 2/6 [00:06<00:13,  3.40s/it]<home>/supratex-0311/venv/lib/python3.12/site-packages/cumesh/remeshing.py:220: UserWarning: Using torch.cross without specifying the dim arg is deprecated.
[  126.41] Please either pass the dim explicitly or simply use torch.linalg.cross.
[  126.41] The default value of dim will change to agree with that of linalg.cross in a future release. (Triggered internally at /__w/pytorch/pytorch/aten/src/ATen/native/Cross.cpp:63.)
[  126.41]   normals0 = torch.cross(mesh_vertices[atempt_triangles_0[:, 1]] - mesh_vertices[atempt_triangles_0[:, 0]], mesh_vertices[atempt_triangles_0[:, 2]] - mesh_vertices[atempt_triangles_0[:, 0]])
[  138.39] 
Cleaning mesh:  50%|█████     | 3/6 [00:07<00:07,  2.41s/it]
Parameterizing new mesh:  50%|█████     | 3/6 [00:07<00:07,  2.41s/it]
Parameterizing new mesh:  67%|██████▋   | 4/6 [00:11<00:05,  2.92s/it]
Sampling attributes:  67%|██████▋   | 4/6 [00:11<00:05,  2.92s/it]    
Sampling attributes:  83%|████████▎ | 5/6 [00:13<00:02,  2.48s/it]
Finalizing mesh:  83%|████████▎ | 5/6 [00:13<00:02,  2.48s/it]    
Finalizing mesh: 100%|██████████| 6/6 [00:18<00:00,  3.55s/it]
Finalizing mesh: 100%|██████████| 6/6 [00:18<00:00,  3.16s/it]
[  144.39] Textured mesh: <home>/supratex-0311/out/r-robot-1.glb
[  144.39] Intermediate files and run receipt: <home>/supratex-0311/out/r-robot-1_work