There is a minor bug when non-CPU device is used for torch:
(3/3): Marching cubes
Traceback (most recent call last):
File "/home/gorbuz/content/TripoSR-Bake/01-mesh.py", line 202, in <module>
main()
File "/home/gorbuz/content/TripoSR-Bake/01-mesh.py", line 192, in main
mesh_result = run_mesh(args, tsr_result, grid_result)
File "/home/gorbuz/content/TripoSR-Bake/01-mesh.py", line 91, in run_mesh
.numpy()
TypeError: can't convert cuda:0 device type tensor to numpy. Use Tensor.cpu() to copy the tensor to host memory first.
It's easily fixed by adding .cpu() to line 91 of 01-mesh.py. It's simple enough to fix so I won't submit a PR for it.
There is a minor bug when non-CPU device is used for torch:
It's easily fixed by adding .cpu() to line 91 of 01-mesh.py. It's simple enough to fix so I won't submit a PR for it.