UDF extract mesh

[zgg311]

Thanks for your excellent research work, now I have a few questions:
What is the method you use to extract mesh when using UDF for reconstruction? The method I use is https://github.com/cvlab-epfl/MeshUDF, but in the code
features_sample = torch.nn.functional.grid_sample(triplane.reshape(-1, triplane_params.resolution, triplane_params.resolution, triplane_params.hidden_dim).permute(0, 3, 1, 2), grid).squeeze(2).squeeze(2).permute(0, 2, 1)
Causes errors:
RuntimeError: derivative for aten::grid_sampler_2d_backward is not implemented
What's your solution, please?

[adricarda]

Hi,Thank you for your interest. It’s really hard to understand what’s going here, but my guess is that there is something wrong with the pytorch version you are not using which is not fully compatible with your gpu.I would double check the PyTorch version maybe.Best,AdrianoOn 3 Aug 2024, at 15:57, 板蓝根 ***@***.***> wrote: Thanks for your excellent research work, now I have a few questions: What is the method you use to extract mesh when using UDF for reconstruction? The method I use is https://github.com/cvlab-epfl/MeshUDF, but in the code features_sample = torch.nn.functional.grid_sample(triplane.reshape(-1, triplane_params.resolution, triplane_params.resolution, triplane_params.hidden_dim).permute(0, 3, 1, 2), grid).squeeze(2).squeeze(2).permute(0, 2, 1) Causes errors: RuntimeError: derivative for aten::grid_sampler_2d_backward is not implemented What's your solution, please? —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[zgg311]

Thank you very much for your reply, which should not be the problem of my pytorch version, so I would like to ask what method is adopted when extracting mesh from UDF, and whether it is possible to provide relevant code reference, thank you very much.