Can't extract mesh

[WXuanyang]

I tried to extract the mesh with the ipynb code, but have the following bug in the third block (search for tight bound of the object)

---

RuntimeError Traceback (most recent call last)

in 24 dir_embedded = embedding_dir(dir_[i:i+chunk]) # (N, embed_dir_channels) 25 xyzdir_embedded = torch.cat([xyz_embedded, dir_embedded], 1) ---> 26 out_chunks += [nerf_fine(xyzdir_embedded)] 27 rgbsigma = torch.cat(out_chunks, 0) 28 ~/anaconda3/envs/nerf_pl/lib/python3.6/site-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs) 1100 if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks 1101 or _global_forward_hooks or _global_forward_pre_hooks): -> 1102 return forward_call(*input, **kwargs) 1103 # Do not call functions when jit is used 1104 full_backward_hooks, non_full_backward_hooks = [], [] ~/mesh/nerf_pl/models/nerf.py in forward(self, x, sigma_only) 98 if not sigma_only: 99 input_xyz, input_dir = \ --> 100 torch.split(x, [self.in_channels_xyz, self.in_channels_dir], dim=-1) 101 else: 102 input_xyz = x ~/anaconda3/envs/nerf_pl/lib/python3.6/site-packages/torch/functional.py in split(tensor, split_size_or_sections, dim) 154 # split_size_or_sections. The branching code is in _tensor.py, which we 155 # call here. --> 156 return tensor.split(split_size_or_sections, dim) 157 158 ~/anaconda3/envs/nerf_pl/lib/python3.6/site-packages/torch/_tensor.py in split(self, split_size, dim) 516 return super(Tensor, self).split_with_sizes(split_size, dim) 517 else: --> 518 return super(Tensor, self).split_with_sizes(split_size, dim) 519 520 def unique(self, sorted=True, return_inverse=False, return_counts=False, dim=None): RuntimeError: start (0) + length (63) exceeds dimension size (42). I'm using the dev branch and trained with 4 1080ti gpu, is that a multi-gpu related problem?

[WXuanyang]

Oh I checked the file and I found the problem:

embedding_xyz = Embedding(10) embedding_dir = Embedding(4)

I changed those 2 lines in the second block and its now stop giving errors

[kwea123]

Yes, I made an update recently, but forgot to update the .ipynb. For those who encounter the same error, please made the modification like above. I will correct this soon.