Jacob Munkberg

Thanks! To increase quality, the first thing I would recommend is to increase batch size and/or rendering resolution, but that is hard on a RTX3060. You can also increase the...

Yes, the .obj file is automatically generated, see https://github.com/NVlabs/nvdiffrec/blob/main/train.py#L646 You can skip validation by specifying `"validate" : false,` in the config

This seems to be a problem in the tiny-cuda-nn setup. I would suggest reporting that issue there https://github.com/NVlabs/tiny-cuda-nn, or testing their bindings in isolation to track down the issue: https://github.com/NVlabs/tiny-cuda-nn#pytorch-extension

Hello, Unfortunately, we did not release the DTU dataset reader and the MLP parameterization of geometry as part of the public release. For the dataset reader for DTU, we adapted...

Thanks for setting up the Colab example, @neurall ! 10 images are on the lower side I think. We typically use 50+views in our tests. Correct, we need foreground segmentation...

In order for me to answer this question, please provide more details.

Hello @jpbalarini Yes, we do require alpha. The foreground segmentation mask is essential for our shape optimization to work. See the paper for details. We should be compatible with the...

Thanks for the bug report. This is very strange. We explicitly added `indexing='ij'` to the `torch.meshgrid()` calls to remove the warnings (I assume the meshgrid warning reappeared for you when...

Actually, I would change the pytorch commandline to `conda install pytorch torchvision torchaudio cudatoolkit=11.3 -c pytorch` pytorch only supports cuda 11.3 . It is fine to have a more modern...

Ok, this is strange. I'll try in a fresh conda env later to see if I can reproduce. I think I have the latest PyTorch 1.11, but they may have...