search

NVlabs / nvdiffrec

Official code for the CVPR 2022 (oral) paper "Extracting Triangular 3D Models, Materials, and Lighting From Images".
Other
2.09k stars 222 forks source link

Getting stuck during installation #156

Open eliyoom opened 2 months ago

eliyoom commented 2 months ago

Hello The following command runs forever :

conda install pytorch torchvision torchaudio cudatoolkit=11.6 -c pytorch -c conda-forge

Breaking it into several parts works, but then the following line fails:

pip install --global-option="--no-networks" git+https://github.com/NVlabs/tiny-cuda-nn#subdirectory=bindings/torch

As a result, I get the following error when executing python train.py --config configs/bob.json (tinycudann missing)

Traceback (most recent call last): File "C:\Users\eli\Algo_studio\algo_studio\TV_Python\nvdiffrec-main\train.py", line 36, in from render import mlptexture File "C:\Users\eli\Algo_studio\algo_studio\TV_Python\nvdiffrec-main\render\mlptexture.py", line 11, in import tinycudann as tcnn ModuleNotFoundError: No module named 'tinycudann'

jmunkberg commented 2 months ago

Hello @eliyoom ,

For the first remark, this is a standard PyTorch install. https://pytorch.org/get-started/locally/ . You can install through conda or through pip (often slightly faster in my experience) inside the conda env. Note per our instructions that you need to match the CUDA toolkit versions in torch and the one you install to build the extension.

For the second remark, AFAIU, this is an issue with installing tcnn, so please check there for help: https://github.com/NVlabs/tiny-cuda-nn

eliyoom commented 2 months ago

Thank you so much! It is working now.