NeRF Tutorial

This example code is largely inspried from the Jupyter Notebook given by the pytorch repository.

It has been adapted to be ran inside a python sheel and not a jupyer notebook. It is GPU only, there is support for CPU.

It cannot run my lab computer: NVIDIA GeForce RTX 2060 (6GiB of VRAM) and thus need to be run on the cluster which has more VRAM.

Installation (using conda)

Please refer to the original Pytorch3D project, here is a quick summary:

conda create -n pytorhc3d python=3.8
conda activate pytorch3d
conda install -c pytorch pytorch=1.7.1 torchvision cudatoolkit=10.2
conda install -c conda-forge -c fvcore -c iopath fvcore iopath 
conda install -c bottler nvidiacub
conda intall numpy matplotlib plotly

You can then clone this repo and run the code

git clone https://gitea.auro.re/otthorn/nerf_tutorial.git
cd nerf_tutorial
python3 main.py

To deactivate the conda env

conda deactivate

Installation (using pip and virtualenv)

This requires Python>=3.8

First, clone the repo

git clone https://gitea.auro.re/otthorn/nerf_tutorial.git
cd nerf_tutorial

Then install pytorch3d inside a virtualenv

virtualenv -p /usr/bin/python3.8 pytorch3d
source pytorch3d/bin/activate
pip3 install pytorch3d -f https://dl.fbaipublicfiles.com/pytorch3d/packaging/wheels/py38_cu102_pyt171/download.html
pip3 install torch torchvision

You can then run the code

python3 main.py

To deactivate the virtualenv

deactivate