PyNeRF: Pyramidal Neural Radiance Fields

This is the official implementation of PyNeRF.

Installation

1. Install NerfStudio+PyNeRF

Installing PyNeRF should include Nerfstudio as a dependency (tested on v3.4.0)

pip install git+https://github.com/hturki/pynerf

Alternatively, clone then install this repo:

git clone https://github.com/hturki/pynerf.git
cd pynerf
pip install -e .

2. Install additional dependences

pip install git+https://github.com/NVlabs/tiny-cuda-nn/#subdirectory=bindings/torch
pip install torch-scatter -f https://data.pyg.org/whl/torch-2.1.0+${CUDA}.html

The following command should now include pynerf as one of the options:

ns-train -h

Data Preparation

Multicam

Download the NeRF Synthetic dataset
Run the conversion script in the mip-NeRF repository
Use Multicam dataparser

Blender-A

Download the Blender-A scenes
Use Multicam dataparser

Mip-NeRF 360

Download the data from the Mip-NeRF 360 project page
Use MipNeRF 360 dataparser

ADOP

Download the scenes and model checkpoints from the links in the ADOP Github repo
Run scripts/process_adop.py
Use the ADOP dataparser

Citation

@InProceedings{turki2023pynerf,
    title = {PyNeRF: Pyramidal Neural Radiance Fields},
    author = {Haithem Turki and Michael Zollh\"{o}fer and Christian Richardt and Deva Ramanan},
    booktitle={Thirty-Seventh Conference on Neural Information Processing Systems},
    year = {2023}
}

koshian2/pynerf