PyNeRF, a Neural Radiance Field implementation in PyTorch

This repository contains a PyTorch implementation of the NeRF algorithm, as described in the paper "NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis" by Ben Mildenhall, Pratul P. Srinivasan, Matthew Tancik, Jonathan T. Barron, Ravi Ramamoorthi, Ren Ng. The paper can be found here.

The code in this repository is based on the original NeRF implementation by @bmild, which can be found here. The code has been however heavily refactored and modified for better readability and understanding.

Download the dataset

Simply download the dataset on Kaggle. Then create a folder called data/ and unzip the dataset in it.

Run the code

As of now, the code only supports training on the synthetic "Blender" dataset. Simply run the following command:

# Example for training with lego scene:
python3 -m nerf.main --config configs/lego.conf --train

# Example for testing with lego scene:
python3 -m nerf.main --config configs/lego.conf

Architecture of the project