jorgemf/nerf

Neural Radiance Fields implementation

Hash NeRF

This is an implementation of HasHNerf in Pytorch. The fundational paper is NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis and this implementation is based on Instant Neural Graphics Primitives with a Multiresolution Hash Encoding.

The code is based on the code of HashNeRF-pytorch and Instant ngp.

Configuration

Dependencies

The dependencies for this project are managed by Poetry. To install them, run

poetry install

Some of the dependencies are:

• Pytorch 2.0
• Python 3.10

Docker

A Dockerfile is provided to run the code in a container. To build the image, run

./build_docker_image.sh

The image name is $HOSTNAME/nerf. To run the container, run

./docker.sh python -m nerf.main --help

Hardware

This code was developed and tested on the Nvidia 4090 GPU with 24GB of memory.

Usage

Dataset

In order to download the dataset, run

./docker.sh python -m nerf.main dataset download

and it will dowloaded under -./data.

Training

In order to train a model, run

./docker.sh python -m nerf.main train data/nerf_synthetic/lego blender 

Validation and metrics

TODO

Inference

TODO