nerf-pytorch

A PyTorch re-implementation

Project | Video | Paper

NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis
Ben Mildenhall*¹, Pratul P. Srinivasan*¹, Matthew Tancik*¹, Jonathan T. Barron², Ravi Ramamoorthi³, Ren Ng^1
1UC Berkeley, ²Google Research, ³UC San Diego
*denotes equal contribution

A PyTorch re-implementation of Neural Radiance Fields.

Speed matters!

The current implementation is blazing fast! (~5-9x faster than the original release, ~2-4x faster than this concurrent pytorch implementation)

What's the secret sauce behind this speedup?

Multiple aspects. Besides obvious enhancements such as data caching, effective memory management, etc. I drilled down through the entire NeRF codebase, and reduced data transfer b/w CPU and GPU, vectorized code where possible, and used efficient variants of pytorch ops (wrote some where unavailable). But for these changes, everything else is a faithful reproduction of the NeRF technique we all admire :)

Sample results from the repo

On synthetic data

On real data

Tiny-NeRF on Google Colab

The NeRF code release has an accompanying Colab notebook, that showcases training a feature-limited version of NeRF on a "tiny" scene. It's equivalent PyTorch notebook can be found at the following URL:

https://colab.research.google.com/drive/1rO8xo0TemN67d4mTpakrKrLp03b9bgCX

What is a NeRF?

A neural radiance field is a simple fully connected network (weights are ~5MB) trained to reproduce input views of a single scene using a rendering loss. The network directly maps from spatial location and viewing direction (5D input) to color and opacity (4D output), acting as the "volume" so we can use volume rendering to differentiably render new views.

Optimizing a NeRF takes between a few hours and a day or two (depending on resolution) and only requires a single GPU. Rendering an image from an optimized NeRF takes somewhere between less than a second and ~30 seconds, again depending on resolution.

How to train your NeRF super-quickly!

To train a "full" NeRF model (i.e., using 3D coordinates as well as ray directions, and the hierarchical sampling procedure), first setup dependencies.

Using pip

In a new conda or virtualenv environment, run using pip

pip install -r requirements.txt

Importantly, install torchsearchsorted by following instructions from their README.

Alternatively, using conda

cd nerf-pytorch 
conda env create
conda activate nerf

Once everything is setup, to run experiments, first edit config/lego.yml to specify your own parameters.

The training script can be invoked by running

python train_nerf.py --config config/lego.yml

Optionally, if resuming training from a previous checkpoint, run

python train_nerf.py --config config/lego.yml --load-checkpoint path/to/checkpoint.ckpt

Cache rays from the dataset (Optional)

An optional, yet simple preprocessing step of caching rays from the dataset results in substantial compute time savings (reduced carbon footprint, yay!), especially when running multiple experiments. It's super-simple: run

python cache_dataset.py --datapath cache/nerf_synthetic/lego/ --halfres False --savedir cache/legocache/legofull --num-random-rays 8192 --num-variations 50

This samples 8192 rays per image from the lego dataset. Each image is 800 x 800 (since halfres is set to False), and 500 such random samples (8192 rays each) are drawn per image. The script takes about 10 minutes to run, but the good thing is, this needs to be run only once per dataset.

NOTE: Do NOT forget to update the cachedir option (under dataset) in your config (.yml) file!

(Full) NeRF on Google Colab

A Colab notebook for the full NeRF model (albeit on low-resolution data) can be accessed here.

Render fun videos (from a pretrained model)

Once you've trained your NeRF, it's time to use that to render the scene. Use the eval_nerf.py script to do that. For the lego-lowres example, this would be

python eval_nerf.py --config pretrained/lego-lowres/config.yml --checkpoint pretrained/lego-lowres/checkpoint199999.ckpt --savedir cache/rendered/lego-lowres

You can create a gif out of the saved images, for instance, by using Imagemagick.

convert cache/rendered/lego-lowres/*.png cache/rendered/lego-lowres.gif

This should give you a gif like this.

A note on reproducibility

All said, this is not an official code release, and is instead a reproduction from the original code (released by the authors here).

The code is thoroughly tested (to the best of my abilities) to match the original implementation (and be much faster)! In particular, I have ensured that

• Every individual module exactly (numerically) matches that of the TensorFlow implementation. This Colab notebook has all the tests, matching op for op (but is very scratchy to look at)!
• Training works as expected (for Lego and LLFF scenes).

The organization of code WILL change around a lot, because I'm actively experimenting with this.

Pretrained models: Pretrained models for the following scenes are available in the pretrained directory (all of them are currently lowres). I will continue adding models herein.

# Synthetic (Blender) scenes
hotdog
lego
ship

# Real (LLFF) scenes
fern

Contributing / Issues?

Feel free to raise GitHub issues if you find anything concerning. Pull requests adding additional features are welcome too.

LICENSE

nerf-pytorch is available under the MIT License. For more details see: LICENSE and ACKNOWLEDGEMENTS.

Misc

Also, a shoutout to yenchenlin for his cool PyTorch implementation, whose volume rendering function replaced mine (my initial impl was inefficient in comparison).