NeRF-XL: Scaling NeRFs with Multiple GPUs

TL;DR

NeRF-XL introduces a multi-GPU method for training and rendering large-capacity Neural Radiance Fields, improving reconstruction quality and scalability by minimizing inter-GPU communication and enabling arbitrarily large models.

Contribution

The paper presents a novel distributed training and rendering formulation for NeRFs that scales with multiple GPUs, overcoming previous limitations and revealing new multi-GPU scaling laws.

Findings

Improved reconstruction quality with larger NeRF models.

Speedup in training and rendering with more GPUs.

Successful application on the largest open-source dataset to date.

Abstract

We present NeRF-XL, a principled method for distributing Neural Radiance Fields (NeRFs) across multiple GPUs, thus enabling the training and rendering of NeRFs with an arbitrarily large capacity. We begin by revisiting existing multi-GPU approaches, which decompose large scenes into multiple independently trained NeRFs, and identify several fundamental issues with these methods that hinder improvements in reconstruction quality as additional computational resources (GPUs) are used in training. NeRF-XL remedies these issues and enables the training and rendering of NeRFs with an arbitrary number of parameters by simply using more hardware. At the core of our method lies a novel distributed training and rendering formulation, which is mathematically equivalent to the classic single-GPU case and minimizes communication between GPUs. By unlocking NeRFs with arbitrarily large parameter counts, our approach is the first to reveal multi-GPU scaling laws for NeRFs, showing improvements in reconstruction quality with larger parameter counts and speed improvements with more GPUs. We demonstrate the effectiveness of NeRF-XL on a wide variety of datasets, including the largest open-source dataset to date, MatrixCity, containing 258K images covering a 25km^2 city area.