pixelNeRF: Neural Radiance Fields from One or Few Images

TL;DR

pixelNeRF introduces a neural radiance field model that can generate novel views from just one or a few images by learning a scene prior, enabling fast, feed-forward view synthesis without explicit 3D supervision.

Contribution

It presents a convolutional architecture conditioned on input images, allowing training across multiple scenes and enabling single-image view synthesis with state-of-the-art results.

Findings

Outperforms existing methods on ShapeNet benchmarks

Effective with as few as one input image

Works on multi-object and real-world scenes

Abstract

We propose pixelNeRF, a learning framework that predicts a continuous neural scene representation conditioned on one or few input images. The existing approach for constructing neural radiance fields involves optimizing the representation to every scene independently, requiring many calibrated views and significant compute time. We take a step towards resolving these shortcomings by introducing an architecture that conditions a NeRF on image inputs in a fully convolutional manner. This allows the network to be trained across multiple scenes to learn a scene prior, enabling it to perform novel view synthesis in a feed-forward manner from a sparse set of views (as few as one). Leveraging the volume rendering approach of NeRF, our model can be trained directly from images with no explicit 3D supervision. We conduct extensive experiments on ShapeNet benchmarks for single image novel view synthesis tasks with held-out objects as well as entire unseen categories. We further demonstrate the flexibility of pixelNeRF by demonstrating it on multi-object ShapeNet scenes and real scenes from the DTU dataset. In all cases, pixelNeRF outperforms current state-of-the-art baselines for novel view synthesis and single image 3D reconstruction. For the video and code, please visit the project website: https://alexyu.net/pixelnerf