# DeepVoxels: Learning Persistent 3D Feature Embeddings

## TL;DR

DeepVoxels introduces a novel persistent 3D feature embedding that enables view synthesis without explicit geometry modeling, combining 3D scene structure with learned image mappings for high-quality novel view generation.

## Contribution

It proposes DeepVoxels, a learned 3D scene representation that encodes view-dependent appearance without explicit geometry, using a Cartesian grid of embedded features and adversarial training.

## Key findings

- Achieves high-quality novel view synthesis across diverse scenes
- Does not require 3D reconstruction of scenes
- Combines 3D geometric insights with learned image-to-image translation

## Abstract

In this work, we address the lack of 3D understanding of generative neural networks by introducing a persistent 3D feature embedding for view synthesis. To this end, we propose DeepVoxels, a learned representation that encodes the view-dependent appearance of a 3D scene without having to explicitly model its geometry. At its core, our approach is based on a Cartesian 3D grid of persistent embedded features that learn to make use of the underlying 3D scene structure. Our approach combines insights from 3D geometric computer vision with recent advances in learning image-to-image mappings based on adversarial loss functions. DeepVoxels is supervised, without requiring a 3D reconstruction of the scene, using a 2D re-rendering loss and enforces perspective and multi-view geometry in a principled manner. We apply our persistent 3D scene representation to the problem of novel view synthesis demonstrating high-quality results for a variety of challenging scenes.