Bringing NeRFs to the Latent Space: Inverse Graphics Autoencoder

TL;DR

This paper introduces an Inverse Graphics Autoencoder (IG-AE) that aligns 2D image autoencoders with 3D geometry, enabling efficient latent space manipulation of NeRFs with improved quality and faster training.

Contribution

The paper proposes a novel IG-AE that regularizes image autoencoders with 3D geometry, facilitating latent space NeRF training and enhancing performance.

Findings

Latent NeRFs with IG-AE outperform standard autoencoders in quality.

Training and rendering are faster with IG-AE-based NeRFs.

Open-source implementation extends Nerfstudio for latent scene learning.

Abstract

While pre-trained image autoencoders are increasingly utilized in computer vision, the application of inverse graphics in 2D latent spaces has been under-explored. Yet, besides reducing the training and rendering complexity, applying inverse graphics in the latent space enables a valuable interoperability with other latent-based 2D methods. The major challenge is that inverse graphics cannot be directly applied to such image latent spaces because they lack an underlying 3D geometry. In this paper, we propose an Inverse Graphics Autoencoder (IG-AE) that specifically addresses this issue. To this end, we regularize an image autoencoder with 3D-geometry by aligning its latent space with jointly trained latent 3D scenes. We utilize the trained IG-AE to bring NeRFs to the latent space with a latent NeRF training pipeline, which we implement in an open-source extension of the Nerfstudio framework, thereby unlocking latent scene learning for its supported methods. We experimentally confirm that Latent NeRFs trained with IG-AE present an improved quality compared to a standard autoencoder, all while exhibiting training and rendering accelerations with respect to NeRFs trained in the image space. Our project page can be found at https://ig-ae.github.io .