Learning Unified Decompositional and Compositional NeRF for Editable Novel View Synthesis

TL;DR

This paper introduces a unified NeRF framework that jointly models scene decomposition and composition, enabling high-quality novel view synthesis and editing by disentangling objects and background.

Contribution

It proposes a two-stage NeRF approach with a novel regularization and pseudo supervision for scene decomposition, improving joint scene editing and view synthesis capabilities.

Findings

Outperforms state-of-the-art methods in view synthesis accuracy

Effectively disentangles objects and background for editing

Demonstrates superior qualitative results in scene editing

Abstract

Implicit neural representations have shown powerful capacity in modeling real-world 3D scenes, offering superior performance in novel view synthesis. In this paper, we target a more challenging scenario, i.e., joint scene novel view synthesis and editing based on implicit neural scene representations. State-of-the-art methods in this direction typically consider building separate networks for these two tasks (i.e., view synthesis and editing). Thus, the modeling of interactions and correlations between these two tasks is very limited, which, however, is critical for learning high-quality scene representations. To tackle this problem, in this paper, we propose a unified Neural Radiance Field (NeRF) framework to effectively perform joint scene decomposition and composition for modeling real-world scenes. The decomposition aims at learning disentangled 3D representations of different objects and the background, allowing for scene editing, while scene composition models an entire scene representation for novel view synthesis. Specifically, with a two-stage NeRF framework, we learn a coarse stage for predicting a global radiance field as guidance for point sampling, and in the second fine-grained stage, we perform scene decomposition by a novel one-hot object radiance field regularization module and a pseudo supervision via inpainting to handle ambiguous background regions occluded by objects. The decomposed object-level radiance fields are further composed by using activations from the decomposition module. Extensive quantitative and qualitative results show the effectiveness of our method for scene decomposition and composition, outperforming state-of-the-art methods for both novel-view synthesis and editing tasks.