Editable Free-viewpoint Video Using a Layered Neural Representation

TL;DR

This paper introduces a novel layered neural representation called ST-NeRF for generating editable, photo-realistic free-viewpoint videos of large dynamic scenes using only 16 cameras, enabling scene manipulation and high realism.

Contribution

It presents the first layered neural approach for editable free-viewpoint video of large scenes, incorporating scene parsing, continuous deformation, and object-aware rendering for enhanced editing capabilities.

Findings

Achieves high-quality, photo-realistic free-viewpoint videos.

Supports scene editing such as scaling, relocating, duplicating, and retiming.

Demonstrates effectiveness on large dynamic scenes with multiple performers.

Abstract

Generating free-viewpoint videos is critical for immersive VR/AR experience but recent neural advances still lack the editing ability to manipulate the visual perception for large dynamic scenes. To fill this gap, in this paper we propose the first approach for editable photo-realistic free-viewpoint video generation for large-scale dynamic scenes using only sparse 16 cameras. The core of our approach is a new layered neural representation, where each dynamic entity including the environment itself is formulated into a space-time coherent neural layered radiance representation called ST-NeRF. Such layered representation supports fully perception and realistic manipulation of the dynamic scene whilst still supporting a free viewing experience in a wide range. In our ST-NeRF, the dynamic entity/layer is represented as continuous functions, which achieves the disentanglement of location, deformation as well as the appearance of the dynamic entity in a continuous and self-supervised manner. We propose a scene parsing 4D label map tracking to disentangle the spatial information explicitly, and a continuous deform module to disentangle the temporal motion implicitly. An object-aware volume rendering scheme is further introduced for the re-assembling of all the neural layers. We adopt a novel layered loss and motion-aware ray sampling strategy to enable efficient training for a large dynamic scene with multiple performers, Our framework further enables a variety of editing functions, i.e., manipulating the scale and location, duplicating or retiming individual neural layers to create numerous visual effects while preserving high realism. Extensive experiments demonstrate the effectiveness of our approach to achieve high-quality, photo-realistic, and editable free-viewpoint video generation for dynamic scenes.