Robust Dynamic Radiance Fields

TL;DR

This paper introduces a robust method for dynamic radiance field reconstruction that jointly estimates scene structure, appearance, and camera parameters, improving reliability on challenging videos with dynamic objects and camera motion.

Contribution

It proposes a novel joint estimation approach for static and dynamic radiance fields along with camera parameters, enhancing robustness over existing methods.

Findings

Outperforms state-of-the-art dynamic view synthesis methods.

Demonstrates robustness on challenging videos with dynamic objects.

Provides extensive quantitative and qualitative evaluations.

Abstract

Dynamic radiance field reconstruction methods aim to model the time-varying structure and appearance of a dynamic scene. Existing methods, however, assume that accurate camera poses can be reliably estimated by Structure from Motion (SfM) algorithms. These methods, thus, are unreliable as SfM algorithms often fail or produce erroneous poses on challenging videos with highly dynamic objects, poorly textured surfaces, and rotating camera motion. We address this robustness issue by jointly estimating the static and dynamic radiance fields along with the camera parameters (poses and focal length). We demonstrate the robustness of our approach via extensive quantitative and qualitative experiments. Our results show favorable performance over the state-of-the-art dynamic view synthesis methods.