Gaussian in the Wild: 3D Gaussian Splatting for Unconstrained Image Collections

TL;DR

This paper introduces GS-W, a novel 3D Gaussian-based method for unconstrained image collection view synthesis that models intrinsic and dynamic scene features, improving reconstruction quality and speed.

Contribution

The paper proposes a new 3D Gaussian point-based approach with separated intrinsic and dynamic features, and an adaptive sampling strategy for better scene reconstruction in unconstrained settings.

Findings

Outperforms NeRF-based methods in reconstruction quality

Achieves faster rendering speeds

Effectively handles transient occluders and dynamic variations

Abstract

Novel view synthesis from unconstrained in-the-wild images remains a meaningful but challenging task. The photometric variation and transient occluders in those unconstrained images make it difficult to reconstruct the original scene accurately. Previous approaches tackle the problem by introducing a global appearance feature in Neural Radiance Fields (NeRF). However, in the real world, the unique appearance of each tiny point in a scene is determined by its independent intrinsic material attributes and the varying environmental impacts it receives. Inspired by this fact, we propose Gaussian in the wild (GS-W), a method that uses 3D Gaussian points to reconstruct the scene and introduces separated intrinsic and dynamic appearance feature for each point, capturing the unchanged scene appearance along with dynamic variation like illumination and weather. Additionally, an adaptive sampling strategy is presented to allow each Gaussian point to focus on the local and detailed information more effectively. We also reduce the impact of transient occluders using a 2D visibility map. More experiments have demonstrated better reconstruction quality and details of GS-W compared to NeRF-based methods, with a faster rendering speed. Video results and code are available at https://eastbeanzhang.github.io/GS-W/.