COLMAP-Free 3D Gaussian Splatting

TL;DR

This paper introduces a method for 3D scene reconstruction and novel view synthesis that does not require pre-computed camera poses, leveraging explicit Gaussian representations and sequential processing of input videos.

Contribution

It proposes a novel approach combining 3D Gaussian Splatting with sequential input processing to eliminate the need for SfM preprocessing in view synthesis.

Findings

Improves view synthesis quality over previous methods.

Achieves accurate camera pose estimation under large motion.

Eliminates the need for pre-computed camera poses.

Abstract

While neural rendering has led to impressive advances in scene reconstruction and novel view synthesis, it relies heavily on accurately pre-computed camera poses. To relax this constraint, multiple efforts have been made to train Neural Radiance Fields (NeRFs) without pre-processed camera poses. However, the implicit representations of NeRFs provide extra challenges to optimize the 3D structure and camera poses at the same time. On the other hand, the recently proposed 3D Gaussian Splatting provides new opportunities given its explicit point cloud representations. This paper leverages both the explicit geometric representation and the continuity of the input video stream to perform novel view synthesis without any SfM preprocessing. We process the input frames in a sequential manner and progressively grow the 3D Gaussians set by taking one input frame at a time, without the need to pre-compute the camera poses. Our method significantly improves over previous approaches in view synthesis and camera pose estimation under large motion changes. Our project page is https://oasisyang.github.io/colmap-free-3dgs