Monocular Dynamic Gaussian Splatting: Fast, Brittle, and Scene Complexity Rules (original) (raw)

Dynamic Gaussians typically originate from static 3DGS by adding a Dynamics module.

Abstract

Gaussian splatting methods are emerging as a popular approach for converting multi-view image data into scene representations that allow view synthesis. In particular, there is interest in enabling view synthesis for dynamic scenes using only monocular input data---an ill-posed and challenging problem. The fast pace of work in this area has produced multiple simultaneous papers that claim to work best, which cannot all be true. In this work, we organize, benchmark, and analyze many Gaussian-splatting-based methods, providing apples-to-apples comparisons that prior works have lacked. We use multiple existing datasets and a new instructive synthetic dataset designed to isolate factors that affect reconstruction quality.

Bibtex

@misc{liang2024monocular, title={Monocular Dynamic Gaussian Splatting: Fast, Brittle, and Scene Complexity Rules}, author={Yiqing Liang and Mikhail Okunev and Mikaela Angelina Uy and Runfeng Li and Leonidas Guibas and James Tompkin and Adam W. Harley}, year={2024}, eprint={2412.04457}, archivePrefix={arXiv}, primaryClass={cs.CV} }