Reconstruct What You Can: Distractor-free 4D Gaussian Splatting
Dynamic reconstruction from casual video must preserve valid subject motion while suppressing transient occluders, shadows, and volumetric effects. Existing methods either suppress valid motion through static-scene assumptions or model all observations jointly, producing ghosting and unstable geometry. We introduce ReCan-GS, a 4D Gaussian Splatting framework that separates coherent scene content from inconsistent observations. It derives a Jacobian-based deformation-sensitivity score from the learned deformation field, using it as an empirical proxy for reconstruction ambiguity. The score guides separation into dynamic Core Gaussians and per-view Distractor Gaussians, or post-hoc pruning of sensitive Core Gaussians. The same criterion supports removal of spatial occluders, photometric artifacts, and evolving smoke or fog without masks, pretrained segmenters, or medium-specific supervision. Experiments on synthetic and real dynamic scenes and synthetic smoke/fog scenes demonstrate its usefulness across the evaluated distractor regimes.
Authors
- Koichiro Niinuma (ORCID: https://orcid.org/0000-0001-8367-3988)
- Ananya Bal (ORCID: https://orcid.org/0000-0002-9592-1085)
- Liuyue Xie (ORCID: https://orcid.org/0000-0001-7125-5084)
- László A. Jeni (ORCID: https://orcid.org/0000-0002-2830-700X)
- Joel Julin
- Mose Sakashita (ORCID: https://orcid.org/0000-0003-4953-2027)
Publication Details
- Journal
- arXiv (Cornell University)
- Published
- 2026-11-12
- DOI
- https://doi.org/10.48550/arxiv.2611.16259
- Primary Topic
- 3D Shape Modeling and Analysis
- Type
- preprint