
UL
Ugo Leone Cavalcanti, Fabio Tosi, Matteo Poggi, Andrea Conti, Vladimir Zlokolica, Valerio Cambareri, Stefano Mattoccia
· 1 min read
ResearcharXiv cs.CV
ProDyGS: Dynamic Gaussian Splatting from a Single Static Monocular Camera
arXiv:2609.32711v1 Announce Type: new
Abstract: We present ProDyGS, a novel dynamic 3D Gaussian Splatting framework for high-quality novel view synthesis from videos captured by a single static camera. While existing methods rely on multi-view setups or significant camera motion for geometric constraints, our approach addresses the challenging scenario where multi-view supervision is completely absent. We overcome this limitation by generating synthetic multi-view supervision through depth-guided proxy image synthesis. Specifically, we estimate temporally consistent depth maps using foundational monocular depth networks, then construct 3D Gaussian representations that generate proxy images from arbitrary viewpoints. A deformation network learns temporal dynamics by warping canonical Gaussians using this augmented supervision. Experiments on the DyNeRF dataset demonstrate that our method achieves state-of-the-art performance while requiring only monocular depth estimation as external supervision, outperforming approaches that rely on stronger priors such as scene flow.
Original source
This story was published by arXiv cs.CV and written by Ugo Leone Cavalcanti, Fabio Tosi, Matteo Poggi, Andrea Conti, Vladimir Zlokolica, Valerio Cambareri, Stefano Mattoccia. SyncAI.news shows a preview; the complete article is on the publisher's site.
Read the full story on arxiv.org


