4DGT: Learning a 4D Gaussian Transformer Using Real-World Monocular Videos

TL;DR

4DGT introduces a novel 4D Gaussian Transformer trained on monocular videos, enabling efficient, real-time dynamic scene reconstruction by modeling complex environments with a unified static and dynamic component.

Contribution

The paper presents a new 4D Gaussian Transformer architecture that unifies static and dynamic scene components, with a novel density control for efficient long sequence processing, outperforming prior Gaussian methods.

Findings

Reduces reconstruction time from hours to seconds.

Outperforms prior Gaussian-based networks on real-world videos.

Achieves comparable accuracy to optimization-based methods on cross-domain videos.

Abstract

We propose 4DGT, a 4D Gaussian-based Transformer model for dynamic scene reconstruction, trained entirely on real-world monocular posed videos. Using 4D Gaussian as an inductive bias, 4DGT unifies static and dynamic components, enabling the modeling of complex, time-varying environments with varying object lifespans. We proposed a novel density control strategy in training, which enables our 4DGT to handle longer space-time input and remain efficient rendering at runtime. Our model processes 64 consecutive posed frames in a rolling-window fashion, predicting consistent 4D Gaussians in the scene. Unlike optimization-based methods, 4DGT performs purely feed-forward inference, reducing reconstruction time from hours to seconds and scaling effectively to long video sequences. Trained only on large-scale monocular posed video datasets, 4DGT can outperform prior Gaussian-based networks significantly in real-world videos and achieve on-par accuracy with optimization-based methods on cross-domain videos. Project page: https://4dgt.github.io