Efficient 4D Gaussian Stream with Low Rank Adaptation

Recent methods have made significant progress in synthesizing novel views with long video sequences. This paper proposes a highly scalable method for dynamic novel view synthesis with continual learning. We leverage the 3D Gaussians to represent the scene and a low-rank adaptation-based deformation model to capture the dynamic scene changes. Our method continuously reconstructs the dynamics with chunks of video frames, reduces the streaming bandwidth by $90\%$ while maintaining high rendering quality comparable to the off-line SOTA methods.

@article{liu2025_2502.16575,
  title={ Efficient 4D Gaussian Stream with Low Rank Adaptation },
  author={ Zhenhuan Liu and Shuai Liu and Yidong Lu and Yirui Chen and Jie Yang and Wei Liu },
  journal={arXiv preprint arXiv:2502.16575},
  year={ 2025 }
}