GIFStream: 4D Gaussian-based Immersive Video with Feature Stream

TL;DR

GIFStream introduces a 4D Gaussian-based immersive video representation that combines feature streams and advanced compression techniques, achieving high-quality, real-time rendering at low bitrates.

Contribution

It proposes a novel 4D Gaussian model with feature streams and integrated compression networks for efficient, high-quality immersive video.

Findings

High-quality immersive video at 30 Mbps

Real-time rendering on RTX 4090

Effective motion modeling and compression

Abstract

Immersive video offers a 6-Dof-free viewing experience, potentially playing a key role in future video technology. Recently, 4D Gaussian Splatting has gained attention as an effective approach for immersive video due to its high rendering efficiency and quality, though maintaining quality with manageable storage remains challenging. To address this, we introduce GIFStream, a novel 4D Gaussian representation using a canonical space and a deformation field enhanced with time-dependent feature streams. These feature streams enable complex motion modeling and allow efficient compression by leveraging temporal correspondence and motion-aware pruning. Additionally, we incorporate both temporal and spatial compression networks for end-to-end compression. Experimental results show that GIFStream delivers high-quality immersive video at 30 Mbps, with real-time rendering and fast decoding on an RTX 4090. Project page: https://xdimlab.github.io/GIFStream