PackUV: Packed Gaussian UV Maps for 4D Volumetric Video

TL;DR

PackUV introduces a novel 4D Gaussian representation mapped into multi-scale UV atlases, enabling compact, video-codec-compatible storage and scalable, high-quality volumetric video reconstruction for long sequences.

Contribution

The paper presents PackUV, a new 4D Gaussian representation with a UV atlas format and a fitting method that ensures temporal consistency and compatibility with standard video codecs.

Findings

Outperforms existing methods in rendering fidelity.

Supports sequences up to 30 minutes with consistent quality.

Introduces the largest multi-view volumetric video dataset to date.

Abstract

Volumetric videos offer immersive 4D experiences, but remain difficult to reconstruct, store, and stream at scale. Existing Gaussian Splatting based methods achieve high-quality reconstruction but break down on long sequences, temporal inconsistency, and fail under large motions and disocclusions. Moreover, their outputs are typically incompatible with conventional video coding pipelines, preventing practical applications. We introduce PackUV, a novel 4D Gaussian representation that maps all Gaussian attributes into a sequence of structured, multi-scale UV atlas, enabling compact, image-native storage. To fit this representation from multi-view videos, we propose PackUV-GS, a temporally consistent fitting method that directly optimizes Gaussian parameters in the UV domain. A flow-guided Gaussian labeling and video keyframing module identifies dynamic Gaussians, stabilizes static regions, and preserves temporal coherence even under large motions and disocclusions. The resulting UV atlas format is the first unified volumetric video representation compatible with standard video codecs (e.g., FFV1) without losing quality, enabling efficient streaming within existing multimedia infrastructure. To evaluate long-duration volumetric capture, we present PackUV-2B, the largest multi-view video dataset to date, featuring more than 50 synchronized cameras, substantial motion, and frequent disocclusions across 100 sequences and 2B (billion) frames. Extensive experiments demonstrate that our method surpasses existing baselines in rendering fidelity while scaling to sequences up to 30 minutes with consistent quality.