English

Locality-aware Gaussian Compression for Fast and High-quality Rendering

Computer Vision and Pattern Recognition 2025-03-13 v3

Abstract

We present LocoGS, a locality-aware 3D Gaussian Splatting (3DGS) framework that exploits the spatial coherence of 3D Gaussians for compact modeling of volumetric scenes. To this end, we first analyze the local coherence of 3D Gaussian attributes, and propose a novel locality-aware 3D Gaussian representation that effectively encodes locally-coherent Gaussian attributes using a neural field representation with a minimal storage requirement. On top of the novel representation, LocoGS is carefully designed with additional components such as dense initialization, an adaptive spherical harmonics bandwidth scheme and different encoding schemes for different Gaussian attributes to maximize compression performance. Experimental results demonstrate that our approach outperforms the rendering quality of existing compact Gaussian representations for representative real-world 3D datasets while achieving from 54.6×\times to 96.6×\times compressed storage size and from 2.1×\times to 2.4×\times rendering speed than 3DGS. Even our approach also demonstrates an averaged 2.4×\times higher rendering speed than the state-of-the-art compression method with comparable compression performance.

Keywords

Cite

@article{arxiv.2501.05757,
  title  = {Locality-aware Gaussian Compression for Fast and High-quality Rendering},
  author = {Seungjoo Shin and Jaesik Park and Sunghyun Cho},
  journal= {arXiv preprint arXiv:2501.05757},
  year   = {2025}
}

Comments

Accepted to ICLR 2025. Project page: https://seungjooshin.github.io/LocoGS

R2 v1 2026-06-28T21:02:18.238Z