LightGaussian: Unbounded 3D Gaussian Compression with 15x Reduction and 200+ FPS

1The University of Texas at Austin 2Xiamen University
*denotes equal contribution

The caption displays both the file size and the Structural Similarity Index (SSIM)


Recent advancements in real-time neural rendering using point-based techniques have paved the way for the widespread adoption of 3D representations. However, foundational approaches like 3D Gaussian Splatting come with a substantial storage overhead caused by growing the SfM points to millions, often demanding gigabyte-level disk space for a single unbounded scene, posing significant scalability challenges and hindering the splatting efficiency.

To address this challenge, we introduce LightGaussian, a novel method designed to transform 3D Gaussians into a more efficient and compact format. Drawing inspiration from the concept of Network Pruning, LightGaussian identifies Gaussians that are insignificant in contributing to the scene reconstruction and adopts a pruning and recovery process, effectively reducing redundancy in Gaussian counts while preserving visual effects. Additionally, LightGaussian employs distillation and pseudo-view augmentation to distill spherical harmonics to a lower degree, allowing knowledge transfer to more compact representations while maintaining scene appearance. Furthermore, we propose a hybrid scheme, VecTree Quantization, to quantize all attributes, resulting in lower bitwidth representations with minimal accuracy losses.

In summary, LightGaussian achieves an averaged compression rate over 15× while boosting the FPS from 139 to 215, enabling an efficient representation of complex scenes on Mip-NeRF 360, Tank & Temple datasets.


Method Overview

The overview of LightGaussian is illustrated above. The 3D-GS model is trained using multi-view images and is initially initialized from SfM point clouds. By expanding the sparse points to millions of Gaussians, the scene is well-represented. Then, the 3D-GS undergoes processing within our pipeline to transform it into a more compact format. This involves utilizing Gaussian Prune and Recovery to reduce the number of Gaussians, SH Distillation to remove redundant SHs while preserving the modeled specular light, and VecTree Quantization to store Gaussians at a lower bit-width.


Result on Mip-NeRF 360 datasets

Visual Comparisons

3d-gs [Kerbl 2023]
VQDVGO [Li 2022]
3d-gs [Kerbl 2023]
VQDVGO [Li 2022]


  author    = {Zhiwen Fan and Kevin Wang and Kairun Wen and Zehao Zhu and Dejia Xu and Zhangyang Wang},
  title     = {LightGaussian: Unbounded 3D Gaussian Compression with 15x Reduction and 200+ FPS},
  year      ={2023}, 
  eprint    ={2311.17245},