Faster and Better 3D Splatting via Group Training

TL;DR

This paper introduces Group Training for 3D Gaussian Splatting, organizing primitives into groups to significantly accelerate training and enhance rendering quality without altering existing frameworks.

Contribution

It presents a simple, universal grouping strategy that improves training efficiency and synthesis quality in 3D Gaussian Splatting methods.

Findings

Up to 30% faster convergence in training.

Consistent improvement in rendering quality.

Universal compatibility with existing 3DGS frameworks.

Abstract

3D Gaussian Splatting (3DGS) has emerged as a powerful technique for novel view synthesis, demonstrating remarkable capability in high-fidelity scene reconstruction through its Gaussian primitive representations. However, the computational overhead induced by the massive number of primitives poses a significant bottleneck to training efficiency. To overcome this challenge, we propose Group Training, a simple yet effective strategy that organizes Gaussian primitives into manageable groups, optimizing training efficiency and improving rendering quality. This approach shows universal compatibility with existing 3DGS frameworks, including vanilla 3DGS and Mip-Splatting, consistently achieving accelerated training while maintaining superior synthesis quality. Extensive experiments reveal that our straightforward Group Training strategy achieves up to 30\% faster convergence and improved rendering quality across diverse scenarios. Project Website: https://chengbo-wang.github.io/3DGS-with-Group-Training/