Decomposing Densification in Gaussian Splatting for Faster 3D Scene Reconstruction

TL;DR

This paper introduces a global-to-local densification strategy and an energy-guided multi-resolution training framework for Gaussian Splatting, significantly accelerating 3D scene reconstruction while maintaining high quality.

Contribution

It presents a novel densification method and training framework that improve efficiency and speed in Gaussian Splatting for 3D scene reconstruction.

Findings

Achieves over 2x training speedup

Uses fewer Gaussian primitives

Maintains superior reconstruction quality

Abstract

3D Gaussian Splatting (GS) has emerged as a powerful representation for high-quality scene reconstruction, offering compelling rendering quality. However, the training process of GS often suffers from slow convergence due to inefficient densification and suboptimal spatial distribution of Gaussian primitives. In this work, we present a comprehensive analysis of the split and clone operations during the densification phase, revealing their distinct roles in balancing detail preservation and computational efficiency. Building upon this analysis, we propose a global-to-local densification strategy, which facilitates more efficient growth of Gaussians across the scene space, promoting both global coverage and local refinement. To cooperate with the proposed densification strategy and promote sufficient diffusion of Gaussian primitives in space, we introduce an energy-guided coarse-to-fine multi-resolution training framework, which gradually increases resolution based on energy density in 2D images. Additionally, we dynamically prune unnecessary Gaussian primitives to speed up the training. Extensive experiments on MipNeRF-360, Deep Blending, and Tanks & Temples datasets demonstrate that our approach significantly accelerates training,achieving over 2x speedup with fewer Gaussian primitives and superior reconstruction performance.