Trimming the Fat: Efficient Compression of 3D Gaussian Splats through Pruning

TL;DR

This paper introduces a gradient-informed pruning method for 3D Gaussian Splatting models, achieving significant compression and speed-up while maintaining or improving performance, thus enhancing scalability of 3D models.

Contribution

It presents a novel post-hoc pruning technique that reduces model size and computation in 3D Gaussian Splatting, addressing scalability issues in current models.

Findings

Up to 75% of Gaussians can be pruned without performance loss.

Achieves approximately 50× compression of the model.

Speeds up rendering to 600 FPS.

Abstract

In recent times, the utilization of 3D models has gained traction, owing to the capacity for end-to-end training initially offered by Neural Radiance Fields and more recently by 3D Gaussian Splatting (3DGS) models. The latter holds a significant advantage by inherently easing rapid convergence during training and offering extensive editability. However, despite rapid advancements, the literature still lives in its infancy regarding the scalability of these models. In this study, we take some initial steps in addressing this gap, showing an approach that enables both the memory and computational scalability of such models. Specifically, we propose "Trimming the fat", a post-hoc gradient-informed iterative pruning technique to eliminate redundant information encoded in the model. Our experimental findings on widely acknowledged benchmarks attest to the effectiveness of our approach, revealing that up to 75% of the Gaussians can be removed while maintaining or even improving upon baseline performance. Our approach achieves around 50$\times$ compression while preserving performance similar to the baseline model, and is able to speed-up computation up to 600 FPS.