Beyond Heuristics: Learnable Density Control for 3D Gaussian Splatting

TL;DR

This paper introduces LeGS, a learnable density control framework for 3D Gaussian Splatting that replaces heuristic rules with a reinforcement learning approach, improving flexibility and performance.

Contribution

It presents a novel RL-based density control policy for 3D Gaussian Splatting, with a tailored reward function and efficient computation, outperforming existing methods.

Findings

LeGS outperforms state-of-the-art methods on multiple datasets.

The approach achieves a better balance between reconstruction quality and efficiency.

The reward computation complexity is reduced from O(N^2) to O(N).

Abstract

While 3D Gaussian Splatting (3DGS) has demonstrated impressive real-time rendering performance, its efficacy remains constrained by a reliance on heuristic density control. Despite numerous refinements to these handcrafted rules, such methods inherently lack the flexibility to adapt to diverse scenes with complex geometries. In this paper, we propose a paradigm shift for density control from rigid heuristics to fully learnable policies. Specifically, we introduce \textbf{LeGS}, a framework that reformulates density control as a parameterized policy network optimized via Reinforcement Learning (RL). Central to our approach is the tailored effective reward function grounded in sensitivity analysis, which precisely quantifies the marginal contribution of individual Gaussians to reconstruction quality. To maintain computational tractability, we derive a closed-form solution that reduces the complexity of reward calculation from $O(N^2)$ to $O(N)$. Extensive experiments on the Mip-NeRF 360, Tanks \& Temples, and Deep Blending datasets demonstrate that \textbf{LeGS} significantly outperforms state-of-the-art methods, striking a superior balance between reconstruction quality and efficiency. The code will be released at https://github.com/AaronNZH/LeGS