Beyond Heuristics: Learnable Density Control for 3D Gaussian Splatting
For 3D Gaussian Splatting practitioners, this work addresses the inflexibility of handcrafted density control rules, offering a learnable alternative that adapts to diverse scenes.
LeGS replaces heuristic density control in 3D Gaussian Splatting with a learnable policy optimized via reinforcement learning, achieving state-of-the-art reconstruction quality and efficiency on Mip-NeRF 360, Tanks & Temples, and Deep Blending datasets.
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