Bake It Till You Make It: Ultrafast Spatial Texture-Atlas Splatting

2026-07-15

AI Summary

This work addresses the high computational overhead in 3D Gaussian Splatting (3DGS) caused by hash-grid-based appearance parameterization. To mitigate this, the authors propose a decoupled radiance representation that models low-frequency geometry and view-dependent appearance using 2D surfels, while baking high-frequency textures into a compact, view-independent texture atlas via a spatial hash grid. Sparse optimization is further introduced to eliminate redundancy. By leveraging GPU-efficient texture mapping mechanisms, the method achieves state-of-the-art visual fidelity while accelerating rendering by up to 5× over the original 3DGS. Notably, it enables real-time 4K rendering at 60 FPS on consumer-grade hardware for the first time.

Abstract

Recent extensions of 3D Gaussian Splatting (3DGS) capture fine color details using hash-grid-based appearance parameterization but incur high computational cost during fragment rendering. We introduce a decoupled radiance representation that models low-frequency geometry and view dependent appearance features with 2D surfels while representing high-frequency textures via a view-independent spatial hash grid that is baked into a compact texture atlas. By including sparsity-enhancing optimizations that penalize semi-transparency and per-primitive falloff, our method aggressively prunes insignificant surfels and achieves significantly faster and sparser reconstructions than prior work. Exploiting geometric sparsity and efficient GPU texture mapping, our approach achieves up to a fivefold speedup over 3DGS while preserving state-of-the-art visual fidelity, enabling real-time 4K rendering at 60 FPS on consumer hardware.

Problem

• 3D Gaussian Splatting
• fragment rendering
• computational cost
• real-time rendering
• high-frequency textures

Innovation

• 3D Gaussian Splatting
• texture atlas
• spatial hash grid
• sparse reconstruction
• real-time rendering