CaT-GS: Efficient 3DGS Rendering for Large Scale Scenes via Inter-frame Caching and Tile Scheduling
2026-07-20 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionArtificial Intelligence
AI summaryⓘ
The authors found that current 3D Gaussian Splatting methods slow down a lot when rendering large scenes because of inefficient processing and uneven workload on the GPU. They identified three main problems: repeating work between frames, re-rendering parts of the scene that are hidden, and unbalanced tasks for tiles in the image. To fix this, they created CaT-GS, which reuses work across frames, avoids rendering hidden areas multiple times, and better balances the workload across the GPU. Their approach speeds up rendering by up to 10 times compared to the original method and improves performance compared to other recent approaches.
3D Gaussian SplattingNeural RenderingRasterizationGPU UtilizationTile-based RenderingOcclusion CullingLoad BalancingReal-time RenderingMulti-frame PreprocessingCaching
Authors
Tingjia Zhang, Bo Chen, Shengzhong Liu, Fan Wu, Guihai Chen
Abstract
Recent breakthroughs in 3D Gaussian Splatting (3DGS) have advanced neural rendering with high fidelity and speed. However, its performance degrades significantly in large-scale scenes due to the computational burden of tile-based rasterization. Existing optimization efforts either require costly scene re-training or focus on narrow aspects of the pipeline, overlooking critical inefficiencies in real-world deployments. Through a comprehensive analysis, we identify three primary sources of redundancy and low GPU utilization: redundant inter-frame pre-processing, viewpoint-based occlusion redundancy, and severe tile-level load imbalance. To address these issues, we propose CaT-GS, a novel and efficient 3DGS rendering pipeline. CaT-GS introduces a speculative multi-frame preprocessing method to eliminate redundant computations across consecutive frames, and an inter-frame caching mechanism to eliminate viewpoint redundant rendering stages. Furthermore, it refactors rasterization tasks with a dedicated kernel to mitigate tile load imbalance, significantly boosting GPU utilization. Extensive experiments demonstrate that CaT-GS achieves a speedup of up to 10 times over the original 3DGS and up to 70% over previous state-of-the-art methods, establishing a new benchmark for high-fidelity, real-time rendering of large-scale scenes.