Faster-GS：分析与改进高斯泼溅优化
Faster-GS: Analyzing and Improving Gaussian Splatting Optimization
对做 3DGS/4D 重建的人，这篇把分散的加速技巧统一成一套可复现的优化配方，并给出数值稳定性、高斯截断与梯度近似的处理细节，可直接作为训练 3DGS 时的低成本基线替换默认实现。
原文依据 Faster-GS achieves up to 5x faster training while maintaining visual quality
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论近期三维高斯泼溅（3DGS）的进展主要聚焦于在保持重建质量的同时加速优化。然而，许多已提出的方法将实现层面的改进与算法层面的根本性修改混为一谈，或以保真度换取性能，导致研究格局碎片化，难以进行公平比较。在本工作中，我们整合并评估了先前 3DGS 研究中最有效且适用面最广的策略，并用若干新颖的优化对其加以增强。我们进一步考察了该框架中尚未被充分探索的方面，包括数值稳定性、高斯截断与梯度近似。由此得到的系统 Faster-GS 提供了一个经过严格优化的算法，我们在一整套全面的基准上对其进行了评估。实验表明，Faster-GS 在保持视觉质量的同时实现了最高 5 倍的训练加速，为 3DGS 优化建立了一个新的高性价比、资源高效的基线。此外，我们证明这些优化可以应用于 4D 高斯重建，从而实现高效的非刚性场景优化。源代码见：https://github.com/nerficg-project/faster-gaussian-splatting
英文原文摘要 1240 字符
Recent advances in 3D Gaussian Splatting (3DGS) have focused on accelerating optimization while preserving reconstruction quality. However, many proposed methods entangle implementation-level improvements with fundamental algorithmic modifications or trade performance for fidelity, leading to a fragmented research landscape that complicates fair comparison. In this work, we consolidate and evaluate the most effective and broadly applicable strategies from prior 3DGS research and augment them with several novel optimizations. We further investigate underexplored aspects of the framework, including numerical stability, Gaussian truncation, and gradient approximation. The resulting system, Faster-GS, provides a rigorously optimized algorithm that we evaluate across a comprehensive suite of benchmarks. Our experiments demonstrate that Faster-GS achieves up to 5x faster training while maintaining visual quality, establishing a new cost-effective and resource efficient baseline for 3DGS optimization. Furthermore, we demonstrate that optimizations can be applied to 4D Gaussian reconstruction, leading to efficient non-rigid scene optimization. Source code available at: https://github.com/nerficg-project/faster-gaussian-splatting