LongSplat：針對隨意拍攝長視頻的穩健無姿態3D高斯潑濺技術

LongSplat: Robust Unposed 3D Gaussian Splatting for Casual Long Videos

August 19, 2025

作者: Chin-Yang Lin, Cheng Sun, Fu-En Yang, Min-Hung Chen, Yen-Yu Lin, Yu-Lun Liu

cs.AI

摘要

LongSplat針對從非專業拍攝的長視頻中進行新視角合成（NVS）所面臨的關鍵挑戰，這些視頻通常具有不規則的相機運動、未知的相機姿態以及廣闊的場景。現有方法常受困於姿態漂移、幾何初始化不準確以及嚴重的記憶體限制。為解決這些問題，我們提出了LongSplat，這是一個強大的無姿態3D高斯潑濺框架，其特點包括：(1) 增量聯合優化，同步優化相機姿態與3D高斯分佈，避免局部最優並確保全域一致性；(2) 基於學習的3D先驗知識的穩健姿態估計模組；以及(3) 高效的八叉樹錨點生成機制，根據空間密度將密集點雲轉化為錨點。在具有挑戰性的基準測試上的廣泛實驗表明，LongSplat達到了業界領先的成果，在渲染質量、姿態準確性和計算效率方面相比先前方法有顯著提升。項目頁面：https://linjohnss.github.io/longsplat/

English

LongSplat addresses critical challenges in novel view synthesis (NVS) from casually captured long videos characterized by irregular camera motion, unknown camera poses, and expansive scenes. Current methods often suffer from pose drift, inaccurate geometry initialization, and severe memory limitations. To address these issues, we introduce LongSplat, a robust unposed 3D Gaussian Splatting framework featuring: (1) Incremental Joint Optimization that concurrently optimizes camera poses and 3D Gaussians to avoid local minima and ensure global consistency; (2) a robust Pose Estimation Module leveraging learned 3D priors; and (3) an efficient Octree Anchor Formation mechanism that converts dense point clouds into anchors based on spatial density. Extensive experiments on challenging benchmarks demonstrate that LongSplat achieves state-of-the-art results, substantially improving rendering quality, pose accuracy, and computational efficiency compared to prior approaches. Project page: https://linjohnss.github.io/longsplat/