3D Gaussian splatting (3DGS) has recently demonstrated impressive capabilities in real-time novel view synthesis and 3D reconstruction. However, 3DGS heavily depends on the accurate initialization derived from Structure-from-Motion (SfM) methods. When trained with randomly initialized point clouds, 3DGS fails to maintain its ability to produce high-quality images, undergoing large performance drops of 4-5 dB in PSNR. Through extensive analysis of SfM initialization in the frequency domain and analysis of a 1D regression task with multiple 1D Gaussians, we propose a novel optimization strategy dubbed RAIN-GS (Relaxing Accurate Initialization Constraint for 3D Gaussian Splatting), that successfully trains 3D Gaussians from random point clouds. We show the effectiveness of our strategy through quantitative and qualitative comparisons on multiple datasets, largely improving the performance in all settings. Our project page and code can be found at https://ku-cvlab.github.io/RAIN-GS.

翻译：三维高斯泼溅（3DGS）近期在实时新视角合成与三维重建方面展现出令人瞩目的能力。然而，3DGS高度依赖于从运动恢复结构（SfM）方法获得的精确初始化。当使用随机初始化的点云进行训练时，3DGS难以维持生成高质量图像的能力，峰值信噪比（PSNR）会大幅下降4-5 dB。通过对频域SfM初始化的深入分析以及对一维回归任务中多个一维高斯的分析，我们提出了一种名为RAIN-GS（放松三维高斯泼溅精确初始化约束）的新型优化策略，该策略成功实现了从随机点云训练三维高斯模型。通过多个数据集上的定量与定性对比，我们展示了该策略的有效性，在所有设置下均显著提升了性能。项目页面与代码可见于 https://ku-cvlab.github.io/RAIN-GS。