Tile-wise vs. Image-wise: Random-Tile Loss and Training Paradigm for Gaussian Splatting

Xiaoyu Zhang, Weihong Pan, Xiaojun Xiang, Hongjia Zhai, Liyang Zhou, Hanqing Jiang, Guofeng Zhang; Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2025, pp. 26923-26932

3D Gaussian Splatting (3DGS) has drawn significant attention for its advantages in rendering speed and quality. Most existing methods still rely on the image-wise loss and training paradigm because of its intuitive nature in the Splatting algorithm. However, image-wise loss lacks multi-view constraints, which are generally essential for optimizing 3D appearance and geometry. To address this, we propose RT-Loss along with a tile-based training paradigm, which uses randomly sampled tiles to integrate multi-view appearance and structural constraints in 3DGS. Additionally, we introduce an tile-based adaptive densification control strategy tailored for our training paradigm. Extensive experiments show that our approach consistently improves performance metrics while maintaining efficiency across various benchmark datasets.