Adaptive Seamless Texture Synthesis for UV Fragmentation in Large-Scale Scene Rendering

The problem of UV fragmentation leading to texture seams and compression issues is well-documented in Jihoon Do et al. (2023), who propose a seam-aware CNN post-processor. But their approach is post-hoc and focused on compressed textures. Imagine instead a system that, during rendering, detects areas prone to seam artifacts (using error tomography ideas from Franke et al. 2023) and synthesizes seamless textures on-the-fly, blending across UV boundaries in a spatially-aware and temporally consistent manner. This could leverage diffusion models (inspired by Liang et al. 2025’s DiffusionRenderer) for plausible texture inpainting, guided by geometric and appearance cues. This dynamic approach could enable artifact-free rendering even in highly fragmented, real-time massive scenes (see Kerbl et al. 2024), and may save on both storage and computation by repairing only perceptually problematic regions.

References:

1. Seam-Aware Rendering Quality Enhancement Network For Compressed 3D Scene. Jihoon Do, Soowoong Kim, Jung Won Kang (2023). Visual Communications and Image Processing.
2. A Hierarchical 3D Gaussian Representation for Real-Time Rendering of Very Large Datasets. Bernhard Kerbl, Andreas Meuleman, Georgios Kopanas, Michael Wimmer, Alexandre Lanvin, G. Drettakis (2024). ACM Transactions on Graphics.
3. VET: Visual Error Tomography for Point Cloud Completion and High-Quality Neural Rendering. Linus Franke, Darius Rückert, Laura Fink, Matthias Innmann, Marc Stamminger (2023). ACM SIGGRAPH Conference and Exhibition on Computer Graphics and Interactive Techniques in Asia.
4. DiffusionRenderer: Neural Inverse and Forward Rendering with Video Diffusion Models. Ruofan Liang, Zan Gojcic, Huan Ling, Jacob Munkberg, J. Hasselgren, Zhi-Hao Lin, Jun Gao, Alexander Keller, Nandita Vijaykumar, Sanja Fidler, Zian Wang (2025). Computer Vision and Pattern Recognition.