Texture Synthesis
Overview
Texture synthesis studies how to generate, transfer, complete, or interpolate surface appearance, supporting uses from image editing and restoration to games, garments, and 3D reconstruction. Methods range from statistical and exemplar-based matching to GANs, diffusion models, and neural cellular automata; recent 3D approaches generate in UV or texture space and use geometry, multiview images, or surface-aware attention to align details across views. Evaluation considers visual fidelity and detail alongside distributional quality, seam handling, unseen-region completion, and consistency across viewpoints or time. A central challenge is balancing fine detail and controllability with coherent, physically plausible materials, especially when training data or observations are sparse and generation must remain efficient at high resolution.
Latest papers
34 papers7 papers in the last four weeks, up 133% on the four weeks before. 0.0% of all new papers.