Efficient Texture Mapping by Homogeneous Patch Discovery

Texture mapping algorithms use mesh parameterization meth-ods to find an optimal map for the vertices of a 3D model in texture space. These techniques vary in the properties they try to optimize such as stretch and skewness of the texture when mapped onto the surface. While most of them do well in terms of quality, they tend to be computationally intensive for large mesh models, which limits their use in in-teractive applications. We propose a greedy alternative that is significantly faster than current algorithms and achieves comparable quality. We use a priority queue to store poly-gons and use tangential vectors to guide the texture over the surface. Our algorithm is simple to implement and can tex-ture over a million polygons per second on a typical desktop. The algorithm does not impose any constraints on the mesh topology and we do not require the model to be cut into patches before texturing. Stretch and distortion measures are stable across models and are comparable to current algo-rithms. We also propose a method to generate self tile-able textures for use in conjunction with our texture mapping al-gorithm. We present qualitative and quantitative results in comparison with several other texture mapping algorithms. The efficiency and robustness of our algorithm makes it use-ful in interactive modeling applications and texture mapping large mesh models such as heritage monuments