A measure-driven method for normal mapping and normal map design of 3D models

摘要

Normal mapping is one of the most important methods for photorealistic rendering. It preserves geometric attribute values on a simplified mesh. A normal map stores normal vectors for high-quality meshes in a 2D form. A simplified model is then rendered using these normal vectors. To keep a surface's normal property in a map it first of all requires 2D parameterization. The most common approach to this is to divide the surface into several patches, where each patch has its own parameterization. However, this approach has some weakness when it comes to designing global normal maps. This paper presents a measure-driven method that can interactively direct design of normal maps on a 2D plane. This 2D plane has minimal distortion and, more importantly, it is possible to zoom in or shrink the area of interest. The resulting, novel framework serves as a powerful tool for normal mapping and normal map design. We provide a variety of experimental results to demonstrate the efficiency, robustness and efficacy of our approach.