Skinning Based 3D Character Model Auto Segmentation

3D character model is the carrier of human in the virtual world. Games, film, industry, and other application areas need a large number of practical and creative character models, but it needs a long time and excellent technicians to create an animated model from the design to the full. To simplify this process, oriented to instance-based modeling, this paper proposes an automatic 3D character model segmentation algorithm. The algorithm is based on existing character models with binding skeleton. Based on the skinning information, the mesh and skeleton of models are segmented into five preconfigured human segmentation parts from a semantic library, and the boundaries of each part are optimized. Finally, an average plane is created for each boundary as its average representation. Experiments on models from the web confirm the effectiveness and reliability of this algorithm, laying a solid foundation for improving the efficiency of 3D character modeling.