GPU instancing method for crowd animation based on motion capture data

Crowd animation with high-complexity geometry model entities becomes the problem. A novel geometry instancing method for motion capture models in crowd animation is proposed by analyzing recent crowd visualization methods. Firstly, by the method, the motion state updating and geometry transformation of the entities are made to fit parallel process through preprocessing of the motion files, then a vertex sequence of the geometry mesh models with different subdivision layers is created automatically, the crowd animation is realized by instancing method. Finally the simulation proves the proposed method achieving high-performance on GPU for crowd animation with high-complexity geometry model entities.