Abstract: To meet the requirement of controlling the virtual crowds in film production,a novel shape-constrained crowd animation system is proposed.Given the source and the target shapes of a crowd,the correspondence between their sample points is computed by using a greedy algorithm.Then,the correspondence is optimized to minimize the path crossing while maintaining the relative positions of the individuals.Finally,a visually plausible crowd animation is obtained by applying the motion data on the path.With this approach,the goal of editing and controlling crowd motion can be achieved simply by giving the crowd size and shape constraints.Our experiments show that the method is capable of controlling crowd motion with different sizes and shape constraints.