Data-Driven Group Animation

Abstract

We introduce Data-Driven Groups (DDGs), an animation technique for groups of discrete agents, such as flocks, herds, or small crowds. DDGs are a natural extension of human motion graphs to groups. They create motion by piecing together a sequence of recorded motion clips. The graph structure identifies clips that can be appended while maintaining continuity in the motion. We discuss a method for building DDGs for discrete agents and algorithms for extracting motion from the graph to meet environment constraints. The resulting animations show realistic motion at significantly reduced computational cost compared to simulation.