This presentation introduces Motion Matching, an innovative new approach for creating high quality, fluid and complex character movement. The team adopted a declarative animation philosophy, where instead of placing small animations in a big structure, they place small structured markup on top of long animations. This manual markup is necessary for logical information that can't be inferred automatically, like attack types and defense stances. For navigation, they don't have to manually organize transition animations like starts, stops, and turns. They just capture 5 or 10 minutes of a person running around and import it directly into the engine.
At runtime, they continuously find the frame in the mocap database that simultaneously matches the current pose and the desired future plan, and transition with a small blend time to this other place in the data. The resulting motion is almost indistinguishable from a raw mocap sequence, while being responsive enough for comfortable control.

plus-circle Add Review

comment

Reviews