Real-time Controllable Motion Transition for Characters
Event Type
Technical Paper
Interest Areas
Research & Education
Presentation Types
Registration Categories
Full Conference Supporter
Full Conference
Virtual Conference Supporter
Virtual Conference
Exhibitor Additional Full Conference
Exhibitor Full Conference
DescriptionWe present a novel learning framework consisting of a new natural motion manifold model and a new transition sampler for real-time, in-between motion generation. It generates high-quality motions in mitigating foot skating and motion transitions. It also is general under unseen control signals and outperforms alternative solutions and the state-of-the-art methods.