Interactive applications including video games rely on motion-captured or carefully crafted motion clips for real-time character animation. Unfortunately, it is impractical to provide motion data samples for all possible behaviors --- even the simplest example of walking at different speeds require samples for many different speeds. Even if they can carry an increased amount of data, the performance of the application then degrades significantly. Moreover, since the animation mechanism is tightly bound to static data samples, it is very difficult to create realistic responses to unexpected changes in the world. Even with significant advances in research, these fundamental issues have prevented real world applications from creating any realistic virtual characters beyond simplistic and unnatural animation due to performance reasons.;We provide techniques to tackle these issues by enabling motion engine to synthesize animation beyond rigidly defined in motion data samples. We create an efficient near-optimal character controller that can intelligently adopt the input motion data to a dynamically changing environment. We also significantly reduce the data requirement by finding the most effective set of motion data to create the desired behaviors. The controllers can be adapted as basic components in global path planning, and it further reduces the computational burden in the interactive applications. We also introduce motion fields to enable agile and dynamic responses to the abrupt user directive changes or physical interactions.
展开▼
