Manifold Learning Techniques for Editing Motion Capture Data
Streamlining the process of editing motion capture data and keyframe character animation is a fundamental problem in the animation field. This paper explores a new method for editing character animation, by using a data-driven pose distance as a falloff to interpolate new poses seamlessly into the sequence. This pose distance is the measure given by Green's function of the pose space Laplacian. The falloff shape and timing extent are naturally suited to the skeleton's range of motion, replacing the need for a manually customized falloff spline. This data-driven falloff is somewhat analogous to the difference between a generic spline and the ``magic wand'' selection in an image editor, but applied to the animation domain. It also supports powerful non-local edit propagation in which edits are applied to all similar poses in the entire animation sequence.