Open Collections

Abstract

Keyframing is the main method used by animators to choreograph appealing motions, but the process is tedious and labor-intensive. In this thesis, we present a data-driven autocompletion method for synthesizing animated motions from input keyframes. Our model uses an autoregressive two-layer recurrent neural network that is conditioned on target keyframes. Given a set of desired keys, the trained model is capable of generating a interpolating motion sequence that follows the style of the examples observed in the training corpus. We apply our approach to the task of animating a hopping lamp character and produce a rich and varied set of novel hopping motions using a diverse set of hops from a physics-based model as training data. We discuss the strengths and weaknesses of this type of approach in some detail.