A New Method for Automatic Generation of Animated Motion

A new method for generation of animation with a quality comparable to a natural motion is presented. Proposed algorithm is based on fuzzy description of motion parameters and subjective features. It is assumed that such processing increases naturalness and quality of motion, which is verified by subjective evaluation tests. First, reference motion data are gathered utilizing a motion capture system, then these data are reduced and only main poses of the action are left. The resulting motion is simplified and its quality is considerably decreased. Then, utilizing the automatic motion enhancement system. ANIMATOR, a new version of the action is generated, based on input poses and subjective descriptors given by the user. Finally, a comparison between the original and recreated motion is performed. Presented method is useful for automatic motion generation and can be paired with motion data reduction procedure for regaining naturalness. Moreover the reduced version can easily be edited in the ANIMATOR system, and in this way a new action can be created.