MoNACA: A System for Anime-like Motion Transfer by Adaptive Partitioning of Articular Trajectories

Abstract

In 3D character animation, frame rate is often reduced to mimic hand-drawn animations, like anime (Japanese animation). The technique includes the omission of motions and emphasis on speed to express movements with a small number of drawings. As such, sampling real motions at equal intervals does not produce the same effect as anime-like motions. In this paper, therefore, a system called MoNACA (Motion transfer by Nakawari Adaptation for Cel-anime Articular trajectories) is proposed to convert motion capture data into anime-like motion. MoNACA determines nakawari poses by purifying the trajectory curves of motion capture data and selecting a set of segmentation points, taking into account the motion speed between each pair of consecutive keyposes and redundancy. Furthermore, it omits minute rotations to reproduce the flatness of anime and automatically locates the optimal viewpoint to enhance the visibility of the motion. The evaluation experiments on the transformed motions and the system functionalities empirically proved that we can achieve more attractive motion transformation effectivity compared to existing methods.