Automatic Composition Software for Three Genres of Dance Using 3D Motion Data

This paper reports on the development of three automatic composition systems for dance that use 3D motion data captured from the performances of professional dancers. The systems can generate short choreographic works in three genres of dance: hip-hop dance, contemporary dance, and classical ballet. The authors label the method of automatic composition as analytic-synthetic choreography, where the dance movements are segmentalized along a time axis into short elemental motions and synthesized them to generate new dance movements. Both the elemental motions and the automatic composed movements are instantly simulated as 3D animation. After investigating the characteristic features of each dance genre, motion data archives and automatic composition algorithms were devised. We experimentally evaluated the usability of our software and verified its acceptability for hip-hop dance and ballet.