Music Driven Dance Animation

Since 1990s, motion capture based motion synthesis has been playing an impor-tant role in games and animations, as well as a hot topic of computer graphics re-searches. However, previous works indicate that smoothness of transitions and connec-tivity of behaviors are difficult to achieve simultaneously. We propose the motion thread network (MTN) to address this challenge. An MTN is constructed by taking as input long motion sequences and creating motion threads between similar frames among these sequences, a traversal of the MTN represents a new motion.We achieve a detailed comparison and analysis with the traditional motion graph method. Especially for the standard motion graph and the structured motion graph, we analyze their features, constructions, advantages and disadvantages compared with the MTN method, and relevant theoretical basis and technological details inside. We im-prove some computations of the traditional motion graph method, such as the motion differences computing, motion fragments clustering and motion traversal calculating. What is more, we improve some common operations of the motion synthesis; for ex-ample, time warping, motion parameterization and interpolation, motion synchroniza-tion and blending. Through detailed comparisons, we successfully prove the advantag-es of the MTN method compared with the traditional method.At last, we present an implementation for rhythmic motion synthesis and synchro-nization to generate dance animations using the MTN method. We introduce the mo-tion thread interpolation and selection. Compared with the traditional methods, the MTN method keeps a simpler process, higher quality in the smoothness of transitions and connectivity of behaviors, as well as smaller rhythmic interval errors, which guar-antees the generated dance animation to be highly matched with the input music.