Multi-graph Motion Synthesis

Autor: Kuan-Yi Chen, 陳冠逸
Rok vydání: 2005
Druh dokumentu: 學位論文 ; thesis
Popis: 93
The recent demands on animation from virtual environment in movies and games have encouraged the recent studies in generating animation data in more effective ways. Conventional motion generation method requires a great amount of cost and time. To increase the effectiveness of motion generation, we should re-use the existing data and synthesize the new motion data. In this thesis we propose a method to increase the variety of motion from the original motion data using low amounts of cost and time. Using the existing data, we can synthesis newly motion that differs from the sample motions. In our method, animation data is segmented into several body parts. We then calculate the frame-to-frame relation by calculating the frame distance in order to generate the multi-graph which represents available motion of each body part. By traversing all motion graphs together we determine the clip combination and blend them to generate new motion data. Since the graph traversal of each body parts varies in paths, we can generate the combination of existing motions.
Databáze: Networked Digital Library of Theses & Dissertations