3D facial movement replication using digital video
Date of Publication
Bachelor of Science in Computer Science
College of Computer Studies
Ronello T. Jose
Defense Panel Member
Florante R. Salvador
Ruiz R. Conrado, Jr.
The Philippines is one of the growing game development countries in the world. As of now the game titles that have been produced by local companies are less than a handful. Some of today's games are becoming more and more realistic in terms of the look, the feel and the motion that is done by the characters within a game. Motion capture is one of the most optimal way of replicating movement for 3D movement.
The research has devised a way to replicate, or at the very least approximate, the movements of human face in a given digital video to a 3D face model. The proponents have utilized image processing techniques for movement detection and XML scripting for movement encoding. As for the #D face renderer, the proponents have used the XFace Toolkit. The system posted an average correctness of 80.60% using frame per second comparison of the input video and 3D animation sequence output. The proponents also conducted a survey to reduce the subjectivity of the visual comparison, which in turn has an average of 77.06% confidence value.
Archives, The Learning Commons, 12F, Henry Sy Sr. Hall
Espiritu, D. G., Mendoza, R. A., Nieto, R. L., & Uyecio, A. C. (2006). 3D facial movement replication using digital video. Retrieved from https://animorepository.dlsu.edu.ph/etd_bachelors/14390