The Massachusetts Institute of Technology (MIT) has licensed Eyetronics' ShapeSnatcher software to experiment with what researchers call videorealistic talking humans that would be used in TV, film and games. ShapeSnatcher Suite 3.0 is an integrated software package that transforms an average slide projector and digital camera into a sophisticated 3D scanning and modeling system. "Our project aims to use ShapeSnatcher to record a human subject's mouth while talking, and then re-animate the subject to say whatever is desired by the animator," says Tony Ezzat, a doctoral student in MIT's computer science department. "ShapeSnatcher enables us to acquire 3D mouth shapes with texture." The goal of the project is to create a videorealistic text-to-audiovisual speech synthesizer. The system should be able to take any typed sentence and produce as output an audiovisual movie of a face enunciating the words. The final output would look like a video recording of a talking human. ShapeSnatcher is being used to scan and render 3D images of people talking so that a full range of mouth movements can be recorded. The ShapeSnatcher system is already being used in animation production, game development, research, engineering and retail to create accurate 3D models from physical objects quickly. The software runs on most PCs using Windows or Linux as well as on SGI IRIX systems, and it is compatible with popular 3D graphics and CAD formats.
