Campus Event Calendar
Event Entry
New for: D1, D2, D3, D4, D5
What and Who
Synthesizing and validating nonverbal behavior for virtual characters
Michael Kipp
Cluster of Excellence
Talk
AG 1, AG 2, AG 3, AG 4, AG 5, SWS, RG1, RG2
AG Audience
English
Note: We use this to send email in the morning.
Date, Time and Location
Wednesday, 16 July 2008
16:00
60 Minutes
E1 4
024
Saarbrücken
Abstract
Virtual characters that move and gesticulate appropriately with spoken text are useful in a wide range of applications, from interactive games to multimodal human-computer interfaces. However, this class of movement is very difficult to generate, even more so when a unique, individual movement style is required. In this talk, we present a combined knowledge- and data-driven approach to gesture synthesis, the behavior analysis tool ANVIL and an emerging standard for behavior specification, the behavior markup language (BML).
We present a gesture synthesis system that is capable of producing full-body gesture animation for given input text in the style of a particular performer. The process starts with video of a person whose gesturing style we wish to animate. A tool-assisted annotation process is performed on the video, from which a statistical model of the person's particular gesturing style is built. Using this model and input text tagged with theme, rheme and focus, our generation algorithm creates a gesture specification that is rendered using either kinematics or physical simulation. As opposed to isolated gestures, our system produces a stream of continuous gestures, so-called g-units, that are coordinated with speech. In two subsequent user studies, we successfully validated (1) that the produced gestures are consistent with the original performer's style and (2) that our use of g-units caused the virtual character to be perceived as more natural.
We argue that a standard for movement specification is required as a high-level interface that allows to create virtual characters without being a computer animation expert. We present an emerging standard, the behavior markup language (BML), using our system to exemplify where and how BML fits in the system framework. The talk concludes with future directions and challenges.
We present a gesture synthesis system that is capable of producing full-body gesture animation for given input text in the style of a particular performer. The process starts with video of a person whose gesturing style we wish to animate. A tool-assisted annotation process is performed on the video, from which a statistical model of the person's particular gesturing style is built. Using this model and input text tagged with theme, rheme and focus, our generation algorithm creates a gesture specification that is rendered using either kinematics or physical simulation. As opposed to isolated gestures, our system produces a stream of continuous gestures, so-called g-units, that are coordinated with speech. In two subsequent user studies, we successfully validated (1) that the produced gestures are consistent with the original performer's style and (2) that our use of g-units caused the virtual character to be perceived as more natural.
We argue that a standard for movement specification is required as a high-level interface that allows to create virtual characters without being a computer animation expert. We present an emerging standard, the behavior markup language (BML), using our system to exemplify where and how BML fits in the system framework. The talk concludes with future directions and challenges.
Contact
Conny Liegl
302-70150
--email hidden
passcode not visible
logged in users only
Conny Liegl, 07/16/2008 15:38 -- Created document.