Communication through virtual reality

Communication through avatar-based virtual reality can, because of loss
of nonverbal communication, be more ambiguous and limited than talking
to someone in person or via web camera. A way to leverage this would be
to map facial expressions captured by a web camera to an avatar, for
more realistic communication.

The company neven vision has a patent, Wavelet-Based Facial Motion
Capture for Avatar Animation, that describes a technique for doing
this. Are there any non-patented alternatives that are more appropriate
for open source projects like Croquet?

Mats

We are using OpenCV which is an open source vision library to support  
Webcam based interaction with kids. OpenCV supports face tracking and  
with some cleaning up, it should be possible to use the face as a  
texture for the avatar. However, transferring the image data in real  
time is another story.

However, I think face mapping on avatar is more suitable for games  
like Second Life in which the avatar takes strictly human forms.  
Croquet's avatars are more playful and it would be harder to map a  
human face to a bird.

We are taking the path of gesture recognition to let kids to control  
avatar using the body language. Right now, just simple run, jump and  
flip wings. Just simple things to get started.

There is Europe IST project COGAIN (Communication by Gaze  
interaction) that's working on web cam based Eye Tracking. I think  
it's more natural integration with Croquet like environment with body  
gesture.

OpenCV
http://sourceforge.net/projects/opencvlibrary/

COGAIN
http://www.cogain.org/

On May 1, 2006, at 5:25 AM, Mats wrote: