The goal of our part in this project is to develop and demonstrate a low-cost 'solution' (or system, if you like) for high-fidelity real-time netcasted VRML-based talk-shows.

The idea is to simultaneously netcast/broadcast a talkshow -think Larry King Live- on TV and over the Internet. Photo-realistic avatars of host and guest will feature high-fidelity real-time facial expressions and mid-fidelity full-body (or maybe only torso) motion.

Added value for spectators: free positioning of viewpoint, self-determined switching between various pre-set camera positions, continuous voting on popularity (or not) of both host and guest (with appropriate cartoon-like visualizations of the consolidated audience vote), ability to walk up to the conversation and ask questions to the guest, *become* the 'instant-made' guest... We're still looking for an appropriate talk-show that wants to join in the experiment (and go live with this on TV + the net). Technological cooperations-and commercial for the future marketing of this solution- also welcomed.