One of the more interesting challenges wireless service providers face in rolling out video in 3G networks is the amount of spectrum bandwidth they have to support real-time video calls or conferencing.   The issue is particularly challenging in the sense the uplink bandwidth is more constrained.

I have recently observed one potential solution to this problem that uses an animation rendering application on the hand set that could take information about the caller's facial features, expressions etc. and sends that information in place of a compressed video stream.   The amount of data sent is a small fraction of the available bandwidth in a normal voice call and gives a reasonably accurate facsimile of the actual talker's face on the receiving hand set.    The rendered animation has complete facial expressions and emotions with lip movement synchronized to the actual speech.   The animation can be overlaid on any background of the user's choosing and can even be altered slightly to eliminate unwanted information (like grey hair or an unshaved look). 

I can see this approach to video calls to be easily accepted in the market place, especially with gamers who are used to avatars representing them.   It will be interesting to see when and where this technology catches on.