Replacing human actors all together?
Well human psychology may get in the way. Humans don't particularly respond favorably to 'real lifelike' robots and or avatars because there is 'something' missing. And the 'missing' part is cohesive body language. With synthetic actors it is 'emotional body language' that is askew or missing. The final fantasy movie was great but the body language and the integration of emotions was horrible despite the photographic realism that was achieved. Thus people didn't like it as much as the creators thought people would - and the public stayed away.

If such things are done properly - human psychology will NOT get in the way as much. With interactive avatars/robots increased acceptance can not be achieved by only focusing on language skills and AI.
With copying human body language it is possible to have humans respond more favorably to the lifelike avatars/robots. A nice trick for increased acceptance of interactive robots/avatars is to have the lifelike avatar/robot make the same body movements (head movements - are researched quite thoroughly) as the human opponent made but then with a 2 minute time lapse. For some reason the human short term memory will pick up on the movements - recognize the pattern subconsciously and the 'mirror neurons' (empathy neurons) in the brain will kick in and make the human feel more empathy toward the avatar and thus accept this robot more readily and even feel some sort of emotional connection with it.

In real life situations you can use this 'trick' of mimicking body language and posture to make your 'opponent' feel more at ease and respond favorably to you. But be careful - by overdoing it or timing it wrong you can achieve the opposite. People mimic postures all the time - it's the mirror neurons accountable for empathy that are at work here. It is a delicately fine-tuned system that is very very hard to fool though.

Synthetic actors will have to use captured human body language and facial expressions in order to display cohesive body language otherwise they just 'will not fly' and the movies will BOMB. Mind you - body language is very subtle and people respond to each other in a subtle way. So simply recording human movement and emotions in a library and using these 'stored' cohesive movement patterns and mapping them on the synthetic actors won't work as they need to respond to each other appropriately in order to have the public 'feel' the interaction. Each person is different in these subtleties. So even if you manage to capture and catalog all the movements of a single actor systematically and map them to some sort of emotional states so you can re-use those movements and expressions to create new scenes with a digital avatar - it will be very hard to make these avatars interact appropriately in a way that feels right.
Say you did something like this for Meg Ryan and Billy Christal - it will be virtually impossible the create the same sizzling scenes from the 'When Harry met Sally' just because the interaction will be askew as intricate interaction, timing and responses of the mirror-neurons typically for both actors just won't be there and it won't feel right.

It will take a long time to get this right - good (method) actors are better at it! So perhaps digital isn't better than analog actors. Method acting is in my opinion superior to any other form of acting as the actors are actually using the parts of their brains that would normally be used in the same 'real' situation.

Xbox 360: Lactobacillus P | Wii: 8151 3435 8469 3138
Armchair arcade Editor | Pixellator | www.markvergeer.nl