I meant to write this about a week ago when I was sent the link, but I’m being hideously disorganized since finishing my degree (woohoo!!!) and so have only just gotten around to it.

Anyway, this article discusses the relationship between how realistic an artificial human is and how attracted we are to it. The basic concept is that as a robot or animation becomes more realistic we get more attracted to it (a microwave compared to R2-D2 or C3P0 perhaps.) However there is a point where the gap between the robot/animation and a real person is so small that the eye focuses on what’s not there. This discrepancy is known as “Uncanny Valley” and facial expressions are generally the hardest thing to get (convincingly) right.

Given that the face has more independant moving parts to it in a smaller space than any other part of the body the complexity of getting an expression right is fairly straightforward. And given that many of these elements aren’t actually that independant at all and each reflects slightly differently on another, you have yet another layer of difficulty. Also given the range of expressions that can be created by really tiny movements and it’s hardly suprising that it’s apparenlty so easy to “just miss” an expression. And there are, of course, other factors.

So, what next?

Do we keep pushing for the potentially unreachable lifelike proto-human, and put up with the Uncanny Valley for a bit longer or do we opt for stylized graphics?