Take a look at this 3D morphable model face demo by Volker Blantz. The automated matching of Tom Hanks' face is amazing, but I was even more impressed with the Audrey Hepburn and Mona Lisa sequences.

Apparently, it can sample multiple facial models and extract bone structures, facial expressions, gender, and more. It does seem to approach the uncanny valley, and at other times caricature, but overall, this is very impressive. I guess that's one step closer to S1m0ne.

[via Motionographer]