He's Ugly. He's Animated. He's not real...

Back when I was a 3D Graphics weenie working at a company creating 3D effect for film and television, there were two holy grails:

1. Create a virtual room or environment that was indistinguishable from and actual photograph of the same environment.
2. Create a virtual human being that looks, behaves and acts like the real thing -- oh, and do it in real time.

Number 1 happened quite a while ago now...but number 2 has proven to be quite elusive...there are subtleties to the range of human emotion, expression and verbal nuance that is difficult not only for the graphics engines involved, but for the algorithms that control the motion and react to the audio.

However, at the Game Developer's Conference currently going on in San Francisco, a motion capture company (of all people), Mova, has achieved this goal with spooky accuracy using their Contour Reality Capture System (an array of motion capture camera mapping out 100,000 polygons and mapping the texture captured directly on them), plus their Unreal Engine 3 renderer.

Check it out - then just try and get to sleep tonight:


Mova Contour Reality Capture - Gangster Demo (not real-time) from Joy Stiq on Vimeo.

Beavis and Butthead: Prognosticators! Dear God...

Please correct me if I am wrong, but...

Does not this article in Wired, "The Natural History of the Only Child," sound disturbingly, frighteningly, like Mike Judge's Idiocracy?