
|
Facial Modeling in GamingJune 3, 2005Just as a segue from the Quake 4 trailer, it is *amazing* to see how realistic game characters are now. From a purely subjective point of view, I'd say that Quake 4's Matthew Kane (shown above) looks far more realistic than, say, j-pop idol Ayumi Hamasaki:
Granted, it's not really fair to use Ayumi Hamasaki as a counterpoint, given how much she's been plasticized and photoshopped, but nevertheless, it's really amazing that id has managed to render such life-like models, since the brain is notoriously good at detecting anomalies in facial images. As a matter of fact, this ability is hard-wired into the brain at the cellular level (see neato diagram). So, just as a question, wouldn't it be interesting to see if there's a critical point where people are physiologically unable to distinguish between a real person and an AI model? I mean, of course people aren't tricked by a low-resolution 500-polygon wireframe, but what if we raise the polycount to 4000? 5000? 10000? A quick search on Pubmed uncovers a bunch of interesting articles (see postscript) but none of them really focus on image degredation, either from a lossy compression / 3D modeling point of view. Anyway, very interesting topic -- shall be sure to look into it later in more depth. * * *
|
