Follow by Email

Saturday, January 29, 2005

Why Animate Human Faces?

I met Neil Krepla at a birthday party today. Neil has been a visual effects supervisor on many movies over the last n years (n > 20). We had a nice talk about the benefits of Renderman over ray tracing, interesting changes in visual effects over the last n years (n > 20), how he watches visual effect films and many other topics.

We spent a good amount of time talking about human faces, and the difficulties and failures of human face animation that make it so unconvincing. I suggested that the Final Fantasy production team did a pretty good job of animating human faces and he said that they did a good job with poses, but once they moved, they gave themselves away. He explained to me one of the big flaws of The Polar Express was that you could not see the childrens' souls in their eyes. I said, "I just don't understand why they do it. If it's unconvincing, if they can't hide the fact that a human face is a visual effect, I'd just rather not have them do it."* He said the new Lemony Snicket movie has a young child, and that they did a surprisingly effective and seamless integration of a CG representation with twin child actors which was important because, hey, you couldn't have a two-year-old performing all these dangerous stunts. Apparently this movie is actually successful in pulling it off. I'm interested in seeing for myself.

Here's an article about CG on Lemony Snicket, and Sunny in particular.

* Ok, I didn't really say that, but I'm making the story flow.

1 comment:

Zach said...

My sister, Jess, is a computer graphic artist and we've talked about this numerous times. We both agree that the reason why it is so difficult to animate the human face is because we, as humans, recognize it. In other words, we spend most of our lives looking at people, their faces, and their movements. We are intimately aware of what a person's face looks like when it smiles, frowns, or makes any other movement. So to animate it is extremely difficult to get EXACTLY right.

It is similar in concept to the movie Spiderman. I felt that they did a decent job of animating Spidey's movements. Given the difficulty in getting human movement exactly right, we recognize the slightest imperfections. So even though they did a good job most of the time, when they mess up even the slightest movement, I realize it and remember it.

It's even more obvious in a face with such tiny fluctuations that you are subconciously aware of...

I have no doubt that some day a software algorithm will be written that can accurately depict a human face. But I wouldn't be surprised if that is a few years from now...