Sunday, August 12, 2007

MoCap, Mo Crap

Judging from their omnipresence at this year's SIGGRAPH (Aug 5-9 in San Diego) this may be the year of MoCap.

MoCap, or Motion Capture (or "The Devil's Rotoscope", depending on who you ask) is the application of ping pong balls all over an actor's spandex bodysuit, supposedly to embarrass them when the "behind the scenes" DVD extras come out. These magical ping pong balls somehow deliver spatial information to a computer, from which at the push of a button an actor in a blue bodysuit is transformed into a seven-headed fire-breathing hydra. Furthermore, the computer automatically composites, scores and edits the film, sends it to theaters and reviews it for the New York Times. And we can all go home.
Seriously, though, great strides are being made to evolve past the ping pong balls and blue spandex. This years' crop seemed more accurate and less awkward than ever, as is to be expected. This trend will continue until usable animation data can be seamlessly extracted from any actor or other moving material within an arbitrary frustum (camera's field of view). At that point, even effects animators will be crying foul play, as CG data for fluids and particles can be gathered from real fluids and particles.
Currently the focus is on character animation, the field that feels most immediately threatened by this technology. Most guys I work with hold to an ethic that animation is all about subtle exaggeration, something the computer can't deliver. However, implicit elsewhere at SIGGRAPH were demos and lectures indicating startling advancement in artificial intelligence, with computers able to follow basic rules of design and composition to spit out abstract expressionist paintings and believable Haikus. Why wouldn't they be able, fed the basic design rules of cartoon animation, to deliver a convincing Bugs Bunny performance better and faster than Chuck Jones’ brightest HomoSapiens pupil? I was being only halfway facetious when I spoke of the computer's ability to edit the film and write the review.
It comes down to this: for every automatable task, humans will probably be outshined by computers in the near future. And this includes so-called creative tasks, since creativity can honestly be derived from a finite set of rules and grammars (the wildcard of novelty being one of them). But does this mean the animator is going away? No more than painters went away when photography showed up. You see, there are two reasons we do things: 1. because we have to, and 2. because it’s fun. Even when we don't have to do our animation, the field will still thrive because it’s fun. It will still have a raging, avid audience. Because it is difficult and laborious to hand-craft animation, it excites and inspires the audience. An audience always wants to see evidence of hard work. Tell them an animation is hand-crafted, and an audience will always appreciate it more, even to the point of forgiving some technical flaws. Pixar is on to this, boasting in the credits of Ratatouille that no MoCap or other "shortcuts" were used in the making of the film. But in fact, MoCap as it stands today is not a shortcut. A fad, if anything, that will make the movies using it appear dated in a few years. Remember when drum machines briefly replaced human drummers in mid-80s pop music? Have you gone back and actually listened to Bananarama or the Jets recently? At the time of this writing, MoCap is extremely expensive and time-consuming to wrangle (it was estimated by those involved that Polar Express took twice as long to animate because of all the complex MoCap data than if they had hand-animated it from scratch). Why, then, is this unwieldy, laborious juggernaut the moment's hot item? Well, for one, investors get excited about these "bleeding edge"-type technologies, the ones that appear to be the "next big thing", like MoCap, dot-coms and drum machines. If you get in early, you get rich, so the conventional wisdom goes. So Zemeckis is a rich man, having so embraced MoCap to his investor's delight. And these investors are told (correctly) that soon, very soon, MoCap will cross the line and become less expensive than traditional (keyframe) animation, delivering better results, faster. Thing is, and we all must remember this; MoCap is not a direct competitor to keyframe animation, but rather a supplement - and in this, the Rotoscope analogy is correct. You can use roto correctly, as Disney's Nine Old Men did, or you can abuse it, as Ralph Bakshi did. Now his films look dated. Soon most major animation and effects companies will be using some form of Mocap, be it full-body systems or small hand-gesture input devices at the animator's desk. It is in the way that it's utilized and integrated that will separate the proverbial men from the boys.

The Expo Floor being set up

Rundown of SIGGRAPH: MoCap aplenty, lots of chat about image-based this'n'that, film and game integration, and touch screens.

Autodesk had a party on an aircraft carrier. Tickets were so limited that only two types of people got in: the VIP set and the college fanboys who had nothing better to do than camp out all day waiting for the mad rush when tickets were passed out. The SoftImage party/riot was equally infuriating. Speaking of MoCap, I went to Zemeckis's party, which was nice; ran into a couple of Pixar blokes who seemed to actually be considering a job there. Don't tell their boss.
All in all, SIGGRAPH is the Expired to Comic-Con's Tired. I'm not cool enough to know what's Wired. But it is still the most legit of the geek conferences, where collaboration and learning can actually transpire.
One more item. Will somebody please inform the art schools: stop bussing all your students there to stand in long lines with their resumes. I've known only one person to ever get "hired" at SIGGRAPH, and he would've been hired anyway. The best way to have your demo reel stand out from the crowd is to NOT throw it into a gigantic bin of thousands.

Glad conference season’s over. Back to work.

On the iTunes this week: Big Star's Third/Sister Lovers.

No comments: