Getting animated about crowds

For film studio animators, creating crowd scenes that merge animation and special effects with humans is an art form, writes KARLIN LILLINGTON.

WHO WOULD have guessed that crowds are so complicated?

Merging into a crowd may be easy for humans, but for the animators and technologists at film studios that make the computer-generated animation and special effects we love to marvel at, a crowd is an art form.

According to two crowd experts, technical director Paul Kanyuk of San Francisco area film studio Pixar, and production technology lead Hans Rijpkema of Los Angeles-based special effects company Rhythm and Hues, the problem is partly one of hardware and software – technical limitations that don’t quite, at least yet, enable animators to do this or that.

And the other difficulty is making a crowd look, well, like a crowd. Even though we tend to think of a crowd as something going on in the background that we barely notice, in fact humans are alert to even very small details.

The two explained such challenges on a visit to Dublin last week, when they gave talks at the Metropolisexhibit on animated crowds at TCD's Science Gallery.

So what makes a good crowd? “One word: variety,” says Rijpkema. “That’s the number one flaw you can often find in computer generated crowds.” Sometimes the motion is all wrong “because there’s a couple of characters who walk the same way. It’s the motion of them, the size of them. They’re all walking the some way.”

"Rats were easier than humans," notes Kanyuk, whose first major job at Pixar was animating crowds of rats in Ratatouille. Audiences are less familiar with rats – animated or not – and are less judgmental, he says, whereas people know exactly how humans should look and seek detailed accuracy.

"Variety is first, then to make the crowd actually look good, the motion becomes a lot more important," he notes. For that reason, animating crowds in Wall-E"was a challenge because the characters all wore a red unitard as a story point – that made it more difficult to get variety".

Kanyuk enjoys working on foreground crowds – the creatures that are the main actors and more noticeable. “You want to catch the viewer’s eye and make people think, ‘wow, I’ve never seen anything like that before’. These are very special things, a lot of these crowd shots. There’s a spectacle to it.”

Creating huge, active special effects scenes for films such as Night at the Museumor The Lion, the Witch and the Wardrobe– both Rhythm and Hues projects – present their own difficulties. Some of the creatures are mechanical, controlled by software-driven rigs, while others are computer-generated and dropped into the scene after the "real" scene with the human actors has been shot.

All the members of the scenes “have their own motions and set of motions,” says Rijpkema. “How would a minotaur fight a centaur? If we have these basic motions we can create a believable battle out of it.” But they will only have the resources – technical, human and time – to do a detailed subset of what they could do, he adds.

They start cutting back, especially on types of motion that might be too complicated to do or where they haven’t quite figured out the logistics of movement. “We’ll have rules; [maybe] no dwarf will ever fight a faun, basically it will not happen. But it’s a lot of planning to come up with the right amount of motions.”

Characters in animated crowds are given movement and some independence of action using basic artificial intelligence in animation software, sets of rules as to how they move, where they are trying to get to and whether they will turn right or left or reverse if some other object gets in their way.

"There are limited set of rules but every character has a different perception of the rules that makes them ultimately make different decisions," says Kanyuk, allowing the rats in Ratatouilleto run realistically across a scene.

But things also get more complicated, as with the battles in Narnia, where there are many subgroups of crowd activity and many different characters doing different motions. Sometimes, says Rijpkema, the "intelligence" doesn't quite work and random characters run the wrong way or two characters run right through each other. Such glitches are removed by hand, but if less noticeable might be left in, as was the case with one rampant centaur in a Narniabattle scene.

And though for viewers, CGI (computer-generated imagery) films are cutting edge and magical, the specialists who make them are never quite happy with the abilities of their hardware and software and push them to the limit.

“In just the four years I’ve been at Pixar, we’ve gone through several generations of computers and faster and faster processors. As soon as we have it, we just eat up all the new resources. We’re always on the edge of what the computers can handle. Our creative appetite typically outstrips the technical resources on every single show we’ve worked on,” says Kanyuk. “It’s hard to say no to directors at Pixar so very often, whatever they ask for we come up with.”

Merging the real and animated worlds has its own set of demands. “You have to rebuild the live environment in the computer. It’s Ben Stiller standing there and he has to have a lion standing right next to him and the shadows have to fall in a certain way. To have that integration work so people don’t think it’s two separate worlds, its pretty complicated,” says Rijpkema.

But fun, they add. So do they have the best jobs in the world? “You’re welcome to be a big dork at Pixar,” laughs Kanyuk. “Its a lot of work but it’s fun. It’s cool.”

Listen to the complete interview with Hans and Paul at www.techno-culture.com/podcasts