Weta Digital Reverse Engineers the Human Face
The special-effects house behind Avatar reveals a bit of its magic
Photo: Left: Weta; Right: Mark Fellman
Furious Face: Avatar actress Zoe Saldana playing Neytiri, using Weta's motion-capture technology.
If Avatar is the bright future of cinema, a great deal of that dazzle is going to come from Weta Digital, the firm that created most of the movie’s Oscar-winning visual effects.
This past January, IEEE Spectrum visited the company’s headquarters in a homey suburb of Wellington, New Zealand, where key officials spoke at length about cinema after Avatar. There were also a few tantalizing insights into Weta’s work for its next blockbuster, The Adventures of Tintin: The Secret of the Unicorn.
Weta’s specialty is motion capture, which relies on sophisticated software and hardware to transfer an actor’s body movements and facial expressions to an animated character. The actor wears a black suit with light-colored dots; to detect his movement, optical systems track those dots.
For Avatar, Weta pushed the state of the art. First, it employed head-mounted cameras, worn by the actors, that tracked dots on their faces. The use of the camera greatly increased the range of emotions that could be transferred to the faces of the animated characters, enabling audiences to relate more closely to the computer-generated creatures.
Second, the Weta system enabled director James Cameron to see the results of motion capture essentially in real time. As the actors performed, Cameron was able to look at a screen near his camera and see, in place of actors in black suits, a slightly cruder version of the blue computer-generated space aliens that audiences would see.
The most complicated software challenge was coping with the essentially unlimited variety of expressions that a human face can convey. The solution, according to Weta specialist Luca Fascione, depended on identifying several hundred ”key poses”—fundamental facial expressions.
”The computer says, ’I want 30 percent of this one expression and 50 percent of this other expression,’ ” Fascione explains. ”And then the rigging and the machinery behind the puppeteering [character animation] system is able to make the face express that particular emotion.”
Meanwhile, as part of a push to advance the state of the art for the highly anticipated Tintin movie, a team at Weta is helping to devise a new generation of motion-capture software built on a foundation of physiological principles. The group is working on software that incorporates the underlying anatomy of the face.
”What I’m trying to do,” says team leader Mark Sagar, ”is reverse engineer all the expressions in the human face so we can understand the mechanical basis of, say, what makes a smile have a dimple. What makes the creases in a face when it smiles? It all depends on the anatomical structure of the face, the substructures beneath the facial tissue: the ligaments, the fat, the muscles, how the muscles are laid out in 3-D space.”
The Tintin film is a joint effort between Steven Spielberg and Peter Jackson (Jackson cofounded Weta in 1993), with a budget rumored to be around US $135 million, well under Avatar’s reported $300 million to $400 million. Tintin is already the subject of sporadic movieland buzz because it’s understood to be a labor of love for Spielberg and Jackson. Both have professed deep affection for the comic-book series about a globe-trotting boy reporter, his wirehaired fox terrier, and his choleric seafaring friend. The movie, which is to be the first of a three-movie series, is scheduled for release in late 2011.
”Working with such directors as James Cameron, Steven Spielberg, or Peter Jackson, there is never a known path that we’re going to go through,” says Weta R&D director Sebastian Sylwan. ”It’s always trying to push the boundaries of what can be delivered and how a better story can be told.”
This article was updated on 8 April 2010