Getting Wild About Facial Animation
At first, Spike Jonze wasn't sold on a CG approach to the facial animation for the creatures that populate his adaptation of Maurice Sendak's Where the Wild Things Are. He wanted to try animatronics, but his good friend David Fincher convinced him otherwise by showing him some of the early tests they were doing on The Curious Case of Benjamin Button (which, ironically, Jonze turned down). And that gave him the faith he needed to go down this path.
Not surprisingly, it was an unusual path -- even for CG.
The creatures were defined; the movie had been shot; and there was a rough cut, but the faces were static because of the masks. But the emotional impact was still apparent.
In steps Daniel Jeannette as animation/visual effects supervisor, who had previously worked on Happy Feet at Animal Logic after a prestigious stint at ILM.
"Early on, somebody said you should use CG creatures because it gives you the freedom to do what you want," Jeannette recounts. "It gives you a lot of control, but there's also a lot of limitation that comes with it. And, ultimately, you're talking about a lot of greenscreen work. This didn't work with Spike's vision of shooting on location or the core of the story, which is the relationship between Max and the creatures. It was better to have him interacting with actors in costumes. It was real, the performances were nuanced and emotionally it hits you in the gut."
They tried a lot of techniques before realizing that the best approach would be limited facial animation enhancement whenever they needed lip-synch. In other words, strictly for the movement of the face, they only animated the eyes, lips, teeth and tongue.
And the faces in Wild Things were brought to life at London-based Framestore primarily using a new twist on an old technique known as "sequence projection" or "projection mapping," often used to create talking animals. For Wild Things, however, it had to be taken to a new technical and artistic level. Character suit performances were filmed with static faces. The faces or heads were tracked in 3D and a CG articulated head was animated on top to achieve the appropriate performance to match suit and dialogue.