 |
|
|
| |
Interview: Making Faces With Team Bondi's McNamara, L.A. Noire
by Leigh Alexander [Console/PC]
|
|
| |
|
March 15, 2011
|
| |
One of the main selling points for Rockstar and Team Bondi's upcoming L.A. Noire is its sophisticated facial capture technology that allows the game's animators to convey an unusually lifelike rendition of an actor's performance.
This drives the gameplay, which casts the player as a detective who needs to read the subtleties of characters' expressions and behaviors to divine their truthfulness and how best to question them.
The gameplay hinges so much on the technology that one wonders which came first. Gamasutra spoke to Team Bondi founder Brendan McNamara on the concept's genesis: "The idea for the facial tech came a long time ago, just out of frustration with working with motion cap," he tells us.
"Motion capture works with rotations to try to find out where a bone is in your body, and as amazing as that has been, it's been frustrating with faces, because faces are all about muscles contracting and expanding," he continues. "There was no real clever or accurate way of doing that using that kind of tech."
"We'd always been able to make great cars and great buildings," says McNamara, who also worked on The Getaway with Sony in London before founding Team Bondi in 2004, "but the faces had never really held up."
The lack of realism in game faces becomes increasingly noticeable as realism in the rest of game animations and objects advances, he says.
So the team, together with sister company Depth Analysis, decided to begin work on the MotionScan technology that L.A. Noire uses to bring actors' facial performances to life in detail.
It took five years, says McNamara. "We have lots of different versions of it working, and the key part before we did any of the actual capture was the compression part, [which] we kind of worked out on paper what the numbers were going to be."
"It was capturing about a gig per second, and we thought how were we ever going to play it on a game console... 30k is the framerate," he reflects. After all that iteration and experimentation, MotionScan now handles some 350 terabytes of raw data for L.A. Noire.
"The pipeline, funnily enough, is one of the most straightforward parts of it," says McNamara. "What I didn't want was any interpretation between what the actor did and the end result. I think that's the point where somebody else is putting their personality onto it."
"This processing doesn't really allow that; what you see is what you get, and once it gets processed it goes right into the game and the head just gets parented onto the skeleton."
The result of the tech development is the opportunity to do something McNamara feels is "truly character-based," especially given they have access to acting talent like Mad Men's Aaron Staton. McNamara says the performances that appear in L.A. Noire are half the designer's intent, half the actor's work, an unusual collaboration.
"And if you're going to do that, and you have access to great actors, you want them to do something different -- with this kind of game, we do." L.A. Noire is slated for release Xbox 360 and PlayStation 3 in spring 2011.
|
| |
|
|
