Friday, August 6, 2010

Inside Xbox 360 Kinect


Microsoft will be introducing "Kinect" in November in order to compete with Nintendo Wii and Sony's new motion sensor piece of shit thing that's coming out. Kinect is gonna be awesome. Click to read the full article and see a few more pixxx!!!



Kinect is as much a product of serendipity as anything else. When Microsoft hired Dr. Ilan Spillinger, VP of hardware and technology for Microsoft's Interactive Entertainment Business, it was to be "deeply engaged on the next-generation Xbox." Microsoft was looking to go beyond the Wii for its next big project, and about two and a half years ago, it started looking at natural user interfaces. At the same time, it had started looking at 3D cameras and input systems. Virtually in parallel, all of the necessary technology pieces to make Kinect fell into place—in particular, PrimeSense's 3D sensor.
What Microsoft considers revolutionary about Kinect—and they do consider it revolutionary—isn't that it tracks your body with full depth mapping, or responds to voice commands, or that it has a standard video camera: It's that it brings all of three of those things together for the first time. It's natural user interface in its infancy.

Raghu Murthi, the general manager for Natural User Interface Hardware, is holding a Kinect, stripped naked, as a dozen people gawk at its innards. The exposed metal seems cold. He's telling us about the optical system—how it sees with the three holes in its head that seem like eyes. Without the plastic housing they look like they're bulging out. We're at the beginning of day-long tour of Kinect, gathered in the Great Room, the living room you wish had, but tucked behind a sliding wall inside one of the many food courts on Microsoft's sprawling campus. 3D sensing has been around for 15 years, Raghu explains. What Microsoft has done, he says, is taken 3D depth-mapping technology that typically costs $10,000 to $150,000, and made it at volume, for cheap.
The way the optical system works, on a hardware level, is fairly basic. A class 1 laser is projected into the room. The sensor is able to detect what's going on based on what's reflected back at it. Together, the projector and sensor create a depth map. The regular old video camera is held at a specific distance away from the 3D part of the optical system in a precise alignment, so that Kinect can blend together the depth map and RGB picture for dynamic, on-the-fly greenscreening.

"Xbox, pause." The Alice in Wonderland clip keeps playing. "Xbox. PAUSE." Johnny Depp's freaky visage continues flitting around onscreen. The Xbox is refusing voice commands, pleas really, to pause the clip, as they're lost in the enormity of the room we're in. Pete Thompson, Xbox Live's General Manager, though visibly agitated, is inadvertently revealing that the most complex hardware component of Kinect might in fact be its audio setup.
The Kinect's size and shape is dictated almost entirely by the 4 microphones located along the bottom. It has to be precisely that large to accommodate the mics and the exact positions they need to be in. The mics, and their placement, is the result of research in 200 homes in the US, Japan and Europe. When you buy a Kinect, one of the first things you'll do is calibrate the audio to fit the room it's in. It's creating an audio profile of the room, mapping out the room's reflectivity. And if you majorly re-arrange your furniture, you'll have to do it again.



Basic voice recognition seems like an easy feat—phones do it everyday. But for Kinect, the situation is different. It's attempting to recognize voices from far away with an open mic without the luxury of push-to-talk telling it when to listen for voice cues. The trick used by Kinect is beam forming, so it can focus on specific points in the room to listen. At the same time, the audio processor is using the echo profile of the room to perform multichannel echo cancellation, so the noise coming out of the TV doesn't mess with your voice commands. That said, there's no way to lock out errant voice commands from your douchier friends: It'll listen to any human being in the room. Even if they have a thick Southern accent, like Hee-haw dipped in red eye gravy, there's a good chance Kinect will understand them: The acoustical model for every country includes regional accents, so whether you're from Boston or Alabama, you'll sound intelligible to Kinect, even if you don't to the rest of the world.




***
A row of Kinects line the wall, 16 robot heads nodding silently, endlessly. The motion is robotically smooth, completely un-biological, but alive and almost sentient. We're inside a Microsoft lab where Kinect is undergoing endurance testing. Xboxes litter the room, their cables hanging out like entrails.
More Kinects are locked in a blue box, a sign warning passersby in all caps, DO NOT OPEN CRITICAL TEST IN PROGRESS. It's a heat test. Kinect has a tiny built-in fan that kicks in on demand in hot environments, when the heat produced by the three sensors and the atmosphere around it mix to create conditions warmer than Microsoft would like. Joel asks Ilan if the fan isn't a just bit of over-engineering, a super-insurance policy against heat after the red ring of death plague that slaughtered earlier Xbox 360s. He replies, "It would good to take it out in the future, and we'll look into it when we start to integrate the silicon, but right now, even if it's a small distribution..." in hot environments, they have to have it in there.
The red ring has been seared into the institutional memory of Xbox, undoubtedly. The way Ilan bristles ever so slightly as he tells Joel and I that Trinity, the fresh Xbox, is "a new device, there's nothing from the past," make that clear.

I'm more focused on the two Kinect prototypes we aren't allowed to photograph, one that looks like the head of EVA from Wall-E, a palm-sized bean shape with two antennae shooting out of the side. It was probably rejected for being too personable. Microsoft wants Kinect to disappear, not be your pet. The second looks a lot like the current Kinect, but more Apple-like, a glossy black center wrapped in a kind of brushed aluminum.
The final design is chosen because of the mics, as explained earlier, but the shape, the angles are set that way because they're supposed to angle from the player to the experience. It's glossy because Microsoft thinks glossy means premium. (Hey guys, guess what? The cheaper matte 360 looks better than the shiny one.)




DOPE!

No comments:

Post a Comment