Features// Natal: Hands-On and How it Works

The joys of GamesCom are manifold: it doesn’t take place in the most depressing city in the whole of Germany, you can get a flight straight to it, and there are hotel rooms available in cities less than an hour’s drive away. And, more importantly, it occurs after E3 and isn’t staffed by scary US PR gatekeepers who overtly laugh at your requests to book appointments if you don’t happen to speak with an American accent. As a consequence, we finally got a chance to have a go on Microsoft’s Project Natal.

Having attended Microsoft’s impressively theatrical press conference at E3, we were impressed with Project Natal, but also had reservations. Could it really work as well as it appeared to in that press conference, or were invisible wires and laborious prior configuration involved? Would it be fazed by the hustle and bustle of the typical front room, where it will be expected to reside? Would we be able to break it, inducing glitches when we took it out of the comfort zone of its customary demonstrators? And how does the bloody thing work?

Well, if we were impressed before, we’re now blown away having had a play on it. Sorry. We tried to retain some cynicism, but Project Natal rendered that an impossibility. Certain aspects of the system were very striking. Such as the fact that it requires less calibration than a Wiimote – none at all, in fact. We were in a group of six people all checking Natal out, with new people jumping in to have a go every few minutes, and never once did it miss a beat.

At the end of the demo, Microsoft showed a screen from the devkit which provided a brilliant visual illustration of how the system works, and explained why it doesn’t need any calibration. If Microsoft had shown that screen at the E3 press conference, everyone would instantly have suspended disbelief. Microsoft’s Kudo Tsunoda, who was in charge of the demo (and used to run EA’s Chicago studio), was at pains to explain that the key to Project Natal is that it is just on the lookout for human bodies. Thus, when someone stands in front of it (it can also cope with people who are sitting), it thinks: “There’s a body in front of me: I’ll scan it and lock onto it”.

The devkit screen showed the different live feeds of information that the system was generating, as well as what it was doing to them with some serious algorithmic programming – it has its own processor, so won’t place any extra load on the Xbox 360. Tsunoda explained: “It has an RGB camera and a multi-array microphone, which allows things like visual sign-in and voice recognition. But it’s the infra-red feed that gives a depth-map, like when you push your finger into one of those toys with all the metal pins. That depth-map gives joints that it tracks – 48 per skeleton – and therefore a full-body skeleton.”