On Tue, 2010-11-16 at 16:40 -0800, Scott wrote: > http://www.primesense.com/?p=515 > > > Is the company that provides the underlying technology, and a > middleware layer. Unfortunately, it doesn't look like they've made it > easy for just normal folk to get them.
I had access to the primesense camera when I was working for another games/hardware company who were considering going in this direction. The depth data from the camera is certainly an order of magnitude better than relying on rgb for algorithmically understanding the scene, but it of course brings it's own problems with it. Anything reflective or with high ir (like patches of sunlight) can cause big spikes in the depthmap for example. I haven't had access to the kinetic system, but I'd like to see how good the skeleton matching is - the reviews I've read have been luke warm on this feature, as it seems difficult to make some of the gestures work. It will be interesting if this gets fixed with a bit more experience from the games developers or whether it's a lower level problem. In any case, I would say that interesting stuff can definately be done with the raw data from the camera. I would also imagine it won't be long till we start to see free software libraries appearing to do different things with the data, and these are likely to end up more interesting/useful for more general purpose stuff than the proprietary APIs. cheers, dave
