Except that the DF article was printed _after_ the PC folks had shown that the depth feed is 640x480. They had proof that the depth did not have to be at 320x240, and yet they still kept their assumption that we were for some reason not using that resolution. They should have re-evaluated their data at that point.That's not a particularly fair view. We (everyone not involved in Project Natal) had two conflicting specs published in stores, one saying a 640x480 sensor, and the other saying a 320x240 depth feed. We also have the PrimeSense reference design and tech descriptions that say the depth output is a quarter of the depth camera's resolution. There's also the PrimeSense marketing gumpf that says one of the advantages of their tech is that it uses off the shelf camera components, nothing special, which suggests for economies Kinect could use the same camera for both vision and depth. And given a 640x480 image camera, and therefore a 640x480 depth camera using the same component which is the resolution one lot of retailers specs specified, and given that the depth data is downscaled according to PrimeSense and the other set of specs claimed a 320x240 depth feed, that logically points to a 320x240 depth sensor because that fits all the facts we had available.
It may be wrong, as speculations often are, but it's disingenuous to say DF's viewpoint was probably just picking one bit of info and ignoring the rest. I don't see how anyone of a logical mind could look at the info we had and not see 640x480 video and 320x240 depth streams as the most probable configuration.
The only thing I fault DF for in this case is ignoring the other specs out there detailing a 640x480 feed in favour of the one showing 320x240. Their conclusions were logical, and their assumptions were good, just based off an incorrect datum.
I believe both sensors have the same resolution, and using what you know about the primesense tech, you can probably work out what it is.