Rumour: 3D Kinect Like Camera for PS3

[dumbo]

Solid_Snake4RD said:

slowmo said:

Solid_Snake4RD said:

starcraft said:

Wait a moment.

There has been an enormous amount of crap floating around making out the marginal difference in processing power between the Xbox 360 and PS3 to be some huge thing.  Now suddenly a PSEye without advanced chips is going to supercede the Xbox 360's Xenon Substantial onboard Kinect computing power with JUST the CELL?

Right.

its just a rumour and sony didn't say whether it will be releaseing it or not and no word whether it will have a processor or not.

we don't have confirmation of the product and you are jumpin onto the technical details...........lmao

 

and nobody said it would superseed 360 and KINECT abilities

"If Sony were to expand on this and use the Cell Processor to calculate extraordinary 3D accuracy over and above what Kinect can achieve, we could see Kinect easily superceded."

While they never said would, could is still a very powerful term to use about a rumour.

they could have meant that the KINECT which is rumoured not using or using an extra processor.if KINECT doesn't have its own processor then it could be true

AFAIK Kinect takes a 'raw' image from the depth camera, and processes it (locating bodies/skeletons) with the Xenon processor/system resources.

[The chip that primesense use for the post-processing only tracks 2 skeletons, (AFAICT) only produces 2d skeletons and it would add lag to the system - so for kinect, this was sensibly not used]

So, yes that would be directly comparable to doing it directly on the cell processor.

.....

However... if the device is (as stated) being used for 3d chat, then it likely has nothing (whatsoever) to do with Kinect.

Expected eyetoy3d: 2xRGB camera.
Kinect: 1xRGB camera, 1 'depth sensor'.

For 'skeletal detection':
- kinect produces an accurate 3d depth map that gets post-processed to a fairly-accurate skeleton.
- eyetoy3d gets 2 images that would need insane amounts of "black magic" to find any objects.

For 'video chat':
- kinect would have to 'estimate/calculate' left/right eye, along with guessing the bits that it can't see.  It should be able to produce *an* image, but not 3d exactly.
- eyetoy3d gets 2 images, sends those directly to the output TV - and immediately has 'perfect' 3d chat.

If true, this looks more like Sony pushing harder into 3dtvs/"3ds-style tech" than Sony showing an interest in Kinect.  (it should also be relatively cheap... ~2x the cost of a standard eyetoy as it's literally another camera increased bandwidth).