Kinect vs PlayStation Eye RGB video quality comparison

onQ

Veteran
YooStar 2 is the 1st game that I seen where we can get a direct comparison of the 2 RGB cameras & from the IGN videos Kinect's RGB camera is looking a little washed out compared to the PlayStation Eye, I'm not sure if it's the game or if PlayStation Eye seems to have the better picture quality of the two cameras






2zodopz.jpg


Kinect



2eds36w.jpg


Kinect

2l9h2d5.jpg


Kinect



______________________________________________________





2ylvjus.jpg


PlayStation Eye

o5tkyv.jpg


PS-Eye


2n66j41.jpg


PS-Eye
 
Last edited by a moderator:
The Kinect keying looks much cleaner. The PS Eye shots show white outlines around the extracted person. Not sure what causes the difference though.

The Kinect depth data is useless to pull an RGB matte because it's too lowres.
 
Kinect depth data isn't useless. It'll help extract the player silhouette which can then be cleaned up optically. PSEye will need optical tracking and background removal which is prone to background noise.

As for comparing cameras, you'll need to compare between games, certainly for PS3, as the quality of PSEye varies considerably between titles. I can't see the Kinect pics to comment, though the game should ideally be adjusting the histogram to match the video better in both versions.
 
The Kinect keying looks much cleaner. The PS Eye shots show white outlines around the extracted person. Not sure what causes the difference though.

The Kinect depth data is useless to pull an RGB matte because it's too lowres.

the depth data is used to remove the background
 
Are you sure about this? The depth resolution is very low, I don't think it can be used to pull a clean matte like this.

If you look at this video it shows how "jumpy" and rough the depth data is when using it for foreground extraction.
http://www.youtube.com/watch?v=6BaWwx5x7nM

but that's homebrew & the software isn't as good as what a dev would use for background removal

e1btpg.jpg


yeah you're right about the depth cam resolution being very low but the devs of YooStar 2 said they had to use the digital zoom to pull parts of the higher res picture instead of using it in the low resolution ,
 
I think that the white outlines you see at various spots in both videos probable is because of the different clothing they wear, some show it better, some don't.
I don't think the RGB-camera in Kinect worse quality, but according to the specs the PSEye can record with 2x framerate at the same resolution. But non of the camera's record at HD-formats I believe.
I would expect these recorded videos to be 24 fps @ 320x240, same FPS as movies with downscaled resolution, wich both cameras should be able to do nicely. :)
 
I think that the white outlines you see at various spots in both videos probable is because of the different clothing they wear, some show it better, some don't.
I don't think the RGB-camera in Kinect worse quality, but according to the specs the PSEye can record with 2x framerate at the same resolution. But non of the camera's record at HD-formats I believe.
I would expect these recorded videos to be 24 fps @ 320x240, same FPS as movies with downscaled resolution, wich both cameras should be able to do nicely. :)

I wasn't talking about the white outline I'm talking about the picture quality , you can clearly see that the PS-Eye video isn't 24 fps @ 320x240. it's 640x480
 
I think that the white outlines you see at various spots in both videos probable is because of the different clothing they wear, some show it better, some don't.
Yes, artifacts like these are to be expected. There is a reason after all why the green/blue screen was invented.
 
I think that the white outlines you see at various spots in both videos probable is because of the different clothing they wear, some show it better, some don't.

The PSeyetoy would have a better situation to clean up edges since it is dark clothes vs bright room color.
 
The PSeyetoy would have a better situation to clean up edges since it is dark clothes vs bright room color.

not really because you can't just remove white because people have white in their eyes & the people skin is white, & that white show up better around the black then it would around other colors,

Kinect is better at background removal because the depth cam don't have to use the colors to remove the background it use the depth map
 
not really because you can't just remove white because people have white in their eyes & the people skin is white, & that white show up better around the black then it would around other colors,

I mean for better edge detection. It's a mather of how complex the edge detection is but sure black easier exposes flawed cut-outs. You get to see results if for example you try the magic wand in Photoshop or other image editor with varying settings and qualities.
 
Back
Top