Hello! I am thinking of buying ps3 eye cameras. Just wondering what is the time period in capturing one frame over USB2.0 connected to computer?(Linux)

I’m asking this because USB 2.0 is capable of 480 Mbit/s (60 MB/s) per controller.

So to capture a frame 320x240@120 hz it would take at a rate of 480Mbit/second is:

[ (320x240= 76800 pixels) * 3bytes per pixel(RGB) * 1000 millisecond(1 second) / (480Mbit = 60,000,000 byte/second) ] = 3.84 millisecond

Is this calculation right? But then on Cogville’s web page at the bottom:

http://sszymczy.rootnode.net/index.php?menu=projects&submenu=webcamsync2&subsub=PS3Eye

he has a image showing that it takes 1 millisecond to zero millisecond to capture each frame?

How’s that possible? Why’s my calculation above is wrong? Why my calculation doesn’t add up?

Can anyone kindly point out why I’m wrong in this case?

On the image (Synchronized frame sequence) there are the following times: 36.11, 36.18, 36.25, 36.31. Difference between consecutive values is 0.06 or 0.07 seconds, so it’s 60-70 milliseconds. This is consistent with the fact that the camera was capturing 15 frames per second, so there is 1/15 = 0.06(6) seconds of difference between the frames (about 66 milliseconds). Where did you get that 0-1 ms?

Anyway, your calculation is right, but it’s the time needed to transfer one frame from camera to computer with the ideal USB 2.0 controller, not the time to capture the frame. As for the 480 Mbit/s for USB 2.0 - it’s a theoretical value, existing USB 2.0 controllers are much slower. My computer (after some minor Linux kernel tweaking) was able to capture two 15 fps 640x480 PS3 Eye streams simultaneously, but it was pretty much the limit. I guess it’s possible to overcome this limit by using multiple USB 2.0 controllers.

sszymczy - 04 May 2011 01:18 PM

On the image (Synchronized frame sequence) there are the following times: 36.11, 36.18, 36.25, 36.31. Difference between consecutive values is 0.06 or 0.07 seconds, so it’s 60-70 milliseconds. This is consistent with the fact that the camera was capturing 15 frames per second, so there is 1/15 = 0.06(6) seconds of difference between the frames (about 66 milliseconds). Where did you get that 0-1 ms?

Anyway, your calculation is right, but it’s the time needed to transfer one frame from camera to computer with the ideal USB 2.0 controller, not the time to capture the frame. As for the 480 Mbit/s for USB 2.0 - it’s a theoretical value, existing USB 2.0 controllers are much slower. My computer (after some minor Linux kernel tweaking) was able to capture two 15 fps 640x480 PS3 Eye streams simultaneously, but it was pretty much the limit. I guess it’s possible to overcome this limit by using multiple USB 2.0 controllers.

Thanks sszymczy for replying. So it was the “capture time” not the time to transfer data from camera to computer. I get it.

Sorry i miscalculated. I made a mistake. I thought the synced frames are being transferred at 0-1ms.

Again thanks for clearing that up.