Hi,

I know that the native chip resolution for the ps3 is vga and it has a bayer array -however for our application we want to use the very high frame-rate qvga resolution. Ideally we would like to loose as little visual information as possible and have minimum artefacts. So my question is: When the ps3 cam subsamples from the bayer array to make the lower resolution bayer array, it needs to make one pixel from where there were 4 before (two green, a red and a blue) - one way I would imagine this is done would be by just picking the relevant colour for the new pixel and throw away the rest, (in case of green it would be great if the two values were averaged to give the new green). But I read that some cameras use a much less accurate way and simply replace a 4x4 patch with the top-left 2x2 patch in the lower res - which would be really bad as visual information is displaced and you might get aliasing etc.

Does anyone know what the ps3 eye does for its low-res image ?

thanks,

Stefan

The image sub-sampling is done on the CCD itself through a process called 2x2 binning.
I am not sure exactly what process is used on the OV7720. The simplest way for you to verify this is to capture the same image with 640x480 in bayer raw format and then capture it with 320x240. By comparing these data you will be able to deduce the algorithm that the Omnivision is using.
Another thing you might do, since you are concerned about the quality, is to grab images in 640x480 in bayer raw format and perform high quality sub-sampling yourself.

AlexP