Hello
I would like to capture the pixel values of a DVcam image in order to use these values in a patch. I'm not interested in displaying the image, so the frame rate is not very important. These values will be used for filling a matrix regularly (in the patch). As this matrix is quite small, the capture X-Y-resolution should be low (possibly the same as the size of the matrix) in order to save computation time.
I think it's possible with GEM, but it may slow the patch. What is the best method for doing this ?
Maybe I should use a webcam instead of the DVcam, too.
Again - what platform are you trying this under? Video capture is platform dependent.
Also, things like scaling will most likely not be done in hardware so decreasing the size of the incoming image will almost always also decrease performance! Take DV for example, it should always be 720x480/576 in every situation since the various capture APIs will do software resizing of the DV stream, and I can assure you that most of them don't do this in the most efficient way. If you want a smaller matrix from a DV image you should capture at DV resolution and use your own sampling algo code. You might want to just use a few pix_info objects ins series to sample various intervals of pixel values. I also seem to recall writing some sort of pixel averaging object for testing purposes - perhaps this might be useful for analysis like you want to do?
Another thing to remember is that nearly every video input/capture device natively uses YUV and _NOT_ RGB color-space. The only version of GEM that doesn't do slow, painful, frame-rate killing conversion is the one for OSX.
cgc
Thank you all,
Julien