RGB int values to OpenGL or Jitter matrix

    May 05 2011 | 5:42 pm
    I am using Java to read video into Jitter, but Java represents pixels as a single integer for the ARGB components, like this:
    -8360914 -7374019 -6785989 -6785989 -6589893
    which I don't know how to use in Jitter.
    I am looking for a fast way to either translate this data into 4 separate ARGB numbers per pixel or even better, send them right to a Jitter OpenGL object for subsequent use with other Jitter objects.
    Any ideas for a computationally fast way to do this?

    • May 06 2011 | 4:17 pm
      A quick search (hint - Google: Java ARGB) reveals an algorithm for deriving 8 bit ARGB values from a 32 bit int as:
      int alpha = (pixel >> 24) & 0xff;
      int red = (pixel >> 16) & 0xff;
      int green = (pixel >> 8) & 0xff;
      int blue = (pixel) & 0xff;
      Translated simply into max to illustrate the process and get the ball rolling...(but there are no doubt more elegant and efficient ways than this):
      No guarantees for either the code i stole, or may max implementation of it, but it may help you on the right track...
    • May 08 2011 | 9:35 pm
      Thank you. You are exactly right about needing to extract individual bytes.
      However I have to do this for relatively huge arrays (76800 length just for 320x240 video), so I think I need a faster way then doing the byte extraction on each element of the arrays.
      Do you, or anyone else know if it's possible to send RGB int arrays directly to an OpenGL texture? Or a fast way to byte extract these arrays.
    • May 09 2011 | 2:14 pm
      Not sure about sending RGB ints to OpenGL. However processing a matrix of that size or larger is fairly typical in Jitter and a matrix version of the algorithm I posted earlier might go something like the patch attached (say that five times fast!). I'm too lazy to actually encode an image to test the patch so the usual caveats apply...