Has anyone messed around with DOF in jogl? I want to blur flat primitives based on Z-depth, or figure out an efficient way to simulate DOF. So far my best guess is to jitter each object's coords and decrease alpha as the Z-depth increases.
so (pseudocode):

for objects:visible
  for z
      jitter[] = jitter[z]
           for jitter
               fill(alpha/z)
               visible.render(x+jitter[0],y+jitter[2],z)


jitter(depth):
return jitter[math.abs(depth)]
where jitter arrays are [n][x,y] of floats

or something like that.
Is there a faster/better way in openGL? I think I could get it to look OK by tweaking the jitter arrays and it wouldnt be too slow.
THanks,
Brian

Well, it works well enough. I have one big static array of jitter points and scale the number of points used by z-depth.
Its still not as good as box blur or guassian, but with more tweaking it might get close.