The behavior of most CICategoryReduction filters such as CIAreaAverage, CIAreaMaximum, etc, is to output a single pixel image that contains the calculated values. Using jit.gl.imageunit when I read back the texture using jit.matrix, I get an image that is the full size of the input image with the constant values for all pixels except for the single one in the lower lefthand corner. This makes sense based on Core Image coordinate space, but what I would like to do is still get a single pixel output. I have tried changing dims and sending messages to the texture internal to jit.gl.imageunit to try to get this to work but I seem to not understand something about how jit.gl.imageunit works. Has anyone had any success with this?