I can force text into the proper position using actual numeric values for X and Y in -annotate +X+Y, but I want to be able to compute the X and Y values (in my program) relative to the image size. I am using -gravity Northeast and NorthWest for both vertical and horizontal images of varying sizes. I want to be able to have the text inset a computed number of pixels from the edges of the image RELATIVE to the image size. In my testing, it seems that when using SouthEast, the X value I insert manually is relative to the right-hand edge of the image, which is counter-intuitive. Am I right?