I recall someone once designed an image format which stored (image width - 1), and added 1 to the image dimensions while decoding. That eliminates an edge case,

I recall someone once designed an image format which stored (image width - 1), and added 1 to the image dimensions while decoding. That eliminates an edge case, but no clue what would happen with MAX_INT image size...