The author really is making a lot of assumptions about processing images in the floating-point space. I wonder if they’ll ever realize that their whites are never white any more.