Have you compared yours to what cavac published as Re^3: Shameless plug and QR japh in April, in the second code block? Since GD::Image can handle multiple image types, cavac's has more input options. But looking at yours, I see that it might have color reduction/compression (like changing the default color on the table to the most prevalent color in the image). I also see "canvas" mentioned in your code, but I'm not sure if that's just there because it isn't using GD::Image, or whether it was doing something unique with a real HTML5 <canvas> , or whether you just borrowed that term from HTML5 (since Perl Monks Approved HTML tags doesn't list <canvas> among our assets).