presumably if I do store the image in my code and binary but succeed in obfuscating the fact that I did, then I win some chocolate? The hand drawn image contains very little geometric repetition in which you can really push the boat out in terms of generation, any program that does this would be pretty much be scan-line based with fixed or obfuscated data describing filled blocks etc or otherwise break the image down into the smallest procedurally replicable parts.

I'd be up for an obfuscation challenge any day or night but if we're going to obfuscate the process of re-constructing an image then it'd be far nicer if that image had interesting structures to replicate.

Plus an obfuscation challenge is very subjective, what may be indecipherable to one pascal coder might be perfectly fine for the next. Unfortunately the language is a very poor choice for obfuscation as it's strongly typed by design. A lot of the fun of a C obfuscation challenge comes from throwing off the reader with obscure syntax or by deliberately exploiting the non-type safe design blunder of such languages.

I think it'd be a challenge just to write any obfuscated pascal code, at least in the classical, small code snippet sense. Obfuscation through sheer size (we've all done that ) or macros used to redefine language keywords etc are boring.