"The spawning a separate thread to get images faster might not be a good idea."

I appreciated your input before, but you really couldn't be more wrong about that particular point. The implementation is so ideal it's ridiculous! :-D

To elaborate, the spawned thread doesn't get images faster, it just prevents page X from blocking the response until the image requests finish.

Granted a user can visit page Y before the spawned thread finishes, but the solution isn't to block a user visiting page X for 15 seconds while the images download, the solution is to block on page Y if the spawned thread hasn't downloaded the images previously (which is the approach I'm taking).

Anyway, I AM interested in the way you're saving the files. How long have you been using that convention? Have you noticed a set of duplicate images downloaded periodically? Again, my concern is that the implementation of the URL structure may change frequently resulting in waste, but if you've found it to be fairly stable, I'd be all over making a change.