yes, you’re right, that would be an interesting comparison too. We’ll think about it, thanks for your suggestion.

Zdenka

thanks for your tip. I’ve looked at the link you’ve sent, and you’re right, this piece of code is a sharpening function. But, it is not a build-in GD function, it is an implementation of some sharpening algorithm which uses GD functions. Having the basic image processing functions, you can implement a large variety of functions, and we’re sure there are plenty or them available online.
In this comparison we focused at build-in GD functions, that’s why we haven’t used any of these algorithms. GD as well as IMagick have the option of building your own algorithms, but that would lead us into programming the best sharpening algorithms. But this was not our intention.
Thanks a lot for digging that piece of code, it’s good to know that somebody already did that!

Your second tip is also very interesting, we will look at that.

however there is a sharpen function in GD – (I haven’t tried it yet)
http://vikjavev.no/computing/ump.php?id=306

I think you should create two files that are a similar file size – e.g if you leave the GD file of the girl at 32k – then convert using IM until you get a file of 32k … I think the quality will then be the same