I've tried the selective compression feature for jpg in Photoshop 6 (well, in ImageReady to be precise). But I can't get it to work as advertised. I get more artifacts and a bigger file size than when use the built-in compression algorithms.
The Photoshop 6 Classroom in a book series has three full chapters on image ready & is well worth reading if you want to get to grips with this package. I now use IR all the time & consider it to be a more sophisticated tool than fireworks, though in reality they both do the same things.
I used to use Ulead all the time. It served me very well.
But the algos in Image Ready are now rather awesome. Common scenario -- a layer of type over a photo. You know how sharp edged type attracts jpg artifacts, but the limitations of gif for photos are also rather daunting.
My "premier" solution has been to use absolute position and the z-index -- stacking a gif with transparency (for the type) over a jpg (for the photo). But you should see what a nice job IR does in preserving those sharp edges even with relatively high compression. In all but the most important spots, I'm now a convert to IR and one image.
I still don't use IR for gif compression. I've learned too much about gifs to yield control to any automation. But the jpeg format is mysterious to me, and IR handles it very well. It's just that I get better results letting the algo do its thing than when I try to protect sharp edges from artifacts using the selective compression mask feature.
How much does IR's compression vary from PS6's Save for Web command? I got hooked on Save for Web with 5.5 (and a home computer that can't handle opening ImageReady while PS is running)... Didn't occur to me that IR might actually be using different algos for its compression routines.
Ahh... I thought maybe you needed to actually click the "go to ImageReady" button, and use IR's somewhat more confusing interface. :) It would seem exceptionally silly for the two programs to have different compression algos, though.