I do some of the same type of stuff, and still I haven't found any way to do what you're asking in one pass. Basically, what I do is record the quiet bits on one track and the loud parts on another. If your tracks are too limited for this, punching in is probably your best option. Of course, the distorted screaming vocal sound is very cool if done correctly....
As far as making the screams smooth, first I cut out everything under 120Hz, then I boost the highs above 8000kHz by about 4dBs. Tracking fairly hot to tape, or using a tape simulator also tends to warm up vocals. I usually use a whole bunch of compression; a 4:1 ratio, quick attack, quick release, and a threshold setting that takes off anywhere from 6-11dBs. Actually, I compress the soft parts too to get more of a breathy sound, but that's just my preference.
Remember, these are the settings that sound good with my voice and my equipment (a Rode NT1 through an ART Tube MP into Cakewalk); you may hate the way they sound with your voice and setup, but hopefully this'll give you something to work with.
Good luck!
Ryan
