16th June 2011
Sometimes the goal is to shape the envelope of a sound to give it more snap and punch (for example kick and snare), so a slow attack and fast release might be a good place to start. If a sound is too dynamic (quiet parts too quiet and loud parts too loud) then a faster attack and slower release can smooth things out. It's all about experimenting.
As an exercise you might try bringing the compressor's threshold down too far and then play with attack and release times to really get your head around how different settings shape the sound.
Sometimes it's not about reducing dynamic range so much as just wanting some of the character that a particular compressor imparts. In those cases I'll attenuate little if at all.
When mixing delicate music I tend to compress less and volume automate more to maintain the purity of things. For more rocking material, part of the glee for me is in squishing the heck out of stuff!
Parallel processing is definitely worth getting familiar with. It allows you to leave the original sound in its natural state while feathering in sometimes extreme amounts of compression, EQ, saturation, etc, to taste. This is a way to get things sounding huge without sounding squashed, when that's the goal.
For saturation I love Soundtoys Decapitator. It's really versatile, and small amounts of it all over a mix can add up in a very cool way.
Sorry for rambling...I hope some of this makes sense!