Personally I strive to get the vocal sound most appropriate for the song BEFORE tracking. I will line up 5-6 mics that I think are good for the singer or song and have them run a few test verses/choruses on each one and then have them come back and listen to them (not knowing which take/mic is which) and have them choose the mic they like best.
I did it last weekend with an RE20, SM7, NT1, Studio Projects B1, SM58 and CAD (I forget which model--I'm not a huge fan of CAD mics).
Anyway, the singer sounded *terrible* through the SM7 and SM58. The CAD was okay but too many overtones, the NT1 was nice, rich and clear, and the RE20 was really fat and warm. However, the mic that was the close winner (versus the NT1 and RE20) was the B1!!! I was shocked because it was almost the cheapest mic there. The B1 captured the warmth of the RE20 but had the richness and air of the NT1 (it was sort of between them in a way).
So we cut most of the stuff on the B1, and did one song on the RE20 to take advantage of the warm tones. We did some screamy overdubs on a SM57 to get that crushed lofi sound.
And then we experimented with some preamp and compressor settings, and a little EQ until we had what we wanted--THEN we did the takes. On his cans (not committed to tape) we ran a little 'verb to smooth things out so it would sound very finished to him in the room.
Moral of story: If the singer can hear close to the finished product in their cans they *WILL* sing better.
Other moral: you can never be sure how a singer will sound on a mic until you try them out. I know some guys that sound amazing on the SM7, but crappy on an NT1--in this case the singer sounded GODAWFUL on the SM7 but huge on the NT1. Always, always, always audition and tape mics and analyze them with the singer to capture what the song needs.