while I'm building the character I keep it as low resolution as possible. when I get to a point that I cant keep it low anymore,
I lay out the uvs -cut merge etc. ( its much easier while it's low)
Then I make a copy of the char and smooth it and go to the next level of detail. That will be the one that I smooth bind and render.
then I cut up the low res and PARENT the pieces to the joints.
I put the high rez on a layer and hide it. I use the low rez parented version just as a stand in while I animate. Since the computer does'nt have to calculate any skin/muscle deformations
I am able to scrub and playback real time and get most of the
animation. then I hide the low rez and do some playblasts with the high rez and do any tweakin I need.
The low and high rez are not really connected. except that they are on the same skeleton rig. Some times I can gain speed by setting keys at the bind pose and detaching the the high rez skin (w/keep hist on) before I animate then reatach later.
Any cloth, jiggle, SDK muscles goes on the high rez since thats what renders and it would only slow down playback of the stand in.
Avoid changing after you bind as much as possible because you get extra tweak nodes that have to calculate and you cant delete history with out losing your skin weights.
That's how I do it and it works good for me.