Hi!
The reason this post isn't getting much attention is rather simple.
There isn't a right solution to this.
How you split up a mesh into proxy shapes depends on a lot of factors, topology of the mesh, amount of joints in the area of the body, mesh definition, etc etc etc.
That's why there is no "one size fits all" tutorial on the matter. It just depends...
Just do it in a way that feels right, the only criteria that matters is that the animatior understands what the rig is doing during animation. Because after animation the skinned rig is brought in anyway.
The second reason is, that this is no longer a practise that is really needed. It was needed between 2005-2012, when definition in static meshes waaaaay outpaced what graphics cards could handle during animation. But today most computers can handle body rigs, given that you don't have any muscle or other simluation in there, without a problem. My POS laptop from 2013, that I use to test my rigs on, can handle most of my complex rigs at 18FPS+ playback.
Today, what most riggers do in order keep animation fast is giving the animators two rigs, one with and on without the facial rig. Because facial rigs have become so complex that they take up 50+ % of the calculation time, so animators can animate the body first and then add in the face later when they are satisfied with the body movement, which is the workflow most animators prefere anyway.
I have one series of tutorials on splitting a mesh for rigging. But as I mentioned, it's old and not really common practise anymore:
https://www.youtube.com/watch?v=MV4XRgmTynY&list=PL8hZ6hQCGHMXKqaX9Og4Ow52jsU_Y5veH
I hope this helps!
