Haven't really got the butt shape quite right...

Anyways, changes!
[+] Obviously, the Mesh topology has drastically changed since the last mesh review, which itself was an in-progress departure from a flow that's been in place since v2.
[+] The Waist now features a rather unusual crosshatched flow (composed of diamond-style quads.)
___[-] Frankly, this thing scares me. I spent a long time experimenting with every thing I could think that might go wrong with using it, including texturing and UV unwrapping. It succeeded but it still terrifies me. I fear I will find myself undoing it all one far off day in v4... but, until then, all signals are go!
___[-] As a side note, I was incredibly frustrated when I first started working on the waist. Every change I made seemed a complete failure. In an act of sarcasm, I told myself I would make it a prototype of a waist. It would be made so easy, even my incompetence couldn't mess it up. I expected it fail, of course. I took two long pyramids, inverted one and rotated it 45 degrees. I used a Boolean union and about 5 vertex merges to unify the shapes. The resulting shape was intriguing, so I cleaned up a few more edges and smoothed it. To my great ಠ_ಠ, the result was an incredibly versatile and flexible shape that worked with my existing skeleton to produce exactly the bending actions I wanted. It took some adapting, but the crosshatch shape of that prototype inspire the waist topology of the model.
___[-] There are a few minor problems, in that the transition between the vertical and diagonal edge flows requires a large number of 5-edge-single-vertex intersection (stars, I think they're sometimes called.) These can produce interesting bending issues, but I've worked hard to minimize those issues. I think once it's textured, they'll hardly be noticeable.
[+] Part of adapting the crosshatch diamonds to the model was dealing with technique I use to mirror the model for symmetry. Since using it on the edges would result in gaps, triangles, or pentagons (you can't have a missing edge on the edge of the model for mirroring!) I opted to include a strip on the front and back that follows conventional topology. These strips worked beautifully (with adjustments) as a spine and abs. The shape is built right in; I love how it works!
[+] It doesn't show to well here, in the unsmoothed version, but the diamonds result in a surprisingly smooth curve for the waist. It can be automatically regenerated through averaging the crosshatch verticies, after adjusting the top of the pelvis and the base of the rib cage, as well as the spine strip and abs strip.
___[-] Everything about the shape happens so naturally and automatic, with this new crosshatch; its amazes me.
[+] The base of the rib cage has been adjusted and aligned.
[+] The curved top of the pelvis has been flattened quite a bit and the little strip's bulging effect has been exaggerated. This helps it read better, but it also makes it deform much better and reduces the amount of skin weight painting I need to do, a bit.

Other Changes:
[+] Torso is has been fully converted to quads only! Cleaner topology and edgeloops, all around.
___[-] Well, depending on what you consider clean. The crosshatch thing seriously messes with programs' ability to identify edgeloops and polygon flow. Another reason it kinda spooks me.
[+] Reshaped the butt. It still needs work. The crack, in particular is awkward.
[+] Rewired the pelvis and butt's connections to the legs so that they bend more properly and cleaner. Still needs a bit of work.
[+] Greatly reduced the number of polygons through most of the system. An excess of polygons, contributing little to the overall shape, seems to be the source of many of the problems during posing.
___[-] This model still requires a great deal of skin weighting and post-skin-binding movement of joints to get it to bend properly. I had hoped to eliminate the need for this, but alas I have barely even reduced the need. :(


So much left to do. Many other joints in the limbs require similar adjustments to the mesh and edge loops in order to deform properly with the new smooth skin bind parameters. Oh, didn't I mention that a while back? Yeah, I stumbled upon a set of parameters, while puttering about, that solves about 90% of the issues I have with skeletal deformation.