I've previously described the arm-shoulder area as fragile, but what do I mean by that? This is what I mean:

In the center, you can see a back side view of the low poly base model. The colors are the weights of the shoulder (bottom-center) and arm (top-center) joints per vertex. On the left-center is the vertex arrangement I selected for this cycle's images and what will be used going forward. On the right-center, an alternate arrangement. That red dot in each of the center images? That is the only vertex that changes between the two sides. Can you see the difference? I know it can be difficult, but the slight differences in the weights should be a clue.

Blue wireframes are the chosen vertex arrangement, green wireframes (except the center because I wasn't thinking ahead :P ) are the alternate vertex arrangement. Due to laziness and not wanting to make a new frame, the bottom two smoothed wire frames have their low poly bases arranged vertically in the center-top and center-bottom with their respective colors.

The most noticible difference is in the center far left and center far right smooth wire frames. That tiny shift in the placement of the vertex had the effect of altering its weight and position sufficiently that when it deforms, it has that much affect on the overall shape of the shoulder.


That is what I mean by fragile.


Too few edgeloops? Yes. But through all the experimentation I've done, this is still the best and most flexible shoulder-arm connection I've produced. I tried a lot in order to introduce new edgeloops for tests with dual quaternion skinning method. I simply don't have the skills to make them look as good but be more robust. It would probably at least require adjusting the skin weights a lot, which piles a lot of additional complexity into my tests. Frankly, I think I prefer the fragile, yet fairly well deforming, arm with zero skin weight painting and a touch of simple weight blend painting.