There have been a few threads recently on IK/FK spine rigging which have been really enlightening
( here.) I figured i’d throw a line out there to see if anyone has any cool setups for the shoulder and hip area.
I’m particularly interested in setups which solve the problems below. I know that most of these can be fudged using corrective morphs or totally steamrolled using complex muscle systems, but are there any more lightweight, elegant solutions to the problems which can be done with bones?
Shoulders: When you bring the arms straight forward in front of the body, the deformation tends to go all crunched up and horrible, and part of the arm tends to intersect the body.
Hip: When flexing the thigh to extreme angles (bringing the knees up to the chest), the deformation can get really nasty. I guess one solution might be to have several bones in a chain at the hip area, to create a sort of ‘U’ bend, but this would be a real pain to control with FK, and IK might behave in completely unexpected ways.
Image: This shows the cheap solution I use for the shoulder (I’ve also seen it used in some game models). By splitting the shoulder into two bones, you can create a curve at this area. The downside is that it’s not realistic, and there is no way to automate it that I know of.
Looking forward to hearing your ideas. This place rocks!
If it’s not for games, you can use a few stretchy bones to quickly fake muscle groups of the shoulder girdle. The setup isn’t at all hard or complicated, and you get the bonus of brushing up on your anatomy knowledge. It takes a bit of experimenting to determine exact placements and skin weights, though. Take care not to place the bones in such a way that they would shorten nearly to zero on some movements.
In my experience, for the shoulder its enough to make the following:
Pectoral. One long stretchy bone starting at the center of the rib cage, inserting about 1/3rd 1/4th down the humerus, front side.
Deltoid. Usually two stretchy bones, one for the front group (from middle of the collar bone to about 1/4th down the humerus) and one for top group (from a bit above the distal end of the collar bone to 1/3rd down the humerus).
Teres group. One stretchy bone from where the scapula would be to about 1/3rd down the humerus, back side.
Trapezius. One long stretchy bone from the top of the scapula to the neck at the back of the head.
(The front/back side means the insertion should be a bit offset from the central axis to that side, also linked to the nearest twist bone to make it look like they work when the arm rotates on the longitudinal axis.)
If you’re only looking to fix the crease that happens front of the shoulder on flexing the arm, 1 and 2 is enough.
In my experience, this approach works better than corrective morphs for cases when you make an adjustable scapula (usually only needed on relatively naked characters), and is overall more versatile in terms of angles, yet it offers significantly less control over the exact shape.
Similarly, a stretchy bone from the top front part of the pelvis (around anterior superior iliac spine to be exact) to about 1/3rd or 1/2 down femur helps take care of the crease at the front of the hip when flexing the leg.
(The “stretchy bone” I mean is the 3dsmax kind that squashes when compressed and becomes thinner on stretch, rather than only scale along length axis. Though in the case of hip, the “scale” one is often preferred as it’s easier to control and there’s no need for “bulging” there.)
Perhaps someone could offer a solution better suited for games, though…
I’ve started experimenting with the setup you posted Zhalktis. You’re definitely right about the weight painting… it’s bloody difficult!
So far I’ve been having some real issues getting it to work with the arms held out straight forward. It works really well for other positions though. I’ve attached an image of my bone placement for the first few bones. I’m pretty sure I got something wrong!
Yeah, it’s a little tricky the first few times.
(I’ve found it also teaches well about what effects you can achieve with just the basic tools: positioning, angles, lengths and weighting of the bones. To a greater depth than skinning ordinary bones, at least.)
I’d suggest some minor changes:
lengthen the top deltoid bone, just extend the end of it in the same direction up to the skin at the top of the shoulder. This is to avoid glitches when lifting the arm more than usual - extreme lengthening can create extreme bulging which quickly becomes odd even with little weighting.
it’s probably the positioning of the pectoral bone that is giving you problems with weighting for front extension; it turns inwards and upwards when you turn the arm into position, ideally it should be mostly being squashed in the front extension position and only turn when the arm is lifted up. Try putting the insertion of the pectoral at the center of the arm bone (not offset) and a bit lower, (roughly where you put the front deltoid bone’s origin), and make the origin of the pectoral about the center of the chest (vertically) so that it was more or less horizontal in this base position, you can also experiment with setting it a little deeper in the chest. Usually mostly the front of the armpit (shared with front deltoid) is weighted to it, the actual pectoral area gets less weight.
Thanks for the advice again! I’ve had to put this little project on the side for a while to focus on uni work… can’t wait to get back to it.
I found another technique online which is quite useful for sorting the shoulder deformation out with the forward angle. I think it works exactly the same in principle as your technique, but uses reactions instead of lookat constraints. Essentially, you just make a little ‘fin’ bone in the pectoral area (using the weighting you mentioned earlier), and wire it so that it goes forward when the arm/shoulder does, but not when it goes back. The advantage of this is that it can give the illusion that the pectoral is wrapping round the ribcage when the arm is pulled back. It’s probably less accurate than the lookat method for other angles however.
I’ve found a free rig online which features a full stretchy bone solution for the shoulder girdle. It’s by a Korean artist known as Sox. It has a really high range of motion, and it uses a really simple but effective method to drive the twist bones on the shoulder with almost no flipping issues. It’s actually very similar to what Zhalktis posted earlier, but the bones are split into two to allow more curvature. The method for powering the shoulder twist is impressive and simple with almost no flipping issues.
I’ve done some tests and turned the bone stretching off, and it still seems to be an effective solution for the shoulder deformation (meaning it can be used in game engines with no bone scaling allowed).
Kinda missed the boat on this one, but a couple times I had decent results simulating the pectoral area by using a ribbon spine run from the chest to upper arm (basically where the pectoral would connect) and just having it bulge out when contracting. But it was a pretty quick and easy setup. that didn’t involve any joint scale.
