Pole target rotating feet

Hello, I am having a little trouble with a leg IK chain. When I create a pole target to guide the knee’s, the feet rotate 90 degree’s inward. I assume there is a simple fix for this, I’ve just been looking at it too long. Help would be much appreciated. Here’s the .blend file of the rig:

There is an “offset” option in the settings which is usually equalised at either 90 or -90.

Indeed, the offset fixes the rotation of the feet when set to 90, but then the knee’s bow out to the sides instead of bending normally. The knee’s bend correctly when offset is set to the default 180.

Okay - had a look at your rig (dial and rapidshare woes make me dread such downloads).

Nice rig (hope you don’t mind me learning something from playing with it).

Unparent the foot from the shin bones. Have its “connected” setting to “off” and then parent it to LegR_IK (or LegL_IK) Efffectively, the foot is a separate object that the shin just likes attaching to as with a magnet.

Thanks for the help. That solves my problem as long as I keep the legs controlled via IK. The character I am making the rig for plays the role of an extra in a short I’m making, so that should work out well.
Though I was thinking of using the same rig for some of my other characters and adding an FK/IK switch. If anyone can come up with a solution that would keep the foot attached to the shin, that would be very helpful.

Feel free to learn whatever you can from my rig. It’s not finished yet, but I’m pretty happy with it so far.

I haven’t gone as much into rigging in Blender as I would like to, though I think for FK and IK leg rigs, you would use two sets of bones; one for each. I’m not sure how the standard Blender FK/IK switch works though, could be wrong.

You could try making he rig bow-legged (feet point outwards) so the offset actually corrects the feet, though I’m not sure how many issues this could give you when trying to marry the rig to your mesh.

sorry to say, but the rig isn’t that clean, you have a number of cyclic dependancies and it’s controls leave something to be desired (and I don’t mean it needs visual widgets, I just means the controls seem a bit inconsistent)
As for the pole issue, that’s easily fixed.
Set the pole offset for both bones to -90, and make sure you have the !actual knees! point forward a little.
Your IK chain is currently a straight line, which confuses the IK as to how it should bend in the first place. Point the knees forward a little and the IK will work flawlessly, real legs usually have a slight bend as well.

You’ve got the main idea right, but I think you’re overcomplicating your own rig, which causes inconsistencies, and cyclic dependancies. It has more constraints and is more complex than it needs to be to do the exact same thing it does now, or would do better.

For IK/FK switching you don’t need to use what I like to call “limb-sets” You could use one arm with an IK constraint, and do the switching by driving that IK constraint. But I’ve learned from Nathan (from BBB fame) that using a second limb set for FK actually makes the transition between FK-IK smoother. And boy, he was right.
It’s perfectly possible to use one limb set, and it will work (little_fellaX uses this) but a seperate set for the FK will make the interpolations prettier.

If you want to keep the foot attached to the shin look for the footroll tutorial from calvin.
Or look at rigs like Cessen’s biped rig or little_fellaX (those are the first I can think off)

A quick method would be to have the foot copy the location of the shin and set the head/tail slider to 1. then have the foot also copy the rotation of the leg’s IK target.

Thanks for the help, but I am a little confused.
“Set the pole offset for both bones to -90, and make sure you have the !actual knees! point forward a little.”
I’m not sure what you mean by that. I tried rotating the bones in edit mode so the knee’s are bent forward. However, when the offsets are set to -90 & 90, the problem remains the same. Is that not what you meant?

I think I will use the copy location/rotation constraints as you suggested. That method will provably meet my needs.

If you have any specific suggestions for how to improve the rig, that would be very helpful.
Thanks for the assistance guys!

from what I can tell, that’s almost the same thing I’m doing. But I’m just grabbing the knees and pull them forward, instead of rotating the bones. I don’t know what it is you are doing different that could cause this. rotating on a set axis shouldn’t affect the roll angle either.
Sorry but I don’t know.