Setting Up Wrist Twist/

[mike-milo]

Can someone point me to some good tutorials or tell me some methods for creating good wrist deformation-specifically twisting the wrist and making that look good??

[joojaa]

first you must understand taht the wrist does not rotate! the rotation actualy starts at your elbow.

YOu cn either constrain two bones to conform to the way a real human hand works OR a equaly accepted idea is to place a joint midway between the elbow and the wrist then run the rotations up from there thisway whan you bind the bone the rotation seems to gradualy grow towards the wrist giving the impression of the same thing as the real structure.

[onscreen]

i set up 2 ik, arm to wrist and wrist to palm.
then I addAttr on the wrist control called wristTwist and connect it to the twist of the palm IK.

to get the twist going up to the elbow - I never liked binding the radius/ulna approach and the adding joints and into the chain method gave me problems with my IK set up when working ny reference. (it involved moving the pivot of the effector...)

so i just add two joints coming off the elbow one third and two thirds of teh way down the forearm. they are not connected to anything else. I then use an expression to set their twist to the twist of the elbow(0) plus the wristTwist divided by .3 or .6 respectively.

note- when I bind, I don't bind the palm joint. its just messy.
also this works best if your charcter is modeled in the arms straight out to the sides pose.

cheers.

[mike-milo]

Thanks fellas!. I'll have to try this... so much to think about for an artist!

QUOTE(OnScreen @ 09/27/05, 07:23 AM)
i set up 2 ik, arm to wrist and wrist to palm.
then I addAttr on the wrist control called wristTwist and connect it to the twist of the palm IK.

to get the twist going up to the elbow - I never liked binding the radius/ulna approach and the adding joints and into the chain method gave me problems with my IK set up when working ny reference. (it involved moving the pivot of the effector...)

so i just add two joints coming off the elbow one third and two thirds of teh way down the forearm. they are not connected to anything else. I then use an expression to set their twist to the twist of the elbow(0) plus the wristTwist divided by .3 or .6 respectively.

note- when I bind, I don't bind the palm joint. its just messy.
also this works best if your charcter is modeled in the arms straight out to the sides pose.

cheers.
[snapback]215748[/snapback]

[thraddash]

both replies are wrong.

to add twist joints and to control their rotation through reading the wrist joint's rotation attributes is just not good - they will flip all the time. make a test rotate the wrist down more than 90 degrees and then start twisting it.

the IK twist attribute thing is not good too, in most cases we are animating the arms in FK mode.

i think the only way to make the twist setup stable is to read the vector angles between elbow and wrist joints. in this way you are going to be independent by the euler angles and the gimbal lock thing ( remember, you are going to lock 2 rotation axes of the twist joints - in this case all the constraints, expressions, attribConnections just break.

Take a look at the Maya Help for some angular functions. Expression or simple C compilation doing the vectorAngles math will pay worth.

[joojaa]

i beg to differ. Theres absolutely nothing taht can lock or flip in athe extra bone setup since theres only one rotation thats takeing plce on that bone so theres no possibility of this here. The last joint is not part of th ik ionb anyways so you couldnt use iks twist eiter just straight kaying of its rotation. And theres no need to lock anything.

And no theres no use for the iks twist attribute.

Now the only problematic joint is the shoulder joint wich has the possibility of gimbal lock on the evry extreme poses of very extreme characters. but theres not much you can do about that.

And the iks ofset is not all that problematoic in my experience atleast i havent noticed any problems. Especialy if i do it with reverse foot typr of thing! tough thats not neccesery. Gnomon used to have a tutorial on the simple way on the net.

Offcourse the controll wont gimbal either if the rotation order is correct becasue it only rotates in 2 dimesnions!

[sheriffcicada]

QUOTE(Joojaa @ 09/29/05, 09:10 PM)
i beg to differ. Theres absolutely nothing taht can lock or flip in athe extra bone setup since theres only one rotation thats takeing plce on that bone so theres no possibility of this here. The last joint is not part of th ik ionb anyways so you couldnt use iks twist eiter just straight kaying of its rotation. And theres no need to lock anything.

And no theres no use for the iks twist attribute.

Now the only problematic joint is the shoulder joint wich has the possibility of gimbal lock on the evry extreme poses of very extreme characters.  but theres not much you can do about that.

And the iks ofset is not all that problematoic in my experience atleast i havent noticed any problems. Especialy if i do it with reverse foot typr of thing! tough thats not neccesery. Gnomon used to have a tutorial on the simple way on the net.

Offcourse the controll wont gimbal either if the rotation order is correct becasue it only rotates in 2 dimesnions!
[snapback]215995[/snapback]

I use a similair setup with several bones for both the upper and lower arms. I find that it is difficult to know exaclty which rotation order to use. Any advice on decifering that correctly. I use an expression with multipliers and a compensation rotating object just in case, but I always rin into flipping. Generally, Iuse Z as the axiz pointing down the bones and the x in the perpendicular horizontal plane then set
the rotation order to ZXY. Any ideas? I like the idea about vectors, this is the only completely fool proof method, as you end up losing a rotation somewhere along the way.

thanks for the thoughts

sheriff

[joojaa]

I dont understand how doing manual vector calcs as opposed to internal vector ones make a difference? Except thet the internals are 100 times faster. I mean thets exactly what mayas solevrs DO!

Can you post a unworking rig as MA filr (MA still means maya ascii NOT frigging MB wich everybody posts, ii mean ANYBODY can embed a number of very nasty commands in mb file, no problem and they get the trojan trough with the users priviligies, all you need is 3 lines of code to hicjak the users sytem, 2 to format it and 1 to crash maya settings)

[sheriffcicada]

tryin a new rig out... i need the twist attribute to work on a sc solver.... I know I've used it before, but somehow it doesn't seem to have any effect at all urg!@!!!!!

Thanks guys...

PNG

S