CompuServe Thread

#IK — Help please!

9 messages in this thread
#170011From: Dennis FenskeMay 14, 1995 9:32 AM
I am attempting to get an IK motion set up an am having trouble. I originally set up an overly complicated (but real) situation and became totally confused. I have read and re-read the manual a number of times. I have worked the tutorial on the "person" and "Randy", but somehow cannot grasp what is "not correct" about my setup. A simplified version of what I need is like the human arm, elbow, and hand. See below. C-D is the hand. It only moves left and right. No up or down motion or rotation at all. There is a pivot point at A, B, and C. So as the hand moves left and right, the arm AB rotates around a fixed point "A". Arm BC pivots at movable pivot "B". And the hand moves left and right with attachment at "C". B /\ / \ / \ / \ A / C \——– D <–> only moves in this direction. It is like having your hand directly in front of you and facing away from you. Your elbow is out to your side. As your hand"CD" moves away from your body, your elbow"B" moves at half the distance of your hand in the hand direction, and also moves in toward the center of your body as it moves. The upper arm is attached to the body, and only pivots"at A". I linked AB to BC to CD. Parent is AB, child is CD. I defined the chain from "AB" to "BC" to "CD" with "End to Start". I didn't set the joint precedence because I believed it to be all the same. I set the joint parameters as rotation around "A". Rotation around "B", and sliding for CD. I set up a follow object that only moves as CD would move. When I play the interactive animation, only the hand moves. AB, and BC do not move. They just sit there. The joint parameters have them set to rotate around the points they should. Is this understandable? Does anyone have any suggestions? Sometimes I feel so dumb. This is one of the times. Thanks for your help, Dennis
#170019From: Jonas Ruikis [ADESK]May 14, 1995 10:45 AM
Hi Dennis, << Is this understandable? >> well….:-) << Your elbow is out to your side. As your hand"CD" moves away from your body, your elbow"B" moves at half the distance of your hand in the hand direction, and also moves in toward the center of your body as it moves. >> This movement seems similar to me (in the IK sense), to bringing your knees up to your chest while jumping in the air and all of the time the toes stay pointed towards the floor. Please review the following file for some possibly additional insights to several different methods of IK approaches.. File Info for FLIP.ZIP Section: Mesh/Project Files Contributor: 73172,1351 Size: 95951 Submitted: 10/21/94 Type: Binary Accesses: 207 Title: IK Backflip Keys: IK FLIP R4 This is an example of using IK a number of different ways. A woman doing a 1 1/2 back flip off of a diving board. There was no keyframing of the board or the woman on my part. All movement solved for by using IK. This zip includes a solved version and an unsolved version so that the R3 users can compare the two files. Note also the flip.zip (a flip.fli) in the Animations section of the lib. IK'd and uploaded by jonas. [dl time at 2400 baud: 7 min]
#170026From: Dennis FenskeMay 14, 1995 12:03 PM
Jonas, I will download it an review it. Yes the legs work just as well as an elbow. Dennis
#170040From: Dennis FenskeMay 14, 1995 5:10 PM
Jonas; >> Please review the following file for some possibly additional insights >> to several different methods of IK approaches.. >> File Info for FLIP.ZIP Thanks for the tip Jonas! I looked at this file, and at first couldn't figure out what was wrong. I had my Heir* links set up correctly. I had a path just as you did. But when I did a "Recalculate" on yours, it did not have an (ill formed chain) like mine did. So I experimented with the chain. I had a dummy possibly lower in the heir* than it should be, so I deleted mine, and made a new one. I also experimented with "Joints" and "solve for orientation" more, and between these items, was able to get it to work. I didn't really know if "IK" could allow me to do what I needed, but this is EXACTLY what I needed for the animation I am doing now. I have four (3 object each) arms that all need to move together. This will really make my life easier. Thanks so much! Dennis
#170078From: Jonas Ruikis [ADESK]May 14, 1995 10:50 PM
Hi Dennis, << Thanks so much! >> My pleasure. 🙂
#170025From: Dennis FenskeMay 14, 1995 12:03 PM
Additional information. I am not sure what recalculate means, but when I click on it, it says, "Chain is ill formed". My animation in the Key framer does exactly what it does in the interactive viewer. I also shortened the chain to AB and BC and used a follow object. The "hand" moved in the direcion it should at twice the speed of the follow object. The arm "BC" moved in sync with the follow object, but did not rotate. And "AB" didn't move. I also added more parts thinking it needed more to link through. Still only the hand moves. All help appreciated. I can upload a file (25K) of what I have if someone wants to look at it. Thanks, Dennis
#170031From: Paul Sanford [LVL5]May 14, 1995 2:09 PM
>> The joint parameters have them set to rotate around the points they should. << Did you select the X, Y, and/or Z axes for each joint that you want IK to solve for positions with? You know, the buttons along the left side of the IK joint parameters module that turn red when on? If they aren't selected, IK doesn't consider them. It's easy to overlook this step. Paul S.
#170039From: Dennis FenskeMay 14, 1995 5:09 PM
Paul; >> Did you select the X, Y, and/or Z axes for each joint that you want IK >> to solve for positions with? I have the Y axis selected because that is the only one that it rotates about, and the X axis for sliding for the "hand". However, since it was not going to move in any axis other than Y, I had NOT selected "solve for positions". Maybe that is the problem. I wrote the above and thought I would experiment before actually sending this, and with your and Jonas' help, I was able to determine that I had two problems. One was incorrectly set joint parameters. I had the hand sliding instead of rotating. The other was I was not "solving for orientation" on the hand. So now my simplified model works. I can now move on to the real thing. Thanks so much for your inputs! Dennis
#170068From: Yost GroupMay 14, 1995 9:06 PM
I can't add much to what Jonas said. (Thanks Jonas.) – G