PDA

View Full Version : Astronaut on IK robotic arm: Shakes too much: ideas?



4dartist
02-07-2008, 06:41 PM
I am animating an astronaut out on the robotic arm. The robotic arm works like IK, when you control it you fly the end of the arm, so there isn't any swinging arc, just linear movement of the end. So naturally i want to use IK because it works the same way, so i rig this 50 foot arm with IK put astronaut on the end, and from far away it's flawless. Works just like i pictured and now i can move on.... BUT.. as i start to animate I notice when you move the arm via IK it jitters. just a few inches up and down, but totally enough to be noticeable.

Can I increase the accuracy of the IK somehow? I tried parenting all the equipment and astronaut to the IK null object so it doesnt inherit the shakes, but you can still see it way too much on the arm parts.

Any ideas?

Ben

SplineGod
02-07-2008, 08:09 PM
Whats the size of your scene (in meters) and do you have keep goal in reach selected?

Steamthrower
02-08-2008, 06:53 AM
Those jitters are small turbulence produced by orbiting the earth in the thin air in the upper atmosphere...turn "real world atmospheric" off and you should be fine...

Nangleator
02-08-2008, 08:08 AM
Those jitters are small turbulence produced by orbiting the earth in the thin air in the upper atmosphere...turn "real world atmospheric" off and you should be fine...
Nah, he probably sized up his null so much that the jitters come from quantum uncertainty or observer influence or something. He'd have to switch to "Classic Einstein" or "Basic Newton" dynamics.

Or not actually look at his animation at any point.

4dartist
02-08-2008, 08:21 AM
Around a 90mx50mx30m scene. I did a quick test with a new scene, and a few bones and set the IK up similar and it's very noticable. AHHhh... that sux.

With keep goal in reach it still shakes, but gives you less control.

I going to try reversing the Bone system and move the bottom of the IK chain instead, and just let the bottom shake and see if it's still noticable. Hopefully not as much.

4dartist
02-08-2008, 08:44 AM
Haha, ya, need to switch to that dynamic system where things just work and shaky math doesn't. Like outside my window.

adamredwoods
02-08-2008, 10:25 AM
Sometimes Lightwave's IK does shake when it does it's solving. Frustrates everyone.

Some tips:
Increase goal strength.
Decrease stiffness on one axis (pitch) and increase on another (heading).
Pre-bend the rest position of the bones.
Adjust the limits on the joints; sometimes -1.5 to -178.5 works better for 180degrees rotation limit (or smiliar).

Make sure BANK IK is off. Less axis that need IK, the easier it is for LW to solve.

4dartist
02-08-2008, 11:35 AM
OK i'll give that a shot. Thanks for the tips.

Castius
02-08-2008, 11:44 AM
You should be able to not have IK shack from what your doing.

But in order to save time trying to help you fix your setup. You might want to try using the PLG IK setup.

http://homepage2.nifty.com/nif-hp/index2_english.htm

4dartist
02-08-2008, 12:44 PM
Thanks for the link Castius. I've spent some more time on the problem and think I have a solution for now. I reversed the IK chain so bones point to the where the arm mounts to the shuttle, thus making that the shaky end. I also took 'adamredwoods' advice and put limits all over the chain and the other end is still shaking but the astronaut end is smooth as butter. I think for the camera's we'll be using that the other end shaking won't be noticeable.

I've never used any sort of IK plugin and am a little worried about investing the time it may take to figure out how to use it. Deadline for my animatics and all. :( But I very much want to see what other options are out there for IK so I'm bookmarking the link and will give it a try when i can take a breath from my deadline. Hehe.

Thanks for the advice guys!

Castius
02-08-2008, 01:18 PM
If you have no time Then you can resort to tweaking the IK settings.

http://forums.cgsociety.org/archive/index.php/t-197961.html

bgraves44
02-08-2008, 01:22 PM
Sometimes i find that I have inadvertently left something parented wrong. Like if you use the "uncheck parent in place" trick to position your nulls and not unparenting them afterward. It seems LW doesn't like goals to be parented to the base object in my (limited) experience.

Also try to have LW using IK to solve for just one axis if possible...

Hope that may help.

B

adamredwoods
02-08-2008, 04:43 PM
...And one last tip.

One of your bones may be getting close to gimbal lock. (Usually seen when a bone gets close to 90degrees Pitch from the parent) You can fix this:

Add a buffer bone, which is very small, 0% effect, and has a slight bent.
By changing the RPR of the bone, which is not always adviseable, since it is harder to manage.

bgraves44
02-08-2008, 04:51 PM
Sometimes i find that I have inadvertently left something parented wrong. Like if you use the "uncheck parent in place" trick to position your nulls and not unparenting them afterward. It seems LW doesn't like goals to be parented to the base object in my (limited) experience.

Also try to have LW using IK to solve for just one axis if possible...

Hope that may help.

B

Sorry, correction. Meant to say don't like to be parented to your root bone (or any other bone). Parenting to your base object is probably ok. Or if you used that technique your goal can end up as the child and the goal of your last bone. That will cause that jittering, too.

B