View Full Version : How to make dynamic hairtails, etc.

12-01-2005, 05:49 AM
What's the best way to make a tail or hairtail with dynamics? IK-Booster might be an option, but what if the rest of the character should have conventional FK/IK? Softbody dynamics?

thanks in advance.

12-01-2005, 07:43 AM
Narf! Seems pretty hard to find good settings for IK-Boosters Bone dynamics... What would be good settings for weight, resistance, viscosity, etc., for a hairtail or a swinging antenna?

12-01-2005, 08:47 AM
Eh - I'd skip IKbooster and just setup the tail to have dynamics.

12-01-2005, 09:19 AM
Uhm, how do I do that? I thought there is no other way to have bone dynamics than IK Booster....

12-01-2005, 09:35 AM
Well - unless you need direct control over the tail you could assign a weight map to the tail itself (called tail) and another to the rest of the character (called fix) - apply the dynamic to the character making the fix weight map the fixed part of the mesh. Now LW will calculate dynamics for the tail. You still may be able to use bones in this setup to get some control over the tail.

oops - had to correct my post - meant to say to call the weight map for the tail "tail"

12-01-2005, 09:52 AM
Ok, thanks, I'll try that.

12-01-2005, 02:29 PM
Two words: FXMetaLink

12-01-2005, 03:39 PM
Metalink or IKBoost would work fine. You have the ability to edit the proxy object when using cloth using editfx but its only very simple controls. IKBoost gives you the ability to edit the bones and edit even further in the graph editor. Like Clothfx it sees wind, gravity, collisions etc.
You can also use IKBoost on ANY item chain, not just bones. If youre using a spline deformer such as Shift Spline Transform, IKBoost dynamics can be used on that as well.

12-02-2005, 04:01 AM
I feel stuck a little: while FXMetaLink gives great movement, is superfast and has no banking problem (because 2pointpolys), it seems impossible to find settings that do not stretch the objects like rubber (ClothFX). I'm working with 3 weightmaps: fix, struct and weight. Struct is strongest at the root of the tail, and gets weaker along it, weight is strongest at it's end.

IKBooster computes slow, gives not so good results, and there is no way to "hold" the structure like with clothFX - but it doesn't stretch the object...

... hmmmmm.

12-02-2005, 04:10 AM
Try setting the stretch limit to 1 or 2%.

You don't HAVE to use a 2-point poly chain. You could make it with a simple tube shape. Just add enough sub-structure to keep it from collapsing.

12-02-2005, 04:11 AM
By the way.. It's my understanding that Sub-Structure has no effect on 2-point poly chains.

12-02-2005, 04:26 AM
Sub structure seems to have at least some effect on a 2 point poly chain....

... ARGH! I did set stretch limit to 0% - that doesn't work !*?#!%&!

12-02-2005, 04:30 AM
WOW! there is another solution: I can create a Null + path from a node of the 2pointpolychain! - as a target for an IK chain! WOWOWOWOWOW - cool!

12-02-2005, 04:35 AM
I think a stretch limit of 0% causes problems. Try going around 1 or 2%. You can go even lower, such as .5%, but the lower it is, the slower the calculation time (of course, for your project that's probably a moot issue).

I think you'll find a dramatic difference between sub-structure on 2-point poly chains and 3 and 4 point poly cages.

12-02-2005, 05:03 AM
Ive used IKBoost to do this and it calculate quickly and behaved fine. Since your using bones they hold the structure just fine. Like clothfx you just have to play with the settings. Its even simpler IMO since you dont have to mess with fix, stuct, hold etc etc.
Heres a couple of examples I threw together in a few minutes. I roughed out the model and quickly did a test with IKB and one using metalink. Both could work fine with some tweaking. In these examples IKBoost computed about as fast as the clothfx version.
IKB Test (http://www.3dtrainingonline.com/examples/ikb_ponytail_test.mov)
Metalink test (http://www.3dtrainingonline.com/examples/metalink_ponytail_test.mov)

12-02-2005, 05:08 AM

The ClothFX version had more natural movement in my opinion.

12-02-2005, 05:09 AM
Also try using a ribbon of polys. The substucture will work fine on that. Since the 2 pt polychain or poly ribbon is simple setting the stretch limit lower wont effect the calculation much. You could also reduce the polygon size a bit.
Another thing that helps is to start the calcuation early so that the proxy object has a chance to settle before the animation begins. :)

12-02-2005, 05:23 AM
I agree to a point. The clothfx version bends the pony tail a bit too much IMO. That could be tweaked more and Id probably add more viscosity to reduce the bounce.I could have added additional bones and tweaked it a bit more etc etc as I said. The main reason I posted the IKB version was in response that it was slow and unreliable. I literally just drew the bones in quickly, added IKB, gave it some quick settings and hit the Calc button. :)

12-02-2005, 06:26 AM
Yeah, but my version was really slow - and had only 6 bones, 2 of them fixed. my best guess is that it was so slow because the character it was attached to had many bones (non-IKB) itself.

12-02-2005, 08:10 AM
Hmmm, seems to be true for ClothFX, too. Having a scene with some stuff in it slows FX calculation down considerably. Turning IK off helps, but it's still relativly slow.

12-02-2005, 02:45 PM
You can make the pony tail a separate object with its own bones. Parent that to a bone in the head and apply IKBoost to just the pony tail. You can hide where the pony tail connects in using a hair tie or something similar. You can use a similar technique for the cloth.

12-02-2005, 04:11 PM
SplineGod, this is exactly what I did. It still slows down....

12-02-2005, 04:47 PM
If you set the bounding box threshold to 0 or a very low value it will speed things up considerabley since it doesnt have to compute bone deformations while the bones are being moved. :)

12-03-2005, 06:13 AM
Nope, sorry, that's not the point. During FX calculation, no objects where shown. I've tried to set bbthreshold down, and it makes no difference. FX calculation slows down as soon as there is a parent - child relationship between the dynamic object and it's parent - the head-bone, in this case (and the head bone has no IK). Anyway, turning IK off helps, but an unrelated dynamic object calculates much faster.

I've just made two other tests, parented the dynamic object to the character itself and to a scene prop (with no keyframes). It always slows down considerably.