Please help to eliminate cyclic dependency

[wesserbro]

Seems no one is here, but still ask...

XSI_cycle.jpg

bone C [position constraint] to null B
B [child of] A
I need C rotation to drive A rotation.
Technically its a cycle, but logically A drives(via B) only location of C, not rotation, so theoretically it can be easily solved.
Maya can do this, blender can do this. Any help with Softimage?

[julius]

You can put an expression in the "A" rotation kinematic, based on the "C" one

[wesserbro]

Thanks, Julius, it works, but... not always. If i select C and set some keys - its good (for manipulating and playback both), but if i select A or B it breaks Then I can go back select C, and until i set another key it remains broken.

[Mathaeus]

I need C rotation to drive A rotation.
Technically its a cycle, but logically A drives(via B) only location of C, not rotation, so theoretically it can be easily solved.
Maya can do this, blender can do this. Any help with Softimage?

Once you draw a bone chain, SI automatically creates IK solver, too (or, keeps the thing involved), so, it does not allow such things in advance. Blender and Maya don't create IK solver automatically. If you want to get Blender and Maya behavior, *maybe* you'll be able to get it with a hierarchy of plain nulls, instead of bone chain. Most likely not, if SI evaluates the entire transformation by object , not by SRT component. 'Entire transformation by object' sounds more logical for my taste.

[wesserbro]

HI, Mathaeus! In my scene bones are just two separate bones, not a chain. Anyway, with nulls it isn't stable either.
Sometimes its completely good, sometimes its a bit of lag, and sometimes its totally broken, so probably its the "entire SRT as a whole" thing you'v mentioned...

P.S: too pity, im trying to rig a shoulder, where upper arm should drive the rotation of a scapula but still should be parented to it. And adding separate control object to drive both of them seemed to me not very elegant idea.

[julius]

can you provide an exemple of the behavior you're asking for ? (video capture within maya or blender)

I'll give it a try.

[Mathaeus]

HI, Mathaeus! In my scene bones are just two separate bones, not a chain. Anyway, with nulls it isn't stable either.
Sometimes its completely good, sometimes its a bit of lag, and sometimes its totally broken, so probably its the "entire SRT as a whole" thing you'v mentioned...

It's looking like time to run

Code: Select all

CycleCheck ();

using JScript. Theoretically you can get a sort of 'SRT split' in one ICE tree. Once it comes to dependency between objects, I'd believe it's 4x4 matrix in SI (or 'entire SRT as a whole'). Actually it's interesting how it's even possible to get this in Maya or Blender - perhaps there's 'SRT split' inside the bone chain, don't know...

[wesserbro]

can you provide an exemple of the behavior you're asking for ?

hi Julius. Here is a blender screencap:

scapula consists of 3 bones just to resemble real scapula shape.
purple color on a rotZ channel of a scapula means a driver. Expression: var/3, where var = upperarm.rotZ
You can see a behavior for manipulation, but it's as smooth at playback. While in XSI it can be normal during manipulation, but breaks on playback.
Hope you understand whats going on

[julius]

ok, I better see what you need.
I tried with a 3D chain and an expression like I suggested you, and I have no problem. It works well during playback too.

[wesserbro]

Who would think that it should work for a chain, but not working for separate objects and bones? Not me ))
So much noise from me! ))
Sorry, it actually works fine for a 3d chain.
Thanks again, Julius!