Useful cycleCheck script

682 views
Skip to first unread message

Michael Boon

unread,
Apr 3, 2018, 8:10:48 PM4/3/18
to Python Programming for Autodesk Maya
I've just started looking into the cycleCheck, and this warning:
// Warning: Cycle on '<node name>' may not evaluate as expected. (Use 'cycleCheck -e off' to disable this warning.)

This has become more important in Maya 2018 because getting transforms using MDGContext can now give different results than the viewport in cases that used to work.

If I look at a node that I know has problems, a simple
cmds.cycleCheck(nodeName)
returns False, because I guess the cycle is somewhere in its history.
cmds.cycleCheck(nodeName, dag=True, list=True)
takes several seconds and returns an enormous number of results.

Looking through the results, the first one is a parent constraint on a joint. This is just how parent constraints work - the constraint outputs the rotation and translation to the joint, while the joint outputs its rotation order, joint orient, and parent inverse transform to the constraint. This cycle is necessary and does not cause problems.

I'm not keen to go through hundreds of fake cycles to find the real ones that are causing problems. In addition, I doubt the problem cycle is even in the list, since most joints in our rig get their transform from nodes in a different hierarchy via constraints (ie via DG connections, not DAG connections).

So my question is, is anyone aware of a script that does a better job of this? Something that can follow history, and ignore the places that cycles are expected, and only return the interesting ones? Or does anyone have a different approach?

Marcus Ottosson

unread,
Apr 4, 2018, 2:39:51 AM4/4/18
to python_in...@googlegroups.com

Are you able to post an example or reproducible of one of the cycle check problems you are seeing?

Looking through the results, the first one is a parent constraint on a joint. This is just how parent constraints work - the constraint outputs the rotation and translation to the joint, while the joint outputs its rotation order, joint orient, and parent inverse transform to the constraint. This cycle is necessary and does not cause problems.

I don’t think this is a cycle. Cycles happen per attribute, not per node, so the parentConstraint itself isn’t to blame. In that case, one node (the joint) has incoming connections (rotation and translation) which doesn’t affect the outgoing connections (the rotation order, joint orient and parent inverse transform). For a cycle to happen, an incoming connection must be affected by one of its outgoing connections.


--
You received this message because you are subscribed to the Google Groups "Python Programming for Autodesk Maya" group.
To unsubscribe from this group and stop receiving emails from it, send an email to python_inside_maya+unsub...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/python_inside_maya/3526bc39-5f56-4e73-a64e-182396c8a12f%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Message has been deleted

Michael Boon

unread,
Apr 4, 2018, 6:13:11 PM4/4/18
to Python Programming for Autodesk Maya
If I create a cube and a cone, and parent constrain the cube to the cone,
then do
cmds.cycleCheck('pCube1.r', dag=True, list=True)
I get
# Result: [u'pCube1.rotateY',
 u'pCube1_parentConstraint1.constraintRotateY',
 u'pCube1.rotateX',
 u'pCube1_parentConstraint1.constraintRotateX',
 u'pCube1.translateZ',
 u'pCube1_parentConstraint1.constraintTranslateZ',
 u'pCube1_parentConstraint1.constraintRotatePivot',
 u'pCube1_parentConstraint1.constraintRotateTranslate',
 u'pCube1_parentConstraint1.target[0].targetWeight',
 u'pCube1.rotateZ',
 u'pCube1_parentConstraint1.constraintRotateZ',
 u'pCube1_parentConstraint1.constraintRotateOrder',
 u'pCube1_parentConstraint1.pCone1W0',
 u'pCube1.translateY',
 u'pCube1_parentConstraint1.constraintTranslateY',
 u'pCube1_parentConstraint1.constraintParentInverseMatrix',
 u'pCube1_parentConstraint1.constraintTranslateX',
 u'pCube1.translateX'] #
This is (I assume) a red herring. The parent constraint is working as
intended, and I'd like a script that will ignore it.

Marcus Ottosson

unread,
Apr 5, 2018, 2:06:26 AM4/5/18
to python_in...@googlegroups.com

That dag argument looks suspicious, how come you include it?

dag (dag) | Also look for cycles due to relationships in the DAG. For each DAG node, the parenting connection on its children is also considered when searching for cycles. - Reference

It sounds like it’s being a little more loose on what is considered cycles, especially since the parentConstraint is part of the dag and even parented to the driven object per default.

I found two cases whereby it didn’t output any cycles, but it’s hard to say whether it’s conclusive without an example with an actual cycle warning too. Would it be possible to share one?

from maya import cmds

cmds.file(new=True, force=True)
cube, _ = cmds.polyCube()
cone, _ = cmds.polyCone()

constraint = cmds.parentConstraint(cone, cube)

assert cmds.cycleCheck(cube + ".r", dag=True, list=True)
assert not cmds.cycleCheck(cube + ".r", dag=False, list=True)

cmds.parent(constraint, world=True)

assert not cmds.cycleCheck(cube + ".r", dag=True, list=True)

To unsubscribe from this group and stop receiving emails from it, send an email to python_inside_maya+unsubscribe@googlegroups.com.

--
You received this message because you are subscribed to the Google Groups "Python Programming for Autodesk Maya" group.
To unsubscribe from this group and stop receiving emails from it, send an email to python_inside_maya+unsub...@googlegroups.com.

Michael and Amanda

unread,
Apr 5, 2018, 2:17:32 AM4/5/18
to python_in...@googlegroups.com
Ah thanks, that's good to know. I was misinterpreting dag. I put it there because the node I was struggling with had no cycles of its own, but its ancestors had some (dozens, if you believe cycleCheck). Without dag=True, I got nothing, and with it, I got a page of text. Still, if this node doesn't have any problems with cycles then I've found a significant bug and I'm not willing to jump to that conclusion yet.

I need to come up with an example that's not as complex as the rig I'm looking at (thousands of nodes, completely unmanageable output) but complex enough to be meaningful. I'll keep trying...



--
You received this message because you are subscribed to a topic in the Google Groups "Python Programming for Autodesk Maya" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/python_inside_maya/FCBpdbA8I9Q/unsubscribe.
To unsubscribe from this group and all its topics, send an email to python_inside_maya+unsub...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/python_inside_maya/CAFRtmOADhP3VkSQz8Mo5CdRH1G2vnw2DxaxvwHHx6WOOEV3XEw%40mail.gmail.com.
Reply all
Reply to author
Forward
0 new messages