View Full Version : What is up with clusters and translation

03 March 2003, 06:07 AM
I have been unable to figure out how to link/parent/constrain clusters to anything but their parent objects with "relative" set to true?

Anything but that and I get annoying double translations:thumbsdow

Can this be true? Is there no way to rig deformations on a character with clusters and parent them to something like a null?

03 March 2003, 11:35 PM
Try this: Make a relative cluster. Then group the cluster to itself and center pivot the group before putting the whole group in the objects hierarchy. No double xforms anymore, when you move the whole hierarchy.

If you have your cluster set to relative and parent it to any other transform it will be grouped to itself automatically by Maya.
For some reason this auto-group feature doesn't work when doing the parent operation in outliner via MMB drag. :hmm:


03 March 2003, 04:23 AM

First of all... thanks for responding. I am still in need of a solution to what I am looking for. I had already arrived at what you suggested by freezing the transformations. I have since tried what you suggested "center pivot" and it seems that they do the same thing... and works on a basic level.

..... but I am trying to constrain the null/group to a path (that is then within the hierarchy) and it then breaks at that step. I am now going to proceed trying the different group options.

For some reason this auto-group feature doesn't work when doing the parent operation in outliner via MMB drag.

Things like this make me cringe:eek: As now I also question the order in which everything is connected.

I need this to work:shrug:

03 March 2003, 06:50 PM
Originally posted by rhythmone

Things like this make me cringe:eek: As now I also question the order in which everything is connected.

I need this to work:shrug:

yes, things like this are really embarassing to do in maya. especially if you don't like nodes hanging around on top level or don't like nodes in strange hierarchy levels in general (happens a lot if you dont have an eye on your scene when doing stuff with paths, constraints etc...)

i suppose you're planning to use the curve-cluster thing on a face?

please reply, when you've found a solution. there must be a solution.


03 March 2003, 05:41 AM

I'm not quite sure if this is what you need though???

03 March 2003, 10:10 PM

Thanks for this post and file.... and the file under facial animation. I keep getting interupted today and it has been hard to focus on understanding you structure but it is interesting.... I did look through your stuff and tried to rearrange my file as you had suggested but it's not working (but I really need to spend some more time on maybe building from scratch and I will). I am using motionPaths instead of geo constraints because.

A - I can have 0-1 values for where I am at along the path
B - becuase I don't like how geo constraints tend to float and move by themselves.

I did want to however, describe my exact scenario for you because I don't yet fully understand your setup (which is good because it is challenging me and I plan to tackle it tonight).

1. I have created the m-lines (spline curves).
2. I have created nulls that will be motionPathed to the m-line
3. I then create the cluster on the face (it is a single pt at level zero of a subD)
4. I then parent that to the null/box (which is frozen) (which is already parented to the face geo as you suggested)

oh... one other observation. When I create a box, parent that to the geo, freeze it, create the "relative" cluster, and parent that to the box.... it does work(overally geo translation) but then translating the box does not apply the cluster deformation, it moves the child cluster but no deformation, you actually have to select the cluster and move it for any deformations to occur.(which defeats the purpose)

I need to study your files more...

03 March 2003, 12:26 AM

OK.... I still don't think I understand your file fully but here is what I see:

I can reproduce you structure and it works with a geometry constraint... but as soon I try to do the same setup with a motion path, that is exactly where it begins to wack out.

I guess I am not sure the difference... but if I do the setup and use a motion path, and NOT parent the curve that is acting as the motion path, then everything is OK.

Can you reporduce or give any insight as to why this happens?

03 March 2003, 03:48 AM

hmm wierd, 'cos I could do fine with motion path. I'm sorry I should have put some info on how I achieve that setup. Anyway, here are the important bits:

1. Cluster should be point constrained to the object/null that is geo constrained or motion pathed to the m-lines.

Parenting won't not work here, cos the cluster won't get translated/rotated/scaled), when the parent is translated/rotated/scaled.

2. Only m-line and cluster is parented to either box or the model

Here i've made 2 version of setup above, one is parented to the box and one is parented to the model

My sequence of work:
1. Create m-lines
2. Create box/null -> geo const/motion path -> m-line
3. Create cluster -> move the cluster's pivot to the box pivot -> point const -> box/null
4. Create box -> parent to the model +freeze -> parent m-line + cluster to this box

Hope that helps

03 March 2003, 08:49 AM
have you tried just linking the translate and or rotate of the null/box directly to the cluster(non-relative)??? then parenting the m-lines/splines and nulls/boxes to the skull joint... the clusters would just not move from the bind positon but when the controls(null/box) moves the direct connection would move the cluster... you just need the clusters to be BEFORE the skinCluster in the deformation order on your geo....

03 March 2003, 11:03 PM
:eek: :eek: :eek: well I'll be damned, that is very interesting. THANK YOU:applause:

Now it is clear as to what I didn't specify as one of the goals were, and that was that since the cluster was parented to the null, the cluster could still be manually tweeked if absolutely necessary.... that is why I was going for that approach.... BUT now.... I see that you can just use the constraint offsets to achieve the same results.... very cool. Thanks so much for the file.

Unfortunately I haven't even gotten to dealing with the joint rig and its impact... I hope that I am not in for another suprise.:rolleyes: I am going to bookmark your suggestion for that step.

03 March 2003, 12:01 AM
Twice now I have noticed that your files contain what appears to be a empty transform, in the latest file it is parented to the overall geometry(pPlane2), but with no children (in essense affecting nothing?)... Is there a reason why these are there?

Also, what is the mlRoto1?

03 March 2003, 01:44 AM
mlroto is a maya live node

03 March 2003, 04:13 AM

glad I can help :)

that empty transform does nothing I just forget to delete it, sorry :D

CGTalk Moderation
01 January 2006, 03:00 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.