Results 1 to 7 of 7
  1. #1
    Join Date
    May 2010
    Posts
    9

    Cluster deformers : I probably miss understand the relative mode

    Hello folks,

    as I m quite a maya beginner, I m looking forward a better understanding of cluster deformers relative mode.

    As far as I know, the relative check box toggle on / off weither maya must group a cluster with itself when this cluster is parented back to an object.

    This functionality is vital to avoid double transformation when the object is moved after cluster have been created over some of its components (lets say CVs for instance)

    On maya 2009 however, it seems that even if the relative check box is UNchecked, maya always displays the warning :"cluster will be grouped to keep their position" (something like that) when I parent them back to the object using shift-select p

    I though that without the relative checked, I would come in a situation of double transformation, but i dont, and that's why I am a bit confused with my understanding.

    Actualy, the only way I found to recreate a situation of double transformation is going by myself in the hypergraph and then middle-click drag-drop my clusters under the object.

    Considering the experiments I have performed, it seems that the relative parameters has no impact on the double transformation situation.

    I would be realy pleased to have some more explanation !

    thanks a lot.
    Last edited by komakTm; 06-05-2010 at 05:40 AM.

  2. #2
    Join Date
    Feb 2005
    Location
    /..
    Posts
    2,597
    The help file explains it like this:

    Mode

    Specifies whether the cluster deformation will occur only when the cluster deformer handle itself is transformed (moved, rotated, or scaled). With Relative turned on, only transformations to the cluster deformer handle itself will cause deformation effects. With Relative turned off, transformations to objects parented to the cluster deformer handle can cause deformation effects.

    For example, suppose you are using a cluster deformer to smooth deformation effects around the wrist joint of a character’s skinned arm. If you create a cluster deformer with Relative turned on, and then parent the cluster deformer handle to a wrist joint, you can rotate the shoulder joint without causing cluster deformation effects around the wrist. But when you move the cluster deformer handle itself, you cause cluster deformation effects around the wrist. Default is on.
    There is an exception to every rule, apart from this one.

  3. #3
    Join Date
    May 2010
    Posts
    9
    Thanks for the quote,

    I should have mentioned that I've been looking a the Maya Online Documentation myself.

    Indeed it explains that
    With Relative turned off, transformations to objects parented to the cluster deformer handle can cause deformation effects.
    So does it mean that the double deformation occurs only if my cluster is a parent of the object ? I don't see the use of having object parented to a cluster anyway.

    As it is explain by Buzz in the maya fundamentals - alien abduction videos, I did not catch the parent relationship to be in that way, but in the opposite !

    The more I get, the less I understand ...
    Last edited by komakTm; 06-05-2010 at 09:59 AM.

  4. #4
    Join Date
    May 2010
    Posts
    9
    Let me add that in the example provided in the help file, the parent relationship is described in the way I thought it was
    [...] and then parent the cluster deformer handle to a wrist joint,
    .

    So in that example, the cluster is parented to the joint, not the opposite ...

  5. #5
    Join Date
    May 2010
    Posts
    9
    So what's going wrong with my understanding then ?

    Maybe can someone give me a simple and concrete counter-example where checking and unchecking the relative attribute would make things behave differently.

    I did not try with the one provided in the online documentation 'cause it seems to be in contradiction with the definition itself
    (I am still unclear about the meaning of 'parented to the cluster deformers' anyway)

    hope I could finaly cope with that nicely

  6. #6
    Join Date
    Jul 2002
    Location
    In space
    Posts
    5,428
    Ah, I think you're missing the purpose of the Relative checkbox. Look at it this way. Maya is asking if you want to be able to deform your clustered vertices relative to the cluster's current location, rather than absolutely. When relative is off, the vertices of the deforming object will always follow the absolute location of the cluster. However, Maya will still parent to preserve position. The (sort of) funny thing is that with Relative OFF, it really doesn't have to. This just tells me that when the Cluster's class was written, that particular functionality simply wasn't tied to the Relative attribute.

    In the end, though, that doesn't really matter much. It's when Relative is ON that you can see the value of that group. When you parent a Relative cluster to another object, the cluster gets an internal offset from the parent's current location. It's easier to see than explain, so here's an example where you can see the actual purpose of Relative mode and then why the group is there. We'll start with Relative OFF.

    1. Create a poly cylinder with a lot of height segments.
    2. Create a locator just above the cylinder.
    3. Select the vertices at the bottom of the cylinder and put them in a cluster with Relative OFF.
    4. Parent the cluster to the locator - You'll get the "Cluster grouped" warning.
    5. Move the locator. The cluster moves along with the parent, and the deforming vertices follow right along.
    6. Lot's prove that we don't really need that group with Relative off. Open the Outliner.
    7. Expand the Locator and then expand the group underneath. You should see the cluster handle node.
    8. Middle-mouse drag that handle node onto the cylinder to parent the handle directly, bypassing the group.


    No change. You can still move the locator around just as you did. The group was unnecessary, but Maya created it anyway. Now let's do the same thing with Relative ON, and you'll not only see the purpose of the setting, but you'll also see why that group is so important to Relative mode.

    1. Undo to get the locator back to its original position. If you can, undo back until the cluster disappears, or simply select and delete it. Make sure to clear out the residual group if you do.
    2. Select some vertices on the bottom portion of the cylinder and place them in a cluster with Relative turned ON.
    3. Parent the cluster to the locator - You'll get the "Cluster grouped" warning.
    4. Move the locator around. The cluster handle seems to move, but the deformed vertices do not. That is the primary effect of Relative to the end user.
    5. With the locator moved away, wiggle the cluster a little. Notice that the vertices only move by that wiggled amount, even though the cluster's absolute position is far away. This is the very heart of the Relative behavior.
    6. Now, open the Outliner and use the middle-mouse button to parent the cluster back to the locator, bypassing the group.
    7. Woah! The deformed clusters moved 100% of the original distance from the cluster handle to the locator!


    This is because in Relative mode, the cluster handle wants to internally consider its distance from the parent as an offset of how far it should move the deforming vertices. By grouping to itself, it creates an intermediate parent at the same location as the handle, effectively making the offset 0, so that the vertices do not fly away.

    Finally, to see the "double transformation" follow the setup above with relative OFF, parent the cluster to the locator, then select BOTH the cluster and the locator (not just the parent) and try to move them. Double transformation! Try again with Relative ON, parent the cluster to the locator, and again select them both and move them. The cluster HANDLE double transforms, but the deformed vertices do not.

    This can be helpful when rigging certain objects (such as a spine) when a cluster's location can be influenced by a hierarchy, and potentially by various constraints.

    Hope this helps!

  7. #7
    Join Date
    May 2010
    Posts
    9
    Woha !
    I could not have imagined more precise and clear answer !

    Indeed I was completely missing the wall thing behind that Relative attribute.
    But from now, and thanks to you, I think I am in good shape to understand exactly the role of it.

    I believe my main mistake has been that I focused on reproducing a double deformation situation as Buzz mentioned it in the introduction chapter on clusters.

    I take the occassion to really thanks you and congrate for the quality of your work with Jason on both these videos and forums. Nevertheless, I was quite suprised that you do not appear at all in the Riggings lessons of "Alien Abduction", why don't you ? Maybe the introduction on cluster deformation would have been - if possible - even greater with both of you !

    Nevermind, the wole videos are just amazing !

    Best regards.

    P.S: I certainly will be pleased sending you clusters of your favorite chocolate chip cookies if you mind tell us where you order them !

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •