Creating a cluster deformer

[jim c]

First off has anyone done something like this and made it available? I looked around, and found several expressing interest in this over the years but haven't actually found anything like it. I've been experimenting a bit, and was just curious.

If you were to do it, is it possible to create an Object level node (like a Null) and then insert a CRegion SOP in it to handle the weighting? It seems like this should work, but I'm not 100% if that's the right approach.

 

Any alternate ideas?

 

[benne5]

I suggest making a group of points and then transforming or softTransforming the group of points.  

[jim c]

Thanks, I tried that approach. What I didn't like was that it's a little difficult to control the weighting, and while the soft xfrm effect is nice, how do you scale it up? For example, on a face rig, you might have 30, 40, or maybe even 100 different nulls/clusters for the various areas on the face you want to control. To do the soft xfrm idea, you need to have either the same number of soft xfrm SOPs in the face Geo node, or some other trick? I tried doing this with a ForEach SOP and while it worked it was INSANELY slow, not at all interactive. So I figured something that could reuse the same weighting techniques found in bones would be ideal for this, you can reuse the capture painting SOP for controlling the weight, and so on.

One of the important things (I think, I could be wrong here) is that the control for this needs to be at the Object level so that you can have other things drive it's position, i.e. through parenting or other nodes.

[edward]

Can't you just use 1 softxform and give it all the points in the Group parameter?

 

To answer a part of your original question, yes can just a cregion primitives from anywhere (including feeding it from the Capture SOP's second input). But if you're not using bones, then you must tell the Capture SOP to use a capture frame to define the capture (aka rest) pose.

[jim c]

I don't think so. For example, lets say you want to control the left eye region, so maybe four null/clusters the run around the top of the eye region, another around the lower eye region. Each null needs to move/affect it's points (and each null, like a bone, would have different points that it affects) independently. As far as I know, you can't control all that with with a single softxform (I could be wrong here). From what I can tell the cluster/joint is like a really stripped down version of a bone.

[louis]

yup, a cluster deformer would be sweet! It is easy enough to build one based around the soft xf with a linked object level controller but a bone with a relative deform mode would be so much faster and surely easier.

[jim c]

But if you're not using bones, then you must tell the Capture SOP to use a capture frame to define the capture (aka rest) pose.

 

If you want to edit the capture volume interactively, how would you build a tool to do so? I don't think I was able to get the Edit Capture Region tool to with my custom object at all, seems like it's set up for Bones only.

Thanks !

[edward]

I think you can still paint though by choosing the cregion from the pull down menu.