Setting envelope weights via ICE

[Jules Stevenson]

Hey Gang, is there any documentation for correctly setting envelope weights via ice? 

I can see from dialing down the available properties of the cluster you have the per point envelope weights per deformer array, plus another attribute of per point envelope weights (so not referenced against the deformer index array, I'm presuming this is just for data visualization). However when I set these I can see the weight colours pop into existence in the view-port, but the actual envelope operator does nothing, it fails to bind these new weights to the object - the mesh stays still as the deformers move.

Is there a best practice involved here?

Many thanks for any help,

Jules

[Alan Fregtman]

Perhaps ditch the classic EnvelopeOp for the ICE envelope equivalent? The "Dual Quaternion Deformation" compound can do Linear (classic) Blend Skinning.

If you want to ditch DQ altogether you can dig into the compound and strip out the DQ logic, leaving only the linear bind.

[Jules Stevenson]

Ahh, interesting, thanks Alan. Will have a look...

[Peter Agg]

Even with the DQ compound you still have to paint a regular envelope, mute the operator and replace it with an ICE one. You also have to be very careful about the group ordering you give the compound as, if the order changes, it can get a bit funky!

Basically I wouldn't really recommend it unless you need DQ. :) Still fun to check out.

[Alan Fregtman]

Yeah, and if you delete the Envelope Op it gets confused when painting. It's rather lame that you have to mute it.

I should've clarified I didn't mean to delete the op itself, but mute it and use an ICE equivalent instead.

[Jules Stevenson]

Yep, just worked this out :). The whole process of envelope generation is automatic, so ordering shouldn't be a problem... I've still got issues with the rig not actually deforming the geo, even with the DQ / ice based setup, but working through it.

[Edy Susanto Lim]

Hi,
Setting the envelope weight using the per deformer weight seems to work just fine as long as you know the deformer index.
In the attached scene, I use a null (E_ref) to define a new weight area of the new influence null (E).

Cheers,
edy

--
Edy Susanto Lim
TD
http://sawamura.neorack.com

[Jules Stevenson]

Thanks Guys,

Turns out the reason the envelope operator was being ignored was something to do with the execution order on the object. For some reason the enveloped object had at some point had a simulated tree on it so had all the simulation stack entries. As soon as these were removed (new object) problem went away.

Cheers,

Jules