Re: Equivalent to Maya's Wrap Deformer...

Tue, 29 Oct 2013 12:46:08 -0700

Ah, that makes sense. And unfortunately for me, throws this solution out the window. Time to look at something else. Thanks David!


On 29/10/2013 3:38 PM, David Barosin wrote:
My guess is that when you extract polygons it makes a new object with the center at world 0,0,0.   The extracted polygons are not really deforming with the source geo but rather getting extracted at each frame in the new pose.  So the cage deformer is seeing this as a new bind pose rather than a deformation.   



On Tue, Oct 29, 2013 at 3:28 PM, Steven Caron <[email protected]> wrote:
did you freeze the weights? the performance should pick up once you freeze the weights. also, use settings like where mentioned earlier in this thread (search distances and such)

also, i don't know much about your pipeline but why does animation need to see these mud patches? surely this could be a job for character effects department? or maybe they are hidden right before point caching?

lastly, an ICE cage deformer is possible, here is one way to do it...


not the most ideal because of the mesh copy required, which can that can be removed if one knows the cavaets with ICE attributes, ref models, and ICE optimizations


On Tue, Oct 29, 2013 at 12:16 PM, Sergio Mucino <[email protected]> wrote:
Reviving this topic...
I got back to working on this issue, and found a couple of issues.
First, I'll describe what's going on. I've got a character's coat, to which we're attaching some mud patches. To do this, I selected the polys under the mud patches, and extracted them as suggested before. Everything works (the extracted mesh does follow the original coat's deformations). My idea was to use these extracted polys as a cage for a cage deformer on the mud patches.
I'm having two issues. First, creating the cage deformer on the mud patches mesh, using the extracted mesh as the cage, has thrown out any sense of interactivity out the window, big time. The rig went from real time to "wait 10 seconds for anything to happen". The animators will definitely not work with this.
Second, I don't know if it's related to the above or not, but it seems the cage deformer is not really working (as expected, at least). There's a mud splat on the forearm, for example. When SI finished applying the cage deformer to the mud patches, I can see the weight points all over the mud mesh. Fine. However when I rotate the forearm up (and after waiting for a while for anything to happen), the patch on the forearm loses all its weight colors and goes completely blue, staying in its place, instead of following the forearm. Its as if the cage that influences it walked away, leaving it with no weights (instead of it following the cage). I don't know if its because the deformer does not understand the procedural deformations happening on the extracted mesh or what. I also tried moving the cage operator from the Modeling stack over to the Animation stack. It made no difference.
In any case, the complete grind in performance may stop me from using this approach at all. I'm open to additional suggestions. If anyone knows of alternative ways (ICE?) of deforming a mesh with another, I'm all ears.
Before anyone suggests this one... GATOR is not an option (the original coat uses a very complex system of layered deformations involving envelopes, lattices, and ICE trees. GATOR seems to ignore a lot of these things).
Cheers!!


On 24/10/2013 6:03 PM, Sergio Mucino wrote:
This is definitely gonna work. Thanks Grahame! I'll continue tomorrow.


On 24/10/2013 5:54 PM, Sergio Mucino wrote:
Ah, if I had known that... :-)
I spent like half an hour deleting polys on my copy. But of course, I ended up with a completely independent object. Sound like what you suggest would retain a connection to the original object, which is what I'd need.
I will give it a go.


On 24/10/2013 5:53 PM, Grahame Fuller wrote: 
Oh I must have missed that, sorry. On the original object, select the polygons that you don't want to delete and choose Model > Create > Poly Mesh > Extract Polygons (keep). The selected polygons will be copied to a new object (the old ones will be left alone). As long as you don't freeze the Extract Polygons operator, it should update properly.

gray

From: [email protected] [mailto:[email protected]] On Behalf Of Sergio Mucino
Sent: Thursday, October 24, 2013 5:45 PM
To: [email protected]
Subject: Re: Equivalent to Maya's Wrap Deformer...

Looks like I will have a few problems here...

1. I cannot reduce the cage polygon count. Actually, it is a hi-res model (not crazy polygon counts, but still). I need to attach to it a copy of itself, but with lots of polygons deleted. Kinda like attaching polygon patches that need to follow it. So, I'm trying to get a lower-res mesh to follow the high-res

2. The high-res mesh is deforming. It actually has its points set to follow another mesh (another copy) via ICE. The original mesh is the one with all the deformation operators on it (pretty complicated stack of deformations). We don't touch this third mesh... that's why this second high-res has a get/set point positions ICE tree. It's the render mesh.

I'm thinking of a different solution... hold on... I might have something that works...

[cid:[email protected]]

On 24/10/2013 5:35 PM, Grahame Fuller wrote:

It's used for a few things, but mostly to keep track of the bind pose so that Softimage knows when a deformer has moved. If it's there it means that at some point you used that object as a deformer. If you're not currently using it as a deformer then you can safely get rid of it, then try to add the object as a cage deformer again. A new property will get created with the current transform values.



gray



From: [email protected]<mailto:[email protected]> [mailto:[email protected]] On Behalf Of Sergio Mucino

Sent: Thursday, October 24, 2013 5:22 PM

To: [email protected]<mailto:[email protected]>

Subject: Re: Equivalent to Maya's Wrap Deformer...



There is. All scale components are set to 1, and all Rot/Pos components to 0.

What is this for? I don't think I had seen it before...



[cid:[email protected]]



On 24/10/2013 5:16 PM, Grahame Fuller wrote:



 Static_Kinestate property already on the object before


Reply via email to