Morph target BLENDING in softimage (ICE)

[TonyMag]

First of all, I'm new in the forum so apologies if I've got something wrong. I will try to present my case in a short direct way:

In softimage when you try to blend between 2 morph targets (normalize) the way it works is that it takes ALL the vertices on the mesh and normalizes their position. What you really want is a 'blend' look like the way it works in Maya, where ONLY THE COMMONLY USED VERTICES ON THE 2 MORPH TARGETS GET NORMALIZED. That way you get a 50-50 look in the common area of the 2 morph targets.

Now, I'm working with ICE for my morph-face rig and I wanted to know if there are an solutions/ideas for this problem.

Thank you for your time

[Bullit]

Maybe using clusters of said vertex? or a filter to most moved vertex?

[csaez]

Sorry for the dumb question but... what's a morph target?
Are we talking about blendshapes/shapekeys?

As far as I know, a shape property stores a delta between the base object and the target. If there's no change in some vertex, the delta is equal to (0,0,0) and nothing happens.

[TonyMag]

Hi and thank you for your replies.

Bullit: Yeah clusters could work, but then you might as well go with weight maps. In a complex/advanced face rig though, you'd need to have a weight map for each blendshape compination (a nightmare). AS for the filter, hm, you might be on to something there.

Csaez: Yeah a morph target is a blandshape. Sorry, different vocabulary due to my years with 3DsMax.

The problem with what you said Csaez is this: In softimage, when a point is moved by 2 morph targets at the same time, the values are ADDITIVE. Meaning that if the vertice moved to position x by 1 and the same by the other morph target, then it's final position would b x 2. In maya though you have the option to tell it to blend, so it would find the middle value between the 2 morph targets.

Practical example >>> If you have 2 eyebrow-up morph targets and you are using them at the same time, then the eyebrow will go WAY UP. The end effect that you want though, is a BLEND between the 2.

Hope I've made it a little clearer now.

[rray]

Would an automatic normalization like this work? This should be easily doable in ICE...

For each shape, each points is assigned an "offset" scalar attribute which is its distance to its position on the undeformed mesh.
That very same value is used as the "blend strength (factor)" for determining the normalized distribution of the mixed shape points.

[TonyMag]

Would an automatic normalization like this work? This should be easily doable in ICE...

For each shape, each points is assigned an "offset" scalar attribute which is its distance to its position on the undeformed mesh.
That very same value is used as the "blend strength (factor)" for determining the normalized distribution of the mixed shape points.

Hi rray,

While what you said makes sense (most of it ), I am now trying now to 'translate' that into ICE language.

A more clear definition of what I want >>>

When/IF 2 or more morphs are activated, the weight of the vertices affected -ONLY THOSE- should blend.

I think I'm close to cracking this.

Close. But no cigar.

[rray]

The only problem I see with this method is that "weak" points will be made even weaker:

Say this is you baseshape
|
|
Now you add shape #1:
--------|
--------|
+ shape #2:
--------|
|
you'd get
----------------|
--------|
which you could say you divide by 2 because your strength total is 2, then you could get a "normalized" result:
--------|
----|
But as you say you rather aim for something like this result from the addition:
---------|
---------|
which you could get by weighing the points by their individual magnitude in the mix: point#2 of shape#2 would have no contribution in that case which is what you want.

The problem I mentioned is if you mix 2 shapes like
---------|
---------|
and
---------|
---|
you wouldn't get
---------|
------|
but more something like
---------|
--------|
because point #2 in shape 2 would be considered less . This is also what I suspect maya does, maybe it's not a problem