Morphs V/S Bones for Facial Rigging

Hey guys, I’ve been working on a Realistic CGI Movie using UE4 since 3 months…Script for the film is ready. Now I’ve started filming by sculpting characters for the film… I’ve sculpted one of my characters in film. I usually work in 3ds max. This is HighResShot of that character (Matthew McCoughaney as Joseph Kerr = Joker) from UE4 while I was testing it out… Now, Which is better for realistic expressions for characters? Morphs? or Bones?… Looking forward for the reply.

Morphs…

Ok, May I know the reason? I mean why Morphs? What are the benefits of using Morphs instead of Bones?

https://forums.unrealengine.com/development-discussion/animation/113489-when-to-morph-and-when-to-rig-face

Based on the explanation of your needs morphs is the more logical way to go with out knowing what software you are using to author the animations as it would be a more direct effect of creating the shape you need and is a feature that most applications, like blender, supports with out any additional set up requirements. You create the shape you need and then animate it.

Cluster shaping requires a lot more set up that effects a large area of vertices so if your character needs to act it does not have the ability to move or slide skin across muscle groups in a proper manner with out a lot of joints that becomes difficult to manage. Just opening your mouth for example the entire face can become involved from the neck up to the eyes and since shaping is controlled by joints it becomes difficult or impossible to hit the shape you need with out adding additional bones.

For example the actress Sigourney Weaver as a big blue avatar would not be able to emote her acting abilities through an avatar with out looking like she has some form of muscle damage using cluster shaping and since your actor is Matthew McCoughaney there is no way to hit the proper shaping that says this is Matthew McCoughaney with out constant adjustments.

I agree with you… Thank you so much for the reply, sir… :slight_smile: Now I can use the ‘Morphs’ way without any negative thoughts, Thanks…

Blended morph targets give summed value, and blended bones give average value.

Morphs are good for blending between layers, for example, to combine facial animation (one layer) + lips animation (another layer atop of facial expression). But two morphs at the same layer distort the face.

Bones are good for blending within a layer: for example, you can blend together different facial bones affecting the same part of face to get desired expression or can blend multiple mouth bones to get phoneme animation. But blending between facial expression and mouth would result both animations in half value.

Hmmm… Thank you so much for the explanation… After “**FrankieV” **answered this question… I chose the ‘Morphs’ way, because it wouldn’t make the Character look common or unrealistic… BTW Thank you for the reply :slight_smile:

Humm not sure what you mean. Almost sounds backwards.

Bones work very well as layers and the same animation sets can be used on 100’s of different characters.

about 16 layers so things like the eyes or mouth opening and closing can be done independent using the same animation takes so clusters works well on video game characters where there is no real need for a high fidelity output. All of the Paragon character models for example are rigged for cluster shaping .

Morphs on the other hand blend based on weight as an additive and only effects vertices that has an offset off of the target

So

Blended morphs gives you a sum total of the additive value blended bones gives you the absolute value

But how about using both? I mean, using Bones for Eye lids & Jaws… and using Morphs for Expressions… But I think it’d be time-taking and risky to use both of them, what do you guys think about this?

Yeah sure why not. Depends on the purpose that needs to be served and in the case of making a video it’s about what it takes to get the render. Once you have the render whatever it took to gain it becomes disposable as the purpose has been served. In a video game using clusters authoring the animation is the same as animating a walk cycle and will work on all of the player models using this pathway. You could even do eye blinks and eye adaptions as a procedural.

Or you could make unique morphs for each character

Or you could combine the two.

That’s the wonderful thing about UE4 is it does not hardcore limitations that you “have” to select one or the other.

Genesis 3 for example makes use of both clusters and morphs so by default you could use either based on impulse

Bones.
Take an animation of facial expression. Take an animation of mouth. Blend them with 0.5 alpha. Result: 50% of one, 50% of another. I.e. mouth is half-closed. Sum of animations - 100%.

Blend spaces.
Take an animation of facial expression. Take an animation of mouth. Blend. Result: both are applied for 100%, mouth is atop of face expression.

No, blended bones give an average value for all affected vertices. You also can override animation for some bones (that’s what you meant by ‘aboslute value’, I guess), but in this case you lose background animation. Sum of animations affecting the same bone is always 100%.

OK so saying the same thing in a different way :wink: Sum total of 100% is by definition an absolute value and all influences has to add up to 100%. Anything else would be like saying the world is flat. :wink:

In the case of layering I was referring to the ability of targeting joint animation to a set bone so the “shape” can be pulled to 100% where another shape can also be pulled to 100%. layered with out effecting the supporting shape. This way eye blinks can be added as a procedural with out effecting the lip sync for example.

Interesting idea using blend spaces. Never considered using them.

I just had my breakfast (it’s morning here), and I’m gonna start working on Morph targets first.

use Pose Assets.
They have the flexibility of Bones ( = Retargetable) while having the possibility to drive them like Morph Targets.
You can make the Pose Assets Additive, which means you can drive multiple Poses from 0 to 1 instead of having a total sum of 1

I just recorded and uploaded it to YouTube (because I don’t know how to upload a video over here)… I created 2 morphs for this character ( Yeah, just 2. because I was working on ‘BatCave - Enviroment’ which will be in film)… I animated the model in 3ds max and exported it as .FBX … and the result came as it was in UE4 which is good… It’s pretty good and smooth in UE4, but in this video which I recorded is pretty laggy (dunno why)… and I attached a photo of that particular Morph which I used… I’ll keep updating…

Hi,videolink is dead.How is going on your project? I am work on Sequencer rightnow. Hope learn something from you…