Face Rigging vs Using Morph Targets for Animating Face Expressions

Swayam_9shah

Hi, I would like to know a bit about Face Rigging vs using Morphs. We are a small studio but we want to add the features of Facial expressions movement. We focus on getting atleast 10-12 expressions on the face. So would it be worth to setup Rigging or can we get away with Morphs using Zbrush? MoCap is still under discussion, but if we use MoCap then Rigging is the way to go as far as I am aware.

Would be glad to hear some good workflows being used. Thanks

Archanex

I'd say it really depends on your needs, but for most small studios I would probably recommend rigging with joints. Joints will give your animators a good amount of flexibility without a ton of setup time. If you're hoping to achieve the level of fidelity seen in feature films like Planet of the Apes or The Avengers you will most certainly need morph targets, or at least corrective shapes on top of a joint rig. Here's a quick list of pros and cons

=======
Bone rigs
=======
PROS
-----------
(relatively) low setup time
good amount of flexibility for animators
Works with real time game engines

CONS
-----------
can be difficult to get really specific face shapes the way you can with morph based setups
Can be difficult to strike a balance between too many and too few controllers that don't negate each other and make things difficult for the animator

=========
Morph Rigs
==========
PROS
---------------
You have complete control over every shape, the only limiting factor is your models topology

CONS
--------------
Can feel very linear when moving from one blend shape to another
Doesn't allow animators much flexibility unless lots of blend shapes are created very specifically to all work together
Expensive for real time game engines

Hito

good summary of pros and cons. There is a huge range of fidelity in face setups... joint rigs can easily be anywhere between 10~100+ joints in the face. blendshape rigs can push into thousands or targets when you count the primary, secondary, and corrective shapes you may need to create.

I'll add to it by saying you don't have to go all one or the other. You can definitely do mocap body with blendshape face, the head can be separate mesh object for blendshape animation. combination can work as well, though a little more technically involved in engine, you could drive blendshape channel from the joint animation clip with an embedded curve, or a curve added in engine.

Additionally Joint rig offers some more flexibility allowing you to generate a blendshape rig very quickly. then you can adjust the blendshapes further in say Zbrush; or regenerate the blendshapes again after adjust the pose using joint rig. Additional blendshapes can be added simply by keying additional face poses. the generation  process can be scripted, giving you very quick iterations.

it's fairly easy to switch back and forth as well by connecting the blendshape values to matching joint rig controls or driven keys on the joint rig. and vice versa.

At the end of the day it really depends on your time budget, strengths of your team, and requirements of other tools in your pipeline if you are using face animation/capturing software. if you have strong zbrush artist and know how many blendshapes are needed, blendshape might be the quicker route. if you have good animator/rigger, and is unsure how many blendshape targets are needed, joint rig might be better option.

Swayam_9shah

Hi, Archanex and Hito, Thanks for the pros and Cons and giving the time to answer me. I think the best idea like Hito mentioned is to have a combination of Joints and Blendshapes.

like you said in your comment: "Additionally Joint rig offers some more flexibility allowing you to generate a blendshape rig very quickly. then you can adjust the blendshapes further in say Zbrush; or regenerate the blendshapes again after adjust the pose using joint rig."

I am not sure if I can setup rig for the face, setup the base expression pose, get that to zbrush and bring back blendshape to Max. Is that the process you were suggesting from the comment above?

Thanks so much guys for your help.

Hito

I am not sure if I can setup rig for the face, setup the base expression pose, get that to zbrush and bring back blendshape to Max. Is that the process you were suggesting from the comment above?

I had a joint rigged face and needed a blendshape version. I posed the blendshape targets I needed on the timeline and then duplicated the mesh at each pose and added that to the blendshape rig. There were 80+ blendhape targets and I was able to generate it less than a day. Further iterations to add/adjust targets only took minutes. Not 100% sure if Zbrush will maintain the vert order after import/export, if not you'd probably have to reproject (IIRC) the targets in zbrush.

If you only need 10~12 targets, blendshape is probably quicker to start with. The characters from Unity's Blacksmith shortfilm are good examples of using joint rig for body and blendshapes for face animations. You can download them from Unity market place.