But the sliders are annoying to use, and even more annoying to create new ones as I add shape keys for expressions (or heaven forbid, mouth shapes for speaking). I don’t want to end up with dozens and dozens of sliders.
Is there a “standard” way to lay out face controls for shape key based faces that won’t drive animators up the wall? What do professional rigs do here?
I’ve picked apart some Rigify-based models but they all use bone deformations for the face. I’m not a huge fan of weight painting faces so I’d like to avoid that…
Most “professional” rigs that I’ve used use bones that correspond to things like mouth corners on the face etc. The reason being that some game engines can’t handle morph targets or blend shapes. But in film, they use bones and then on top of that, they use morph keys to enhance things like wrinkles, creases etc. because that is hard to do with bones. But the sliders are bones so that you can put multiple shapes on one item. For instance on one bone you can easily put 18 shape keys. However you need to explain that to animators or your going to get killed. Organization and a roadmap become critical. Put it in a text window as a script. Make use of bone groups so you can turn the parts of the face on or off at will. IE Mouth, eyebrows, forehead creases, etc.
Put your big movements on bones and subtle details on shape keys. If your doing a character for a script, look at the script and only put in what is needed. If the guy don’t smile, you don’t need a smile eh? If you miss a couple expressions you’ll get it back to add it later. Usually in a hurry I might add. Weight painting is fun. I also like watching paint dry.
Thanks for the detailed response. Interesting, good to know about the game engines. My models are for animating in Blender only, so I can use as many shape keys as I want (up to the point where the animator stabs me at least).
When you say that on “one bone you can easily put 18 shape keys”, do you mean that, as that bone is moved, you activate a bunch of shape keys that are blended together? (e.g., as you raise the corner of the mouth, also wrinkle the corner of the eyes and raise the lower eyelid and …)
For organization, my biggest issue is the sliders/handles themselves. Do they usually get put off to the side? Are they attached to the face (if using shapekeys only, probably the handles would be on the face if they’re moving around deformation bones)? Is there any “standard” layout?
Excellent point about only making what you need based on requirements though. My characters don’t need to speak or do lip syncing, so that eases a lot of work. I’ll go look carefully at what else I have planned to see what I can drop for now…
Not that stilltrying is necessarily wrong but a lot can be done with shapekeys and its very dependent on your final goals. When I look at your current setup I don’t think you are aiming for the most complex all encompassing thing on earth. You want something that works well enough for basic expressions (not even speech) and shapekeys are fine for that.
The slider approach is a little old fashioned though. Usually multiple shapekeys are joined together in joystick like controls. The book is a little old by now but Jason Osipa’s “Stop staring, start animating” is a great resource for shapekey based facial rigs.
To get an inkling of how something like that would look there is this awesome thread on a web series called Boon & Pimento with very, and I mean very, extensive breakdowns of various parts of production including rigs.
I suggest you take a look, I think it will be very insightful.
And in summary; I don’t feel you have to go of the deep end and go for full bone based facial rigs at this point. There are a lot of solutions in between that and what you currently have.
What is generally done is to have a mouth corner bone, when it’s moved up it drives a smile_R shape, when moved down it drives a frown_R shape , because you don’t use smile and frown shape at the same time it works.
You can also use the left/right location of the same bone to drive other shapes that extend or contract the lips. That way one bone can control 4 shape keys already.
Both setup are doable, the main issue with controlers parented to the face is that it can get hard to animate when the character is in a strange position. One way to workaround this is to parent a camera to the head , like in the Boon & Pimento rig.
The other solution is to have a dedicated panel. that is cleaner, that also allow to really see the face without any disturbing controllers, but that takes more screen space :
There are no standards for interface but there are good practices that you can find on many professional rigs, try to look for some professional rigging demo-reels for inspiration. If you work with professional animators ask them what they need and design the rig’s functionality with them.
Also don’t be afraid of making mistakes, world isn’t perfect and rigs either, a good animator will always try to work around limitations and will make something good out of it. Of course it’s better for them if they can concentrate on their work rather than put their energy on messing with the rig. Once your project is done try to get feedback from animators and they will point out what to improve on the next project !
Good luck ! and thanks @3dioot for still remembering B&P thread, I hope you’re doing well !
The Rain V2 rig really shows off a super advanced face rig. I think they did it with bones, but it is so complicated with custom code and stuff I have trouble understanding half of what they did so I could be wrong. I can’t even get the CloudRig tab to show.
Looking at the video you get a good idea of what a big time face set up is like.
