I thought it was about time that I tackle this subject as I see the question a lot as I read questions on different forums and groups. The question usually comes in the form of “Which facial rigging method is best for…” insert your production of choice, and is almost always asked by some one new to the business and doesn’t really understand the scope of what they are asking. I figure it is about time that I address the larger question and give a complete answer as so many of the answers that I see are either inaccurate, incomplete or highly biased to the only solution some one has used them selves.

Lets get started.

“What facial rigging method should I use for my production”?

First off you shouldn’t be trying to define what the facial rig will be until you have defined exactly what your production needs are. Almost never do I see this question accompanied by a clear understanding of what will be required from the final results yet person after person, as well meaning as they are, gives them the answer. This is often the very first problem as the solution can not be defined if the problem that needs to be solved isn’t well defined. 

Now the second problem in this question arises, the problem gets defined as “it’s a game” or “it needs to be really high quality”. Neither of these define anything at all and really only serve to muddy the waters even more. 

So how do you define your needs? 

Lets say the project is a game, first what sort of game is it? Battlefield 5 just came out and damn it looks good. It can be described as high quality for sure. How ever facial animation isn’t at the forefront of that game, there is the odd time that you see a characters face give you a glance but for the most part they are unimportant to the over all game. For the cut scenes they need to have more abilities but generally not during game play. For a project like this speed is really what matters the most. With 64 online players and just about everything blowing up around you lag isn’t something the player wants to deal with.

So what if you are doing a first person high end PC boxing game where you are facing your opponent? With this sort of setup really strong facial work would be needed. That right cross to the chin needs to distort the entire face and even have beads of sweat flying off it. But does the character ever speak? 

What about a children’s TV series? Even if the final renders are looking good the big question is how much time does the animation team have to work with the facial setup? If you over complicate the rig and add way to many controls the animator may find it difficult to hit dead lines. All the flexibility in a custom FACS based muscle system with a bone based dorito over driver and corrective targets will not help them at all. What they really need is a supper fast rig that allows them to hit the major phonemes and emotions with the least amount of work. Lets call it less is more in this case. 

If what you are looking at is a full on 300 million feature film with realistic characters that are going full screen in an IMAX theater then the sky’s the limit. If you are that person you are probably not asking this question on some Facebook group looking for advise from students in their second year at a school no one has ever heard of. So if this is the sort of rig you think you need and you are reading this… think again, you don’t. 

One of the first things that I do when I have been contacted by a client and asked to rig a character is I ask for the story boards and script. Once I have read and understood those then I ask to speak directly with the animation team that will be responsible for working with the rigs that I have created. One issue I have found is if I ask ten animators what they want I will usually get ten answers, some times I will get all ten answers in one. For this reason I usually only want to speak with a senior animator and work out with them what will really be needed. I have also had the answer “Well at Pixar we had this rig that…” and I have 2 days to build the entire rig for a TV commercial not a Pixar film. Sorting through this and coming to a good solution takes experience but without any of this information you are running blind.

What if you can define your needs?

So lets say you have done your home work and you have defined the needs for your production, what then? We need to know what the options are and what they can do for us if created properly. Properly is the key here, I have seen lots of riggers tell me that method X is no good at all but when I look at the system they built it is a complete mess, you need to understand, good facial rigs take time to create and to develop a work flow that allows you to create them quickly. Another thing that is very important is the ability to script tools to aid in the development of the rig as there are a lot of steps that can be automated to give you more time to finesse the rig.

What are the options?

There are a number of ways and combinations of ways to create a facial rig. Here is an incomplete list in order of complexity that some one will argue with me about either not including something or being in the wrong order.

Something to note is different software will call certain features different things. When I’m talking about Targets this is the same as Morph Targets in 3DS Max and Blend Shapes in Maya. C4D, Blender and others might refer to them differently as well.

1.  Absolute Targets
2. Bone Based
3. Blended Targets
4. Hybrid Bone & Targets
5. FACS Muscle Based with a nice cream sauce and side of UI design.

Of course the above is by no means complete or factual but it is a good place to start the discussion.

Absolute Targets: (Morph Targets / Blend Shapes)

This is where most budding TA’s start learning the process of facial rigging. Absolute targets refers to creating a copy of your mesh and lets say modeling it into a smiling position and adding it as your Morph Target / Blend Shape, what ever you want to call it. This is fast and easy to do but it does have some major drawbacks for many productions. One of the biggest drawbacks is that every time you want your character to smile it will be the exact same smile, the second is that it will not allow the animator different ways of getting to the smile. The advantage is if all you need is the character to turn to the camera and smile once at the end of a commercial then this is a fast and simple way to get there. You could also use it in the case of those glances that happen in Battlefield as a player runs by you cowering in the corner and says “common, move up”! All you need in that case is a look and it would be easy to program that into the game to play the, get your ass moving, sequence and maybe randomly choose from a couple different looks.

Targets have a down side for games as they are slower than bone based systems but do allow for more fidelity in the shape of the face. That means it is easier to shape in finer details like winkles or other subtle changes in the face that would take far more bones than you might want. How ever you are stuck with a pose and don’t have many options as an animator unless you create many targets and even then each is an absolute shape.

Pros: Fidelity, ease of creation.
Cons: Slower then bones, not as flexible.

Bone Based

Bone based rigs are created with a series of bones where the face is then skinned to the bones and the bones via control objects or directly are animated to create the shapes that the animator needs. Bone rigs can take more time to create and finesse especially for a new TA that doesn’t have their skinning refined. Bones create a system that is more flexible for an animator so they have control over what final poses they are after just by moving, rotating or scaling then bones or controls. They can also get to a pose anyway they like as they can work with the animation curves or add keys where ever they like. This can produce more variation in the animation and produce a more believable look.

The issue with a bone based system is fidelity, you only have the bones that you have created and unless you have created a bone at every vertex you don’t have the control over the final shape that you do with targets. The more bones you add the more complex the rigging gets and the more time it takes to get right. One advantage here is reuse, a basic bone based facial rig can be created and then added to any model that is similar easily. With targets, unless the vertex order and number is identical you can’t reuse them directly.

Most game rigs over the years have been driven with bones for the speed reasons mentioned. Also most games don’t need complex facial setups as it often doesn’t affect game play. It is just there to add a bit more realism and depth to the characters. Bone rigs are used for other production types but often mixed with targets or other tricks to produce the fidelity that might be needed. That being said for Bioshock 2 I created the facial rig for Ryan entirely with targets as he was a speaking character that would be full screen and needed to have subtle shapes created to make it believable.

Pros: Flexible animation, Reusable, Game friendly, Fast
Cons: Rigging time, Low fidelity

Blended Targets: (Morph Targets / Blend Shapes)

Marcin Porebski

This is the system that I have used the most over my 24+ years of rigging characters as it will provide excellent results quickly and produce a flexible animation rig that has a high fidelity. With this rig it is mostly created with targets as I always use a jaw bone to open the mouth. The targets how ever are not absolute poses but instead and designed to be mixed together to create final shapes and poses. For instance there is no smile target but instead a mouth corner up, mouth corner out, cheek up, cheek out, right and left squint and so on. Each of the targets is driven by a control and in the case of the corner of the mouth several targets are controlled by one control object.

One comment on system like this that I have heard is that you need lots of corrective targets to correct issues were multiple other targets are blending together. This how ever is not the case. For the most part I never create corrective targets as I have done a clean job on the original ones. This all produces a rig that is quick to build, has the fidelity that the final product requires and allows the animator to get to different poses in different ways making it flexible.

There are down sides of course to a system like this. First off it would be more work to create then both the previous system but does produce a better result. It isn’t really game engine friendly but still could be used with some in engine rigging. Although in a DCC like 3DS Max or Maya the speed of the rig will be fast enough, it will generally not be as fast as a pure bone rig.

Pros: Flexible, High Fidelity, Easy to setup
Cons: Speed, Not really game friendly, Speed

Hybrid Bones & Targets:

A hybrid system blends together both worlds and if done right gives you the best of both. If done wrong just creates a complete mess that is over complex and doesn’t really give the animator anything extra that a Blended Target system would have got them if it was created right.

Almost all the facial rigs that I have created over the years will use both bones and targets but it is the degree of bones and how they are mixed with the targets that changes. I will often have a bone for the end of the nose or a few controlling the brows but general don’t find the need to mix both into the mouth as I don’t always have time and the animators don’t have time to take advantage of what it would provide them.

Creating a fully hybrid system that works well can take time to setup and get it working in a way that will provide animators a whole other level of control. The idea is that a bone system gives the animators flexibility as they can translate, rotate and scale everything but they don’t get the fidelity, targets get the fidelity but not the flexibility. Mix the two together and you should get a rig that can do just about anything. You pay for this in the time it takes to create it and the time it takes to animate it. If the animators don’t have all day long to tweek facial performances it doesn’t gain you anything at all.

Pros: Flexibility, Fidelity
Cons: Time to rig, Slower rig speed

FACS Based Uber Rig with Dynamic Muscle cream sauce:

FACS stands for Facial Coding Action System and it is something that every TA should read about. It describes the needed shapes that a face can attain to do everything that it can do. For most productions it means little to nothing because if you are working on the next Cloudy with Meatballs, or an Elastigirl rig they don’t follow the rules of the real world anyways. But it is great knowledge to have and always fun to drop into a conversation around the boardroom when you want to confuse the crap out of your boss so they stop asking stupid questions.

Muscle systems are a neat idea but again unless you have an unlimited budget, time and know how, I would stay away from them as they don’t produce a better result, they produce a different result needed for different reasons. So unless you are that guy from ILM that just happened to drop into my site to find out how to rig faces then don’t bother with them until you have perfected the other methods first. If you are that guy, can I add you on Facebook 🙂

Pros: All your friends will think you are really cool
Cons: Some producer with no time and no budget will want you to make it for their production for no money because it will be great for your career and they will send you all the other work that they will not be getting in the near future.

Facial Control Systems

So now that you have defined what type of rig you will be needing you need to decide on the UI/UX that will be built for it. There are a number of options that can be created and all sorts of variation of those. Once again the best way to start is by talking to the animation team since it is generally never going to be used by the TA how would you know what is best. After 24+ years in the business I have an idea of what animators want but often they ask for things that surprise me. I don’t care if it is the dumbest idea I have ever heard, if that is what they want then I will create it for them. This is only if the budget allows for it and when I say budget that might be time and not money. It doesn’t really matter if the client has $10K budget for me if they are only going to give me a day to do it.

I personally prefer facial rigs that are on the face and not floating off to the side, I want it to feel as tactile as possible and not removed from what I’m working with. I guess a good analogy would be working with clay instead of working with a marionette. I have created all sorts of variations of facial control systems, coded floating panels with joystick controls that allow for pose saving, floating spline controls beside the characters head, controls right on the face and a couple other bad ideas that I have had from lack of sleep and to many frappuccino’s.

I have also seen some really odd ideas that take off for a while and then die out because they were fads, this has happened a lot over the years and it is due to people like me that will create a cool tutorial on how to do it and everyone will think it is how I would rig a character. Some tutorials that I have created over the years have been to pander to the latest craze and get my name out there, I can say this now because my name is out there. I remember making a tutorial on this crazy muscle system, I never showed it working with a skinned character and people thought that it was the coolest thing ever, the only time I created it was for the tutorial because I knew it would get people to my site. I can honestly say that everything that is on my site and youtube channel now are methods that I use all the time for rigging so don’t shy away from them.

One of the ideas that I see my students trying to build all the time is a rig not on the character but off to the side and it just loads the results as one large target to the final character. I can see some reasons for this sort of system but in general I don’t see it as being a good solution. If the aim is to have a way to just view the face then just add a camera to the character rig that is locked to the head. In some cases it is a way to solve dependency loops while trying to resolve dorito type effects with a rig. It isn’t that I think this is a bad idea, I just don’t see a good reason for most productions I have worked on. So if you have a good reason to do it and not just because you watched some guy/gals tutorial on youtube that you have never heard of until 20 min ago then do it. Otherwise rethink how you spend your time. I have seen way to many tutorials over the years that you can tell they did my tutorial once and then created their own and it is way off base and you can tell they really don’t know what they are doing. Make sure that you research who’s tutorial it is, if it is that guy from ILM and worked on the last $300 million production then they probably know a thing or two but it still may not be the best solution for what you need to do on your $300 production.

One last think I should mention is ensuring that you have very clean topology for the mesh that you are working with. Without excellent placement of edge loops and the right amount for the type of production and what you expect to get from your rig you will not achieve the results that you are after.

For what it is worth these are all the rantings of an old and tired technical director that started when chrome spheres were a really cool idea. I have never worked for ILM or Pixar but have worked on feature films, TV series, tripple A game titles, mobile games, medical work, aerospace and just about everything else that you can think of. I wish I could go back to the days of shinny flying text, things were just so simple back then.