PDA

View Full Version : Animating the face...


eek
07-18-2004, 09:51 PM
When animating the face which do you prefer?

1.Sliders

2.Viewport controls

3.Mixture of the two

The system im working on will have 3 tiers,

High level controls: i.e Facial poses
Medium level controls: corners of mouth, top lip, bottom lip, jaw, eye brows etc etc.
Low level controls: individual parts of lips, eye lids etc

And these controls will be put on sliders in the viewport.

eek

disrupt
07-20-2004, 07:56 PM
I think you can't go wrong with a mixture of the two. I am looking forward to seeing your setup. I hope you will let us know when it is done.

eek
07-20-2004, 09:18 PM
basically thats what i was thinking, In the setup you make the character talk with the top level controls i.e poses mm,ooh, ahh etc. This is the first layer. Then you start to add character to the face using the next level basically: corners of ,mouth, jaw, top lip, bottom lip. This build the character nuance. After this the third level is finite adjustment i.e lip controls. to tweak shapes and performance.

So the aproach is similar to blocking out, adding overlape etc etc. Im try to make it user friendly. And the systems really fast and smooth. Ive got alot of tweaking as im using expressions. To generate pose. But it makes it very stable and really tweakable. The nice thing is that the rig can be simple or a complex as you want ,just depends on the face.

eek

P.s once ive got some basic shapes done, ill post to a few of you, for testing. See if the tweaks etc.


also one question:

when say creating a mouth shape. Do you want the jaw to move with the shape. Or a seperate control for it so you can adjust how much the mouth opens for each shape.

Also this could be used to block out basic lip sync before additions are added?

Aluuk
07-20-2004, 11:33 PM
Hey man,


I think the jaw should def be seperate.....give the animator as much freedom as you can...don't limit him/her by not letting them open/close the jaw seperate from the phonemes. :)

Leif

eek
07-21-2004, 09:27 AM
so create the mouth shapes with a closed mouth?, or have additional control over it after shapes have been made?


eek

kevinw
07-25-2004, 04:01 AM
my 2 cents...

Personally I think you should model all of the basic phonemes with all of the jaw movement that would normally occur. Making the lips and jaw seperate and requiring them to be keyed independently on every target would be alot more work for the animator and give a greater chance for inconsistancy for repeated words/expressions. I wouldn't think of putting targets on different layers, rather, just adding more targets to the entire rig. Do all the speech/letters and then do additional controls for the jaw dropping, lowering and raising the top and bottom lips (seperate sliders) sneering, squinting blinking, furrow, the list goes on and on....... This will give you all the freedom to hit any position you want. Those secondary controls will be helpful on long vowel sounds when you want to slightly vary the shape of the mouth to get a more natural flow through the dialogue. Also, don't forget things like squash and stretch (two separate targets) to add more control over the general shape of the entire head. I find those shapes to be invaluable when trying to really get expressions across. Anyway I'm just speaking from my own experience. I'm no Jason Osipa.

Good luck and keep us posted on your progress. I'm always interested to see what people come up with.

Ravis
07-28-2004, 08:09 AM
Personally, the only viewport controller is for the Jaw open and closing, then on that controller I assign sliders that would control the facials.

mehdianim
08-15-2004, 07:18 PM
I agree The Jasw should be separate. Have the phonemes animated as you would normally but allowing the animator to open the Jaw a bit more if necessary gives much more range to the character's expressions with only one control. Plus oyu get the twist and sideways movement on the Jay which canadd a lot as well if you are not trying to be realistic.
Just don't do what most films do with facial expression which is ruin great acting by automating the animation of the mouth and not trying to match the tone of the scene...:banghead: Save for Gollum of course.

JBarrett
08-15-2004, 09:07 PM
Personally I think you should model all of the basic phonemes with all of the jaw movement that would normally occur. Making the lips and jaw seperate and requiring them to be keyed independently on every target would be alot more work for the animator and give a greater chance for inconsistancy for repeated words/expressions.Speaking from experience, it's not that much more work, and the end result generally looks a lot better than baking the jaw movement into each pose (depending on the animator, of course).

As far as inconsistencies go, I'm all for them, because that's what naturally happens when we speak. There's no way to really determine what type of jaw movement "normally" occurs when making certain sounds because no two shapes/sounds are exactly the same. There's so much blending of shapes going on when we speak that little inconsistencies are popping up all over the place. Sure it's possible for major inconsistencies to pop up in some cases, but that's partly what animation leads/directors are for. If you're doing a private piece, then you are your own animation lead/director, so part of your job is to maintain consistency in the character's performance.

Lip sync is all about syncing to sounds, not to words or even pieces of words. Because we're humans and not robots, the spoken sounds we make are not perfectly consistent, even when saying the same word with the same emotional delivery at the same volume, and it follows that the contortions that our lips, jaw, and tongue go through to create those sounds are also not going to be perfectly identical from one instance to another. I greatly perfer a system that allows me to mimic all those individual bits of movement, because I know that I'll get something far more lifelike and believeable than a small batch of pre-baked shapes for certain sounds. Even if you're on a tight time/money budget for a project, there are still ways to create a series of controls that keep the lips, jaw, and tongue independent of one another while still allowing for efficient facial animation production.

natsirk
08-16-2004, 10:17 AM
Yes i agree with whats been said
But if you can afford it i recommend things like Facial Studio :thumbsup:
i just thought i would mention it -
cya

kevinw
08-16-2004, 08:43 PM
Speaking from experience, it's not that much more work, and the end result generally looks a lot better than baking the jaw movement into each pose (depending on the animator, of course).

As far as inconsistencies go, I'm all for them, because that's what naturally happens when we speak. There's no way to really determine what type of jaw movement "normally" occurs when making certain sounds because no two shapes/sounds are exactly the same. There's so much blending of shapes going on when we speak that little inconsistencies are popping up all over the place. Sure it's possible for major inconsistencies to pop up in some cases, but that's partly what animation leads/directors are for. If you're doing a private piece, then you are your own animation lead/director, so part of your job is to maintain consistency in the character's performance.

Ok I'm confused. You're all for inconsistancy.... but we should all be responsible for maintaining consistancy?
I agree with you that there should be some variation in jaw and lip positions throughout a dialogue, but most of that can be achieved within the given jaw movement for each phoneme. If your blendshape for the letter "A" is set from 0-1 (1 being the most extreme) then sometimes, depending on the situation you may only need to set a value of 0.7 sometimes maybe 0.3 with a little bit of "E" or "0" mixed with it to get another variation in the lips. If you take all those variations and add another blend for just the jaw, now you have the ability to exaggerate even more when needed. But to simply have blenshapes that move the lips and then have to animate the jaw seperately everytime you want the mouth to open sounds like more work. maybe not tons of extra work for one or two scenes but over an entire movie it would become time consuming. I'm all for having full control to hit any position I want, but I think it's also very important to be efficient with your time. At leat when you have a deadline.

BTW: I also agree with your comments about syncing to sounds rather than the letters and words. That's one of the biggest mistakes I see less experienced animators make. This is when the man in the mirror becomes an animators best friend.

jschleifer
08-16-2004, 11:18 PM
I prefer not to have the jaw motion built into the phonemes. if I do, I ALWAYS end up counter-animating which causes way more of a headache than just keying it where I want.

JBarrett
08-17-2004, 02:51 AM
Ok I'm confused. You're all for inconsistancy.... but we should all be responsible for maintaining consistancy?
I agree with you that there should be some variation in jaw and lip positions throughout a dialogue, but most of that can be achieved within the given jaw movement for each phoneme. If your blendshape for the letter "A" is set from 0-1 (1 being the most extreme) then sometimes, depending on the situation you may only need to set a value of 0.7 sometimes maybe 0.3 with a little bit of "E" or "0" mixed with it to get another variation in the lips. If you take all those variations and add another blend for just the jaw, now you have the ability to exaggerate even more when needed. But to simply have blenshapes that move the lips and then have to animate the jaw seperately everytime you want the mouth to open sounds like more work. maybe not tons of extra work for one or two scenes but over an entire movie it would become time consuming. I'm all for having full control to hit any position I want, but I think it's also very important to be efficient with your time. At leat when you have a deadline.

BTW: I also agree with your comments about syncing to sounds rather than the letters and words. That's one of the biggest mistakes I see less experienced animators make. This is when the man in the mirror becomes an animators best friend.To clarify the (in)consistency issue, I'm all for little micro-inconsistencies, like minor variations in the overall shape created for an "e" sound. Huge inconsistencies -- like those caused by an animator who opens characters' jaws way too wide all the time, or by animators who form clearly improper shapes for certain sounds -- should be nipped in the bud.

Your example about how to achieve jaw and lip variation is interesting. I've done almost exactly as you described on previous projects, so I know that there is still some flexibility to be found in blending between these "whole mouth baked into a shape" controls. However, from my experience it involves almost the same amount of effort (perhaps even more) to blend between those whole-mouth controls to get a certain look as it would to sculpt that same look via a series of isolated jaw and lip controls. (My biggest pet peeve in this regard is with "L" shapes, which often have the jaw open pretty wide at the 100% point, where the tongue finally touches the roof of the mouth. What if the character is speaking softly? Using the "L" control at less than 100% means the tongue isn't where it's supposed to be, so the shape doesn't read accurately.)

A big problem comes in the editing stage. If I look at the result of my first pass on a line of speech after blending several whole-mouth controls, and I decide that the upper teeth are showing just a bit too much at a certain point, it's a pain trying to figure out which control has the greatest impact over the upper lip. I would much rather grab the "Upper Lip U/D" control at the proper point in time and tweak it.

The biggest reason that I prefer separate jaw, lip, and tongue controls, though, is that you can offset the movement between the different parts of the mouth if the controls are isolated, which is impossible using whole-mouth controls. By adding some subtle overlap here and there, especially between the jaw and lip movements, it feels much more organic and alive than when everything is moving all at the same time, similar to how offsetting various body parts feels much better than when the whole body is changing shape all at once.

I also agree that efficiency is important (see the end of my previous post). Depending on the project and the time/money budgeted for it, a fully comprehensive mouth rig with tons of isolated controls may not be feasible. I still believe, though, that a simplified version incorporating the same basic principles could be used in such instances, keeping production efficient while providing great flexibility at the same time.

eek
08-17-2004, 09:41 AM
Wow thankyou Justin and jason,

Its great to get input from you guys! So basically my rig doesnt rely on morphs or even shape morphs but rather a network of muscles. Each muscle is driven by expression and node to paths. Its pretty complex as each muscle as a hierachy to its neibourgh and an influence, like pulling on a sheet of rubber but keeps its volume.

The rig itself handles pretty much every expression( i need to add buchinator muscle) and currently has around 200 combination controls before eyes/eyelids etc etc.

It can also handle poses in different ways, you can extract numerical data and turn it into poses. You can move general controls ala corners of mouth, top lip, bottom etc etc. And you have finite controls being parts of lips/eyes/eyelids.

The rig ive previously used had around 300 controls and had a seperate jaw, and a 2 stage system.Basically you would animate the full character and just its jaw movement. Then this animation would be placed onto a hires character with full facials.

So my rig can handle both. If i do take out the jaw then, all mouth poses happen on a neutral jaw. Personally i feel that the jaw movement is so small in poses generally(other than ahh, but thats the jaw on itself) its not a problem to add it in, but ill do two versions.

The rig is nice in that you can animate all the talking i.e ahh, ooh, h--e---l---l--o, then subtlely tweak it making it smile, frown etc whilst keeping the mouth shapes working.

The rigs is on sliders atm, but how do guys like to work? With a gui in the face or sliders?

eek

kevinw
08-17-2004, 03:20 PM
Your example about how to achieve jaw and lip variation is interesting. I've done almost exactly as you described on previous projects, so I know that there is still some flexibility to be found in blending between these "whole mouth baked into a shape" controls. However, from my experience it involves almost the same amount of effort (perhaps even more) to blend between those whole-mouth controls to get a certain look as it would to sculpt that same look via a series of isolated jaw and lip controls. (My biggest pet peeve in this regard is with "L" shapes, which often have the jaw open pretty wide at the 100% point, where the tongue finally touches the roof of the mouth. What if the character is speaking softly? Using the "L" control at less than 100% means the tongue isn't where it's supposed to be, so the shape doesn't read accurately.)

Justin- First of all thanks for participating in a healthy discussion.

I agree with you example about the "L" shape. A seperate control for the tongue is a must. I've found that after animating a character for a couple scenes you get to know which combinations of certain shapes work well together. Then the "effort" doesn't seem to be much more at all.

I'd be interested to try animating a character with control like your talking about. Perhaps my problem is that every rig I've used with a seperate jaw/lip controls just wasn't configured properly. I always felt like I was lacking in certain shapes of the mouth and had to key 10 different things to get one position. It drove me insane.



The biggest reason that I prefer separate jaw, lip, and tongue controls, though, is that you can offset the movement between the different parts of the mouth if the controls are isolated, which is impossible using whole-mouth controls. By adding some subtle overlap here and there, especially between the jaw and lip movements, it feels much more organic and alive than when everything is moving all at the same time, similar to how offsetting various body parts feels much better than when the whole body is changing shape all at once.

Very good point.... The current characters I'm working with also have seperate controls for upper and lower lip U/D. When combined with the Phonemes I can get that same sort of overlap. However, once or twice I've run into a situation where the lip control just crinkles the lips and it looks like crap.



Jason- Thanks for the 2cents... Always good to here from you

Eek- Your rig sounds interesting. I'd like to give it a go sometime. Personally I like having a seperate window with sliders. This way I can move the window around, hide it if I want. If all of the controls can be easily selected on the face then that may work to but with over 200+ controls, wouldn't that get a little messy?
Please keep us posted..

eek
08-17-2004, 03:47 PM
the system has three layers, each getting more complex like a pyriamid at the top you just have poses, then middle general controls then at the base finite. Each will have its own window so you can dabble into each when you like. If the animations working nicely with just poses then fine, if you want to tweak the corner of the mouth, go into the mid controls, if you need more refinement then open the finite controls layer.


The system is also additive positive and negative, so you dont have to counter animate.

The bug ive got atm minute, is mouth intersecting the teeth, to get the lips flowing over when more than 2 poses are combined.

eek

JBarrett
08-17-2004, 07:09 PM
I've found that after animating a character for a couple scenes you get to know which combinations of certain shapes work well together. Then the "effort" doesn't seem to be much more at all.The same goes for a rig with discrete controls for different parts. It becomes second nature after a while.

I'd be interested to try animating a character with control like your talking about. Perhaps my problem is that every rig I've used with a seperate jaw/lip controls just wasn't configured properly. I always felt like I was lacking in certain shapes of the mouth and had to key 10 different things to get one position. It drove me insane.Sorry to hear it wasn't a good experience. It does takes some time and practice to get such a rig set up properly. A simple version, though, should be fairly easy to implement without too many headaches.

I made a rig of this type early last year just to test it out, and ended up using it for the "Animate a Face" tutorial CD-ROM set that was recently released. It's probably a low-to-intermediate level rig in terms of overall complexity, and the core mouth controls that are most often used are pretty sparse. In the lip sync demo, I'm guessing that it only took an average of three or four controls to get each shape, and they all blend together pretty well.

Eek: I'm a slider guy, as long as the sliders are well-organized. Setups where you're moving around circles inside squares on some hefty GUI thing drive me nuts (no offense to the developers of such rigs...it's just not my personal preference). Give me three objects -- one each for a list of mouth, eye, and brow controls -- and let me slide to my heart's content. :)

eek
08-17-2004, 10:14 PM
Hey Justin,

That cool, i was heading down the route of sliders anyway. Mixing circles thingy around just dont seems to gel. Hey Justin, Kevinw if youve got max would you like to be a beta tester?

Give me three objects -- one each for a list of mouth, eye, and brow controls -- and let me slide to my heart's content. :)
cool ill set it up like this:

3 sets: mouth, eyes, brow

then in each set have three layers:

Poses, main controls, and finite controls.

wadda you think?

eek

JBarrett
08-18-2004, 01:46 AM
Sorry, but I don't have Max. From what you described, though, it sounds like a good setup.

kevinw
08-18-2004, 05:43 PM
Yeah I've got a copy of Max 6. I'd love to check it out. email me at digitalwisdom@hotmail.com and we can discuss further.

whiterat
08-21-2004, 05:25 PM
excuse my ignorance (and my english)

if I have the jaw motion outside of the morph targets, how must do my morph targets, with the jaw open or close?

My guess is with the jaw close, am I right?


Miguel

JBarrett
08-21-2004, 05:57 PM
excuse my ignorance (and my english)

if I have the jaw motion outside of the morph targets, how must do my morph targets, with the jaw open or close?

My guess is with the jaw close, am I right?

MiguelCorrect. These targets should only deform the lips...pulling them up or down, making sneers, moving the corners, etc. By creating these targets with no jaw motion, they should blend well with the jaw motion once you start animating.

To give a little more detail on the jaw, the jaw-open pose should only deform the lips enough to get them into a natural "relaxed" state, as if the character were saying "ah." Be sure not to pull the lips up or down in relation to the teeth in this shape...just sculpt them into a pleasing round form for when the jaw is open.

In the rig I made, I created this jaw control using bones, with a primary bone that has influence over the jaw, lower teeth, most of the lower lip, and some surrounding skin, and is a parent of the tongue bones. I then added some secondary bones that had the same pivot point as the primary jaw bone, followed the primary bone with varying degrees of influence, and controlled the mouth corners and other parts of the lips (this could also be done via skin weighting to the primary jaw bone). I then tied the primary jaw bone's movement to three sliders: one for pivoting up/down, one for pivoting side to side, and one for sliding forward and backward. By rigging the jaw this way, there was no temptation to move the lips into some non-relaxed shape. I just played with the influence of the secondary bones to get natural deformation in the mouth as the jaw opened.

whiterat
08-21-2004, 10:23 PM
Thanks for the quick response and for the time to explain.

Now I have a clear picture of the matter.

Thanks!

mttjss
08-23-2004, 06:13 PM
If I may jump in briefly..

I am starting to animate my very first facial model. I have never even read into this.
I know Osipa has the book, but I could use some other advice and links as well.
I was thinking about a bone setup like the toad facial setup or with belndshapes.
I am using Maya and I am more into the cartoon/Disney/Pixar look.

any help?

I do understand the 3 layers with seperate controls for eyes, mouth and brow and then 3 levels of control for each..is this the best way or easiest way to go??

Didnt mean to jump in posts but I figured it would be better since most of you are all dealing with these different ways of facial setup.

thanks.

eek
08-24-2004, 10:10 AM
Im just saying from a production standpoint, the rig i used in the past had this similar three tier setup. Im trying to build a production setup for animators so they can easily understand and if they need, add more controls; crucially without interrupting there work. So it just goes back into the pipeline.

You dont need 3 levels of controls in each tier, im just doing it as my rig can support it. The main thing is a good perfomance, and generally you as much control as possible.

yep the 4 big setups to go down are: blend shapes, shape animation, bone setup and muscle setup.

Btw shape animation is different to blendshape. Essential your applying morph targets to shapes then skining the mesh to points on the shape as path constraints.(makes a nice setup! the hardes part is working out the shape network)

The cane toad rig is great for bones, Jasons/gollum is great for blendshapes, for shape animation look at "henrys garden" a fantastic rig!. Muscle setups - well research this one, once mine goes live ill post a site for you.

eek

mttjss
08-24-2004, 02:34 PM
thanks Eek -

I know for the face blendshapes are probably the simplest way to animate. I have heard of morph targets but is this for any 3D software?

I have looked into Cane Toad as well and must say that it seems like an interesting way of animating as well. I guess if I am going for a top notch animation reel, I need to have an idea of different ways of rigging and animating for a future job I land.

I will check out the animations and rigs you mentioned.
keep me posted on your muscle setup as well..that has always sounded exciting to me.
-basically you have skin to muscles to bone so when you move the joints the muscles deform..correct?

eek
08-24-2004, 02:52 PM
btw, blendshapes are morph targets(Its just what maya calls them.)

eek

CGTalk Moderation
01-18-2006, 06:00 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.