Friday, April 24, 2009

Exaggerated naturalism (part 2)

A pretty good example of what I call exaggerated naturalism in motion (excluding facial animation)....

An example of motion that is not what I'd call exaggerated naturalism...

And an example of the two mixed together... (hint: Collette yes, Linguini no)


Andreas said...

Hey Keith,

Well let me start by saying thanks for all the juicy info you post each time in your blog.

I think in your post about "Exaggerated naturalism" you are making some good points however i think that even in mocap although some of the things you mentioned like "weight shifts, subtle footwork, living holds, micro gestures, a sense of muscularity in action, a sense of physical presence (ie: weight)" come for free* you still need a lot of work in order to make them feel natural but of course it still takes less time to create mocap animation than actually hand key animation.

I think "Happy feet" was a great film and it was mocap but you couldn't understand at all if it was mocap or hand key(at least i didnt) but when it comes to recreating reality especially if you dealing with realistic biped characters like Beowulf which i found awful (no disrespect to anyone just my opinion) mocap simply doesnt work and no matter how cheaper that is the audience will feel that something is not right.

Also another thing i dont get about the people,directors,producers who wanna make mocap is why have a realistic performance rendered in 3d so you can have realistic 3d animation and not just shoot live especially when the strength of animation is that the only limit is your imagination and that you are able to create stuff that you cant simply replicate when you shoot live. (ok i get it i just dont want to admit it, less money).

One last thing i wanna mention is that on Henson's Sid the Science Kid although the animation is super and i can see all the small moves of the hips, the fingers, the body mechanics working perfectly i still dont get the same emotion (just look at his eyes) iam getting when i see Pixar films(i usually cry)or Miyazaki films. Iam not sure why this is happening but one guess would be that the human brain can adsorb so much information that all those micro movements or perfect body mechanics doesnt matter as long as you can have the sense of realism and they are not completely off.

Lets not forget after all that all storytelling/filmmaking is just an illusion it doesnt matter if it cartoon or live its just an illusion.

P.S cant believe i wrote all that :D

*I have to disagree about muscularity though, since mocap captures external points in the body so no s&s is being captured or no sense of s&s and thats why in my opinion mocap looks unnatural when it comes to biped characters.

Alonso said...
this picture shows how much the animator's had to alter the mocap data to fit Gollum's body, looks like probably the data was just used as reference and the animator did it all from scratch, with Serkis getting all the credit :P (of course it's possible that these two images are not from the same point in the film)

As you say the Henson people are amazing performers, but their skills are weighted towards getting expressive bodies because their faces have traditionally been where they have less control (all though on characters like Hoggle they have a team of pupetters working on the character, 1 in the body, and 2 or so on the face remotely)

As Josh pointed out, motion capture data takes manipulation if the character doesn't match the actors body, it helps if you alter the actor like the muppeteers did with those big feet in the video, but still as with the Serkis example the further the difference the more animator's are needed to alter it. I think the Mickey Mouse example is excellent, an actor in a turtle suit is not going to move the same as an animator would make Vern from Over the Hedge move.

Richard Lico (game animator) has an example of some mocap data he cleaned up on his reel ( Which shows what a large quantity he has to add. The muppet stuff looks a little more clean in capturing, and it's just a matter of time. But I wonder what stage the technology is at right now, it's hard to tell because the PR machines keep touting it as perfect and animator's as just button pushers. But then you hear from animator's stories like Zemekis liking a shot on Monster House and asking "why can't all our shots look like that?" and him being told that that shot was all hand keyed.

Currently I'm big into Blue Sky's Horton. A lot of that stuff is just a celebration of movement possible only in animation. But having such an exaggerated motion style makes me wonder if it's possible to also have close quiet moments of real pathos like you find in the Incredibles. And that's the real advantage of Incredbiles and Ratatouille, the ability to portray deep feelings and thoughts. Probably the right story could mix both styles, but it's hard for me right now to imagine.

Anyway, I agree that Ratatouille is not taking full advantage of the animation medium. But at the same time, if it were mocapped, the actors may not synch up the body to allow an animator to put the strong facial work on top. The muppeteers do it, but it takes practice and rehearsing, and I wonder if the pose and attention are stronger when it comes from a single mind (1 animator vs 3 pupeteers).

(man you're a posting machine today :)

Keith Lango said...

@Andreas: Thanks for adding your thoughts. When I speak of muscularity I'm not referring to deformations. Rather I am talking about the sense that a character is being moved from within by muscle effort, rather than being manipulated externally- which is a very common vibe one gets from Cg animation.

Keith Lango said...

@Alonso: good thoughts. One thing about re-mapping of motion-- this is definitely a current focus for the software developers. They're also looking into filtering the motion, applying different interpretive style's for the data- kinda like a Photoshop filter for the motion.

As for facial animation I don't think mop-cap looks good at all with that yet. Even the Henson stuff, while intriguing, leaves a lot to be desired. That stuff is still best handled by hand right now in my opinion.

Regarding the need for 'animators' to work on mo-cap data, this is where I fall in the camp that you are more of a motion technician than an animator. To me animation means bringing things to life- the soul of it, the source of it. When working with data the soul belongs in someone else's work. You're just the assistant cleaning up the work, much like assistants in 2d clean up and tie down the key animator's drawings. No 2d assistant would pretend they were creating the soul of the performance.

But again this isn't really about mo-cap good or bad or the various applications and tasks involved. It's about recognizing a particular style of motion in CG rendered films and asking some open and honest questions as to what method might be the best one to use to get that desired motion style. I am open to the notion that perhaps good motion performers can deliver that style better than an animator copying video reference by hand. This isn't even taking into account the performance integrity issues, which are not insignificant.

Sunny Kharbanda said...

Nice choice of clips.

In the second clip, I definitely enjoyed Linguini's part more than Colette's (especially the beginning, before she enters). Some of my favorite moments in Ratatouille were the scenes where Remy works Linguini like a puppet. This scene clearly shows why. The other characters move just like regular people. Where's the fun in that?

It's not just CG movies, though. I had a similar thought while watching a bit of Aladdin yesterday. The animation of the main character, while technically great, isn't all that delightful. The movement of the Genie is what I loved about that movie. It's cartoon animation, doing what only cartoon animation can do.

Of course, your mo-cap argument only applies to CG, but would we be at this point if Walt Disney's obsession with realism hadn't shaped the animation industry since the 30's?

Raf Anzovin said...

Interesting post Keith! I find myself both agreeing and disagreeing with you. "Exaggerated Naturalism" is as good a term as any to describe the most common approach to feature animation today, though I think there are counter-trends too (more on that in a moment). And I'm very much on board with your assertion, made here and in other posts, that there are more interesting effects to be had in other directions.

But I'm actually not quite convinced that the particular kind of exaggerated naturalism present in these hand-keyed clips could necessarily be so easily achieved by puppetry/mocap, even though I agree that there are certainly similarities between them and what Henson is doing. There are subtle qualitative differences present that I think still make the hand-keyed work stand out--a certain clarity of staging, a highly specific sort of movement that seems carefully "tuned" for it's aesthetic effect. I have not (yet) seen this particular feel replicated through puppetry in any context (which is not to say that puppetry doesn't have many other virtues, which you have enumerated in your previous post).

Could further development allow for this sort of feel to be added as a post-processed "filter" to the motion? I'm not sure. I guess one shouldn't be quick to say "never" (cause it usually makes you look silly later) but it seems to me that this feel is dependent on the specific pose choices, not something that it's easy to develop rules for.

Of course, to some degree this is beside the point. Because the real question is: would Ratatouille have been diminished as a film if some of the motion had been created by mocap puppeteers? I think that there are some enjoyable qualities that might be lost, but I'm willing to accept that those qualities would be dwarfed by the overall larger question of the quality of the film. Ratatouille was good, and it would still have been good if it hadn't been 100% hand keyed. So perhaps it doesn't matter.

That said, I've been happy to note that there are films that really seem to be breaking that mold. I have to concur with Alanso that Blue Sky's Horton really did it well. I loved Wall-E for it's storytelling and direction, but the animation itself wasn't something that stood out much. I'm sure that was intentional, and is really an example of how well Pixar achieved it's goals of involving me in the story through their technique without drawing attention to it. On the other hand, I loved Horton for the pure joy of its animation, a completely different kind of entertainment. Audiences clearly do enjoy those qualities, so I don't think that their existence within the feature world is threatened. And if the addition of puppeteered mocap means we might get to see more quality CG films, I'm not going to complain.


Alonso said...
This comment has been removed by the author.
billburgNYC said...

Wow, Keith, I forget to check your blog for one week, and I come back to find all this great stuff to read!

I think Sunny makes an excellent point above that exaggerated naturalism is also frequently found in 2D animation. For a time in the 90's, that style seemed to dominate the way it does in CG today. Consider the animation in Pocohontas, Road to El Dorado, Atlantis, Sindbad etc. In the best examples, the influence of the live-action reference is muted by skillful drawing and animation. But in my opinion, many of these films gained little by being animated rather that shot in live-action (in terms of character performance).

The trend toward exaggerated naturalism seems like it goes back at least as long as the time when Disney made the distinction between "straight drawing" of Bambi and the "caricature" of Dumbo (the latter of which he seemed to regard as inferior).

Great post; great discussion!

Dimos said...

Just to comment on the two posts on "exaggerated realism".

Mo-Cap or understanding how to use it does not make one "relevant" in today's industry. Understanding motion, acting and performance in terms of "believability" and having the skills to use interpretation to depict a beautiful performance is. Mo-Cap is great and wonderful when used as a tool for certain application but calling it a style doesn't make it so.

I agree with Raf on when he discusses about the ability for Mo-Cap to pick up the subtleties in motion that an animator can more appropriately apply when hand keying. It takes a good animator using his/her skills of interpretation to actually portray a believable action or emotion. That is something an actor in a Mo-Cap jumpsuit can not get no matter how hard they try. There will always be a need for a skilled animator that understands motion to apply those details.


Keith Lango said...

Dimos-- Motion can have many styles. There are signature attributes that can be identified from one style to another. The point I've been hinting at- which a lot of folks seem to miss- is that the signature differences between the hand keyed CG style of exaggerated realism and well performed mo-cap are not as significant or distinct as many assume them to be. If the currently dominant goal of exaggerated naturalism for CG hand key animators continues on the trajectory it has been on for the last 10 years those differences will continue to blur. This is because hand key CG animators keep trying to make their stuff look like real life motion (which is a necessity due to the overall naturalistic visual language of the films they are employed in making). I propose a time will come when these differences and distinctions in the flavor of the motion will blur almost to the point of insignificance. At that point there will be little reason to pay someone to hand key from video reference something that looks (for all practical purposes) indistinguishable from well performed mo-cap. This is not a re-hash of the old mo-cap/hand key debate. This is looking at the trajectories of the evolution of both methods and realizing that there will come a point of intersection some day- from both sides. When that day comes one technique will become practically redundant.

You are indeed correct- there will always be a place for those who understand motion. However to me that is not the essence of being an animator. Being a motion technician is not why I got into this biz (although for many that seems to be sufficient motivation). Instead, creating believable characters who come to life on screen from imagination- that to me is the core essence of animation. Motion styles are mere servants to that end. When a film heavily relies on mo-cap, yes somebody will still be needed to plus it or tweak it. However they are not acting as animators. They are motion technicians. The core of the performance- the inspiration, the creation, bringing the character to life- that belongs to the mo-cap performer. Take that away from animation and I say you take away the soul of being an animator.

Keith Lango said...

One point of clarification-- when I wrote "When that day comes one technique will become practically redundant." I meant that one technique will become redundant in accomplishing the specific & popular motion style of exaggerated naturalism. For other styles of motion where the distinctions from mo-cap are so great that it is practically impossible to eliminate them (Tex Avery style cartoon motion or Aardman stop-motion, for example) then there indeed will continue to be a need to employ good animators who can animate in those styles. However most CG animators these days are not focusing on learning those styles- for a variety of reasons.

Dimos said...

In that case then, it all makes sense. And... I welcome the day when Mo-Cap does not need animators to tweak the data. That way we can all just animate happily away on other projects that really need that attention to character. ;-)

Oh, and for the record... understanding motion (and how to use it more effectively) is one the underlying skills needed to help create those believable emotional performances. A character animator should not want to just create motion or movement but rather a performance that is believable and strikes an emotional connection in the audience(s). That wont necessarily happen unless the animator has a good grasp on fundamental movement or mechanics. I'd rather strive for plain and simple "believability" and not worry so much about exaggerated reality or what ever it is. All styles strive for that anyhow. No matter if it's cartoony or "real".

That all being said, I think then I agree with you but was a little lost in what was being said. I blame that on you being smart with words and me being "me".

Thanks for taking the time. All the best. Do you mind if I link ya on my blog/site?


sarah said...

Exaggerated naturalism sometimes divert the attention of viewers and i have noticed that it bring a humorous look , even if the animation is serious.

Anonymous said...

I've always thought that Linguini writing in the notebook "Always do something unexpected" was a horribly cliche way of showing us and Colette what he's thinking.