Thursday, April 23, 2009

Exaggerated naturalism (part 1)

In a comment on my post on Henson's Sid the Science Kid, reader Mike said...
I think it's a fun show and have watched the animation in detail trying to determine if it was mocap or taken from video reference.

Interesting. I think it's safe to say that the goal for many Cg animators is to achieve a kind of 'hyper-real' sense of motion. I'll call it exaggerated naturalism. It seems obvious that this has been the prevailing trend in CG films for the last 15 years. Some films have done a better job of hitting that bar. None has done it consistently, but most aim for it (although occasional exceptions do exist).

When I see the work of a Henson puppeteer transposed onto a CG character, I see this exaggerated naturalism nearly perfected. It is exaggerated because a puppeteer has a keen sense of how to stage and present a performance in a way that reads clearly, pushes things a bit and is entertaining to watch. But due to the fact that it is mo-cap all the elusive goodies of natural motion -- things like barely noticeable weight shifts, subtle footwork, living holds, micro gestures, a sense of muscularity in action, a sense of physical presence (ie: weight), etc.-- these all come along for free. Meanwhile, to achieve something that even comes close to this many animators are relying on working from video reference of themselves acting out the scene. The basis of the performance and motion are highly derived from the video source material (we'll take up the artistic flaws of this approach in another post). But even if the animator is skilled enough to capture the nuances of the motion convincingly (based on a lot of scenes I see in films that is a mighty big "if"), it is an extremely time intensive task. This poses an interesting question: If the end result of animation taken from video reference is similar to well performed motion-capture, then why have animators animate from video reference? Why not just hire seasoned, trained stage performers to do the job in suits? The goal is naturalistic motion, right? (or an exaggerated variant thereof). Not only is it faster, but the physical performer can do many takes before lunch, providing the director with choices as well as the opportunity to improve on the performance quickly. Contrast this against having a highly skilled hand key CG animator do one take of similar quality (if they're really good) in about a week. Or longer, as is often the case. And I think it's a valid argument to say that if exaggerated naturalism is indeed the goal, then the resulting motion from the puppeteer is qualitatively superior. I've seen many a scene in high budget CG animated films from the biggest studios that would have been MUCH better had they been mo-capped by a Henson body puppeteer. Seriously. The goal of the scenes was to have a human(ish) character move in a slightly exaggerated, yet still highly naturalistic manner, doing some physical action. In every CG film there are any number of scenes where the animator was not able to meet that goal half as well as a Henson puppeteer could have. And the times they did match that goal with equal success, they did so at 50-100x the time investment.

In the past a major argument against mo-cap was that it didn't match stylistically to the types of motion and performances needed in animation. There was a time when that was a valid argument. But life is never static. As time has passed animators themselves have kept creeping ever closer to the very style of motion that mo-cap actually does better! Thus when you consider the continued improvements in mo-cap technology, plus improvements in understanding what kind of performance is needed for mo-cap to work, combined with mainstream CG animators' blind march toward an exaggerated naturalistic style of motion- that 'style-gap' has narrowed. Before much longer it will have narrowed to the point of insignificance.

This is why I think the current popular trend for exaggerated naturalism among CG animators is creatively a dead end street. Where you want to end up, a puppeteer already lives. The technology will only get better and the physical performers will only become more savvy as to how to use the medium. Time will not march backward just to appease our predilections. The question you might ask yourself is this: Do I want to be relevant as a performance artist, or do I just want to work on Hollywood 'animated' films? I firmly believe that if a CG animator wants to remain relevant as a performance artist in the years to come (relevancy as a performance artist is not the same as being employable in the animation business, by the way) they're going to need to develop a style of animation that cannot be achieved any other way than to hand key it. It may sound crazy now, but the time will come when I really do believe I will be proven correct.


JHN said...

We are a small shop that both use hand key animation and motion capture. One thing to mention is that a mocap session usually is quite expensive (at least a mocap operator, a studio manager and sometimes a director and offcourse a performer) if you don't have a mocap setup in house, then keyframing is still a good way to save a couple of thousand dollar.
But when a production gets bigger then there's a break even point which favors mocap very fast in our experience.
To comment on your conclusion I can only agree and with that I think mocap is going to stay for some time to come. That's why we are not betting on a single technique, we try to make the handkey animation more zany and the mocap more stylized. As that seems to be pleasing our clients most.

Anonymous said...

Since WHEN is "exaggerated naturalism" the goal of animation?

There's so much more than animation can bring us than convincingly-moving biped characters of human proportions. I mean, COME ON.

And Sid the Science Kid can NOT be considered animation. I think Henson himself calls it "digital puppetry". The difference in semantics is significant if you stop to think about it.

Josh said...

Good post Keith. I'm pretty sure that it is indeed "performance capture" as I saw an article on the show in a magazine (think it was popular science of all things). One thing that would make it tend to work more would be that the suits the captured in forced the actors to the proportions of their characters - almost like wearing a mascot costume at Disneyworld or something.

There is definite strength there in terms of certain performances however, you couldn't put someone in a mickey mouse costume and and expect to capture a mickey quality performance, people just don't move that way.

However, when you're trying to create a game or movie with realistic human proportioned characters doing human looking things, mocap makes a lot of sense. A big if on that is if the mocap is planned for and casted well. When I was working in games, too often people saw it is as a quick way to get things done and let the motion capture animators figure out how to get two segments to blend together well or to get that 190lbs guy's motion you captured to work on a 110lbs female...

It's gonna be interesting to see how the "performance capture" debate continues.:)

Keith Lango said...

@Anon: I didn't say that the only goal of animation was exaggerated naturalism. I said the predominant apprarent goal for much of CG feature film animation is exaggerated naturalism. If it weren't then video reference would not be such a popular technique for animators. Anybody who watches films like Bee Movie, the Shreks, The Incredibles, Ratatouille, Over the Hedge, Monsters Inc., Toy Story 2, etc, etc.. - anybody who watches these films sees that there are many, many scenes in those movies where the goal is to get very humanly proportioned characters to move in very convincingly natural ways. Not every scene, for sure. But a lot. To deny this is to be an idealogue and we'd not be intellectually honest with ourselves. So if the goal is to get those characters in those moments to move in an exaggeratedly natural manner, hand keying might not be the best approach for those moments. Not ALL moments in every film. Just a lot of them in many films. In some cases more than half of them.

You are correct, the Sid stuff isn't 'animation'. I don't think I said it was- I'm not that stupid. :) But nobody but animators cares, really. General audiences don't care. However the Sid stuff is CG rendered. What have been commonly called CG animated films are starting to become CG rendered films. There are already many Cg rendered films that have met commercial success that employ the mix of mo-cap and hand key. Several of them have been nominated for Academy Awards, and even won an Oscar. Thus looking at all CG rendered projects that have a certain type of motion as a style, then it makes perfect sense to look at the results of different methods of creating that motion.

The point I'm trying to make is that if the goal of a scene or an entire CG rendered film/show is to have exaggerated natural motion, then it makes sense to look at how to best achieve that style of motion.

Keith Lango said...

You're correct. Mickey Mouse, when done properly, cannot be done using mo-cap. The style of motion that applies to a character like Mickey Mouse, in general, is not what I'd call exaggerated naturalism. I'd call it highly exaggerated performance. It's a style of motion that can only be hand crafted. Very, very little about the way classic cartoons move can be naturalistic (even if you make CG versions of them). In those instances it makes absolutely no sense to try and use a mo-cap method for generating the motion.

Up til now casting for mo-cap has been very slap-dash and sloppy and poorly thought out. I accepted that as the status quo until I saw the Henson stuff and then I realized that people are waking up and they are starting to really think about how they cast and direct for mo-cap.

I don't see this as a continuation of the old tired debate. I'm trying to have an honest look at the final Cg rendered product on screen, and from there seeing how different methods for creating it are either more or less successful. It's not as cut and dry as some folks would like it to be.

bclark said...

Great read Keith, glad to see it written up that way.

For any one thinking that there still is a debate, look around at the production world and you will see that the only people still "debating" are the ones that are in denial. Mocap is just about every where and being used well.

It is a good tool when used correctly, you don't use a hammer on screws or a screw driver on nails... and you don't give either to a person who has no idea what they are or how to use them safely.

Yet this is what happens on project after project. People useing the wrong tools for the wrong reasons with out any idea how best to get the correct end result-something that is entertaining to watch and enjoyable for the audience.

ILM, WETA, Henson, get it right and have for a long time now.

Wonkey the Monkey said...

I think you've hit the nail on the head. The only reason animators still have jobs is because nobody else can do what we can do. To future-proof our careers, we have to focus on developing new things that only we can do instead of polishing our soon-to-be-irrelevant skills.

I think this kind of "do the things that nobody does better" philosophy applies to all kinds of fields, especially in the arts. Stage plays are best when they aren't trying to be movies. Radio is best when it isn't trying to be television. Paintings are (arguably) better when they aren't trying to be photographs. If more people would observe these boundaries, crossing them only when there was a purposeful reason for doing so, there would be fewer creative dead ends in this world.

I have a lot of respect for what you do because you don't just write about the problem. You're actively anticipating the future of animation and adapting your skills in preparation for it. Keep up the good work, and please keep documenting your progress so we can all be inspired.

Anonymous said...

even if the mocap systems would be cheaper as they are now.. I don't think it will ever be better than a hand-keyed animated shot.. a well acting shot...even if the mocap performer could make tons of shots/takes very fast to get the director's aproval.. it wouldn't achieve the subtext, the right acting choices, appealing and good handling of shapes on screen that a good handkeyed animator could achieve... I often see in live-action films/series.. 'bad animated' actors/actresses, with those 'exagerated realism' on their motion(of course cuz they are human beings) but not convincing or entertaining enough to the audience..

Anonymous said...

JOE here . . .

I think you're right to say that exaggerated naturalism will end up in the domain of mocap.

However, I think another thing to keep in mind is that exaggerated naturalism will still be done by keyframers in the future with regards to animal animation. For creatures with long necks or long legs or wings, it is unlikely that they can get a good mocaped performance out of a critter, so keyframers will still have to do that, AND if exaggerated naturalism is the style for that picture - they'll have to keyframe it like that.

Anonymous said...

Aaron Here-

So my question is how to adapt? I'm a character animation student at a certain online school right now, and for our assignments we are shooting ref video etc. I guess according to your thoughts, this could be a dead end. So what should we be looking to do? Maybe we should embrace Mo-Cap? I mean a trained character animators eye would be a bonus for direction and input for Mo-Cap right? Anyways, I guess this post got me a little freaked out, because I'm so new to this field that seems to change all the time.

Anonymous said...

Matthew Here-

Aaron I know how difficult it can be to be studying animation theory and principles when a new technology comes along and "outdates" what you're learning. But, being an animator in this day and age there is no better way to stay afloat then to have as many tools in your toolbox as possible. And learning the mechanics of motion by filming yourself or watching other video reference and then exagerating that motion with animation is a very invaluable tool indeed.

Justin said...

Interesting read but I find it problematic for some reason. I don't understand how you talk about that stuff divorced from all the other areas of a production. If the writing, direction, cinematography, sound, visuals etc. etc. in a production are going for "exaggerated naturalism" then shouldn't the style of the animated performance follow suit? I guess I wonder if the article would work better if it wasn't a warning to animators to examine their goals of what kind of style they want to be known for; but instead, was a plea to writers to come up with more appropriate stories to tell with animation and for executive producers to have guts to greenlight those projects.

Presto would be a great example of what is not "exaggerated naturalism" if the surfaces, lighting and rendering were consistently "supernatural" or "impressionistic."
I was trying to think of a good example that was feature length..."Cloudy With A Chance of Meatballs"? Might be better though there's still some inconsistency in the trailer.

Anonymous said...

Aaron here again-

Keith, I just wanted to say you rock man!!! I'm glad you post whats on your mind, I always find your site interesting :) Thanks Man!