Sunday, July 11, 2010

Realism in Animation

This post is slanted towards realistic character animation. Realism in animation is incredibly convoluted, but the general trend as well as I've been able to figure it out, seems to be this: In order for your character to be believable once it delves heavily into precise realistic detail, you have to have to start dialing back the realism in the way the character moves. If the character is less heavily rendered though, like in hand painted forms of animation realism, you can actually get away with a more realistic design approach, but have to keep the character's actions theatrical to adjust to the limitations. These distinctions alone wouldn't mean much, but I had to reconcile yet a third, more thematic, storytelling and character related approach. Both of these approaches mean that in order to have a character that looks most realistic, you're best dialing further back from visual realism in order to have characters that behave realistically in a more human sense.

For animation realism, I'd like to distinguish between full visual realism, strict realism, and what I consider pushed realism. Full visual realism is the complete commitment to portraying life as accurately as possible, a feat which has not advanced very far yet, as Beowulf and Final Fantasy show. Strict animation realism is a feat which is not so much suited for hand drawn animation as hand painted animation, I'd say. It's just about portraying what are generally realistic forms and realistic movement, though with the limitations of confining technique that keep the possibility of full on realistic detail and nuance remote. Pushed realism is more about visual styles which, while possessing a very high amount of realistic observation, aren't so much concerned with depicting real characters, but more or less, heavily stylized characters with a bent towards realism in the way they behave.

It's tough to be too specific on what is the most detailed work of art or what detail means, but I'd say that there's a general trend towards detailed works demanding a certain amount of realism simply by the observation required. In 2D forms of animation, there seems to be a general dichotomy in animation, though rarely approached too straightforwardly, between emphasizing drawing with little movement and vice versa. Sure, there's plenty of people using limited animation with very simple visuals, UPA and Zagreb for example, but I think that this is not too effective in the long term because works simply start to look alike. This is only realism of course, in a superficial visual way, but the more details you add, the more you end up looking like one realistic thing or another. But why do they do this? Useful detail. Each approach gives you enough useful detail to plant a story or general visual impression on the viewer while leaving them in the abstract in another. But what if you have both heavily rendered artwork and 24 fps motion?

It's by far the most detailed approach, but it uses up it's creative gases pretty quickly. What you start ending up with is a lot of works which are gorgeously detailed but look very similar because you're beginning to run out of the artistic choices that are essentially premade. It's like a one player card game you start playing by drawing a 45 card hand from a 52 card deck. There's not much variety to be had if everybody starts playing this card game.

Here are what I consider to be three of the best examples of strict animation realism. There's a clip from Heroic Times, Hercules Visits Admetus, and The Old Man and the Sea.

Heroic Times has some pretty exaggerated motions at times, though due to the technique, it's often hard to tell for subtle movements. Some of these jousts are not just unlikely, but the way the riders get thrown off is clearly exaggerated. The Knight's facial expression is certainly a bit exaggerated before he puts on his helmet. Some moments seem less exaggerated in animation than in lighting, like the loving embrace towards the end of the clip.

Anatoliy Petrov's Hercules visiting Admetus is an exercise in the most full blown realistic visuals I've seen in animation. The short uses Petrov's technique of photographica, which involves a layer for lighting effects superimposed over cel images for a greater emphasis on light and shade on characters than if there were only one layer. There are quite a number of glaring flaws, and I think that the film frankly works better when the viewer isn't given the opportunity to see the visuals work alone. That's not to say that I think that they're irredeemably bad, but that I think that there are some glaring flaws. There's some awkward repeated motions, shots where characters come off as haloed rubber mannequins, and frankly blatant shortcomings in character movement which slant toward flat planes. There are, like in many other films which are heavy on realism, everything from dark shots of characters to the trademarked eyeballs on static heads, and hair blowing on static heads, tricks you'll see plenty of in anime. But when you see all the personality which the characters convey when animated properly, the technique is revealed to be as legitimate as any other. The backgrounds are more oriented towards visual richness than complete realism, and the characters are certainly stylized by the way they're lit. There are some especially mythological characters, and some of them are a bit exaggerated, but for the most part, they're rather grounded.

Alexander Petrov's The Old Man and the Sea, is a different form of painted animation, paint on glass. This film isn't as faithful an exercise in straight realism as the other two are, a much more impressionistic approach. There are some points where Petrov overused his reference footage, for the sea especially, and the boat's movement, but regardless of these concessions, it's a stunning work of art. The characters are of course, distant, due to Petrov's rendering, but they are intriguing in silhouette and their hinted personalities work well with the sea setting. This film seems to be much more true to real human motion than Heroic Times, but less so than Hercules, simply because it's not as realistically rendered.

These don't show what I'd consider strict realism perfectly, but artists of course don't make their work for poor amateur theorists like me.

Now onto the twelve principles. Some of the classical principles of hand drawn animation aren't as cut and dry as they seem.

A majority of the principles, like arcs and overlapping action are sound, but there's a few that demand further examination. Always be wary of the example that proves the principle and ask why the principle needs to be there at all.

Here are the ones I'm skeptical of:
1. Squash and stretch
Squashing and stretching is a perfectly valid principle, but it's also a fairly subjective observation. You could make up a principle and call it 'warping and folding'. Squash and stretch is supposed to be valuable for allowing fluid motion, but the observation of 'warping and folding' can allow for fluid motion as well. Folding doesn't just have to apply to pieces of paper, as you fold a sleeping mattress but think of that more as a three dimensional form. When you fold a sleeping mattress, you inevitably squash and stretch the material. When an object's shape is warped, you can say that part of it gets squashed or stretched, but that's not the emphasis.

2. Exaggeration. So since you can't capture life perfectly in animation, you're obliged to consider something animated to exaggerative by comparison? Just because an animation comes off as exaggerated compared directly to real life doesn't mean it intends to be exaggerative on its own terms. And the terms 'caricature' and 'comedic' seem to have an obscure relationship that isn't fully described.

The fact of the matter is that most cinema can be seen as exaggerated from one angle or another. Even many films which are labeled as serious are often light headed or have subtle hints of comedy to lighten the mood. I can't claim that I know exactly what makes a film serious or not, but many more seriously intended animated films look ridiculous when compared to one another because they vary such widely in their content. Now here's where I'm going to talk about serious animation, because this is a subject that I've rarely seen brought up at face.

But I still haven't addressed the major point for serious and more particularly realistic animation. You can't make an animated character that emotes as effectively as their live action counterpart. No you can't, in terms of making a character with the same amount of nuanced expression, but this point can be referred to an interview with Satoshi Kon. He talks about how there's less detail in each frame, and how this guides your eye quicker, allowing you to cut faster. So by this interpretation, if you had a scene in animation compared to one in live action, you'd simply go through the scene quicker because you understand it faster visually. Essentially, you show somebody less so that they see more. This is the justification for making realistic animation of a wide variety of sorts, some more in visuals, others more in theme, like Everybody Rides the carousel.

But there's other ways that animators have come up with more serious animation. One way is simply to cut down on character expression, emphasizing the scene more than the character. This might be considered by those who subscribe the idea of caricature as a sort of 'under caricature' depending on how its done, whether it's getting rid of character's faces or deliberately under-emphasizing their emotions. After all, exaggeration goes both ways. This is a point that doesn't seem quite clear, whether it's abstraction or caricature when you start deliberately diminishing a character's presence by doing things like getting rid of prominent facial features. I'd say that this is an area that's up for contention. Other not necessarily mutually exclusive ways seem to be going for a more theatrical approach to character acting or aiming for more symbolic expressive characters. Sometimes the solution seems to be to take what seem to be more cartoony designs and just avoid funny expressions, taking advantage of their especially direct emotional appeal.

I think that the works above have shown that you can use painting to suggest forms clearly enough to bring out what resembles more of a theatrical real person than a theatrical character exaggerated from a real person.

3. Then there's the means of animating, the principles of straight ahead and pose to pose. Straight ahead I'd say is a pretty misleading term, since you can animate something from backwards to forwards or forwards to backwards. Straight behind animation is conspicuously ignored in most mentions of the principle, however, with the wording towards 'first' drawing in the scene. Why not start with the last drawing and work your way backwards to the first drawing? Maybe it's supposed to go without saying, but does it? I can deduce logically that in cutout animation, this order could be pretty important. You're animating with paper that has little detail, and you want to show a simple ball rolling forwards towards the 'camera' in a shot. If you animate straight ahead, it takes more paper to animate that action than it does if you animate it straight behind, because if you're animating straight behind, you can cut smaller balls out of the larger ball. These principles are of course, related to hand drawn animation primarily, but I still think that they need to be examined and considered.

I've brought up all my concerns with the 12 basic principles of animation because they're surely going to come up in the likely event that somebody should argue this post with a different point of view on realistic animation.

Since I've explained that animation realism is not cut and dry and cannot be construed directly, and I've clarified that it's a continuum of approaches I'm going to go a bit deeper into how it can be expressive in the animation medium. Keep in mind that this subject leans towards visual realism, not thematic realism.

Those stricter realistic animation films I showed, while artistically brilliant, don't make as full use of the more superficial aspects of visual realism as some more stylized films. One short film I want to use as example is a Soviet propaganda film, Shareholders.

The film's major characters lean towards realism, but are clearly caricatured. They exist in a stylized satirical USA. More minor characters have various levels of abstraction and the short gleefully exploits the skewed widescreen angles and limited color schemes. The characters exploit animation with their blatantly exaggerated facial features and gestures. Just look at some of character's eyes and the shape of Pearson's snout. The contrast between the more realistically rendered characters and the often simple and/or abstract backgrounds gives the short some wondrous beauty. Background characters are often flattened and silhouetted to stand out from the major characters. The characters may have form that's heavy based in realism, but they're still quite flatly colored.

There's all sorts of exaggerations here, characters jumping unrealistically high, running through cityscapes that change shape, and a skeleton with exaggerated eye cavities to emote like a human being.

I hope that by showing you this short, you see some of the things you can do with more superficially realistic animation.

You can still exaggerate the way people move, even if they look realistic. Just look at the man's jump near the end of the auto race. You could probably pull off an exaggerated jump even with a more realistic looking character. There's not so much to do with squash and stretch or other distortions, but there's still quite a lot to do with motion blur. Don't forget about growing and shrinking either.

I'll just list things that can be captured better in realistic animation than live action? For this argument, I won't be very lenient for effects. If it takes animation to do that one thing, no point in turning a fox into a fur coat.
1. Aging. You can have the character look right without being forced to one person for the part. Not everybody has the patience of Richard Linklater to shoot a film over 12 years and pray that none of his actors die.
2. Serendipity. You learn so much by trying to emulate life that you wouldn't know by just filming it, and get to use it to your advantage.
3. Selection. In some ways you can just be more selective with what you portray in animation. In the 2d forms of animation, you often have to sacrifice camera angles, but you get to put everything in the film your self. Some things you just aren't likely to pull off in live action regardless of the number of takes.
4. Don't forget about realism beyond the limitations of human sense. There's frequencies beyond our hearing, electromagnetic radiation beyond visible light, slow and fast speeds, and sizes below the human threshold of sight. A realistic portrayal here might be more valuable as these are events that are not ordinarily experienced. Call it the National Geographic effect, I guess. In a way, it's the same reason we read that sort of magazine or watch nature documentaries.
5. First person perspective. There are many things to do with first person perspective in various sorts of animation which can't be done in live action. Just look at the short film The Fly for example.
There's plenty more, but that's all I care to type right now.

Animation realism

Here's the best way I can think of to describe how stricter forms of realistic animation uses the medium. Here's what one person using this technique might say if they dared: "I can't capture real life perfectly in animation, but I can give it a different spin of imperfection at any point in time, and I can use such imperfections to justify the use of the medium in spite of the fact that it is negligible to the realism of my portrayal. I trust that the people in the audience will be able to appreciate this peculiar distinction better than animators who are, in this particular situation, bored geologists who have a predisposition to feel every bump on a rocky road."

I need to bring up storytelling here, simply because at higher levels of visual animation realism, it's a loaded design choice. You have to have a good reason to use it, especially when your characters reach the point of strict animation realism. You don't have the freedom of exaggeration that is available when you make something more abstract, which effectively weeds out most gags. It makes more sense at this point to create a work that exploits the technique for a narrative that demands the realistic look, create a more experimental work, or combine both approaches. You don't get the meat and cheese of animation, but you get the two slices of bread on the other end of the sandwich.

It goes without saying that these characters of stricter animation realism are not as essential for the media as others. Animation could live without the character Admetus but suffer with the loss of characters like Oswald the Lucky Rabbit. On the other hand though, some of the most easy to animate characters are also by extension, the easiest ones to copy in terms of basic form, which, let's face it, can degrade their individual value in the long run while it strengthens their importance. The fact is that strict realistic animated characters are not and never will be as well suited for animation as more abstracted and/or exaggerated ones. That, however, does not degrade the artistic value of the few done well, because they cannot be held accountable for these traits when the number of works remains small as they do not exhaust their potential right away.

Now I'm going to address something you've certainly been waiting to hear about. You've certainly been told that in order for animation to function effectively, motions have to be accented. Shouldn't strict realistic animation be unjustifiable as a result? Well, my response is this. Paint on film animation allows for more leniency than hand drawn animation in this regard, and where hand drawn animation is designed to be easy to understand visually, hand painted animation is better at hinting at what you see and does not require the animator to outline the forms clearly. With its extra steps towards capturing a person's look without bolding out the detail, it allows just enough freedom to portray a character which looks like somebody who's overacting in the midst of fancy cinematography.

But what about all the arguments towards simplicity in animation? Here's a few paraphrased, hopefully correctly, which I've seen people make.

1It's harder to hide your poor draftsmanship in the high level of detail that realism provides and reducing your artwork to it's simplest form demonstrates your understanding of art principles and the essence of what you've portrayed.

A valid point. It certainly is difficult to hide poor decisions made in a simple artwork because they're not masked by the complexity of the work. However, the simplicity in reducing a work to such simple form also gives the artist the opportunity to weed out details that they're not so good at portraying and focus on those which they are, which is not so feasible in a more realistic work. This point seems to come out stronger on non-human characters which members of the audience may not be entirely familiar with. Sure, if you deliberately clutter a drawing or an animation shot with details, it's possible to obscure the work's flaws. If you deliberately go from overloaded quick shot to overloaded quick shot in an animated film, it's possible to hide it's lack of quality. Heinz Edelmann admitted in an interview to overloading Yellow Submarine with if my memory serves correct, 20-30% extra detail because he knew the story would not stand up to scrutiny. Certainly plenty of realistic artists are guilty of hiding their errors and shortcomings in a forest of bombastic detail, but I think that while details don't convey as much in a realistic work due to often having so much, an artist can still show their effectiveness of craft in repetition of lines, contrast, and use of color.

2. Simpler work reads quicker and thus makes a better, more immediate visual impression. Maybe so, but this comparative inefficiency is dealt with fairly naturally. Realistic characters are less easy to exaggerate, and it's less appealing to do so, which in effect means there's more time to study them in their greater complexity due to the greater consistency in their appearance. Silhouettes can read well enough with realistic characters for an observer to tell the difference between a good and a bad pose, so I say it's simply good enough. On the other side of the equation though, what is the character good for after they catch your attention? Simply rendered images may be more iconic, but I'd say that there's another side to this, that simple images can be distinct but not distinguished. Plenty of type fonts can be bold but scarcely distinct from one another when compared and so can an overwhelming number of simple cartoon characters. Plenty of cartoon characters from Disney and Warner Brothers from the 30's and early 40's manage to come across as united in their simplicity of form in spite of being very different characters.

In having to rewrite this post as Blogger's save function fails, the internet malfunctions, and accidental erasures stemming from the first, I've been able to rethink this post half a dozen times. I apologize for the blatant shortcomings in this post, because it's been a real nightmare.

Now I'm going to explain how I feel about full visual realism in animation. First, how should it be done? The obvious answer, at least at the moment, is that it would be done with computer animation. Motion capture I think is good for little more than reference and maybe not even that. I think that characters ought to be completely virtually developed, from every aspect of the way they look to the artificially produced sound of their voice. I think that the technique if successfully pulled off, would be a completely different form of animation than we think of it right now. It would be best for more of a treacherous realism, so to speak, where everything would look real, but in a way that's entirely unusual and unfamiliar to us. There would be bands playing unusual instruments, unfamiliar vehicles, fictional ethnicities of people, nonexistent breeds of dogs, stop lights with different colors, and best of all, characters with more legitimate conflicts than those that are perceived in real life.

What would you think of you saw an animated film that seemed completely realistic in every last visual detail where you saw characters emerge from inside of wombs, age and die realistically, and stop at purple-orange-blue traffic lights? With characters that don't look quite look like anybody you've seen before and never appear again anywhere else, and everything you're familiar with appearing in an unusual form? I think that full visual realism, while limited by the aesthetic standards of animation, would counter its limitations by the ways it takes your everyday familiarity and twists it to unparalleled advantage in a way that visually simpler art does not attempt.

This would be an outrageously difficult feat to achieve, but if it worked, would it really degrade animation? I don't think it would be much more valuable than an intriguing antithesis to what's done in most of animation. Would it really make most animation less entertaining by comparison? I personally don't see it adding up to anything more than a black sheep compared to the rest of what's done in the medium. Much of the hype and resultant worry about motion capture I think has steered people off course to how virtual realism would happen. Motion capture only captures the data of one person, so how's it supposed to make much progress in the long run? It seems to me that we've mistaken the superficial realism from motion capture as the way full visual realism (regardless of animation)would be achieved. It seems to me though, that artificial evolution in virtual environments presents a far greater long term possibility of achieving the higher levels of faux realism. How else but by artificial generation can somebody dream of mimicking life to the fullest degree? Link to example In the long run, this seems like it would provide more possibilities for figuring out how to simulate virtual environments which look real because the knowledge gained is additive.

If you wanted to go underwater, it would be easiest to start with a suit and air hose, but eventually if you want to go deeper, you'll need a submarine. I suspect it's the same way with animation realism. Sooner or later, the hallmark of artificial realism means departing from real people into creating fictional computer generated people, from the way they look from the way that they talk. At that point, motion capture would be worthless because imitating real life does not make you better suited towards entirely inventing it from scratch.

Animation realism is extremely convoluted. When it looks almost completely photorealistic, rendering and all, you have to use motion accents. When it's in the general direction of photorealism and hinted at, you do not necessarily need motion accents. Since it's so difficult to make a character at extremely high levels of realism, less realistic characters can actually behave more realistically because the audience accepts them more easily.

So that's 3 different sorts of realism embodied by very different depictions. It's no wonder that people are confused. If you look realistic, you can't move completely realistically, if you move realistically, you can't be completely realistically rendered, and if you behave realistically then you have to take a few steps back from realism altogether. What a conundrum. Throw on the laziness of filmmakers cheating the technique, the lack of proper plots for much of it, experimental filmmakers who blow boundaries, conflicting sorts of animation realism, and lack of recognition for the best works, and you've got utter madness.

So, wrapping this up as best I can. There's plenty else to say. Russian animation has plenty of examples of poor strict animation realism and Americans scarcely have done it right. The Japanese seem to have avoided strict realistic character animation in 2D, but they've done some brilliant work in other areas. Just look at some of the scenes in Wings of Honneamise and Akira. My ideas on realism in computer animation stem largely from the post of Ward Jenkins where he shows how to handle the characters from Polar Express better with a few tweaks in Photoshop. I've got too many things to say and too little time.

I suspect I'll be making adjustments soon, but here's what I have to say for now. Onto a less intense subject.

edit-confused abstraction with under caricature, a stupid mistake

No comments: