Sunday, March 27, 2016

Sumptuous Costume Colors: Introduction (Part 1/5)

Among the recurring color concepts that helped create the sumptuous look of Walt Disney's animated features, the quasi single-colored costume - all parts in different shades of the same basic hue - is particularly powerful. In these two posts I will have a look at the concept of analogous costume colors, its precursors and how it became so prominent during the 1950s.

During the classical Hollywood era, the lavish look of Disney features did not only stem from fluid three-dimensional animation (in contrast to most other studios experimenting or simply giving in to limited animation). More than anything, Disney's trademark richness of texture relied on highly detailed (i.e. labor-intensive, thus expensive) character designs and sumptuous (Techni)color coordination well grounded in 19th century illustration styles.

With its rich textures and opulent costumes, SLEEPING BEAUTY (1959) is doubtlessly the epitome of this style. It has mostly been praised for Eyvind Earle's medieval background stylings with their finely chiseled wood textures and angular designs. The character/costume colors, however, are equally important to our impression of the images.
Shapes and colors define the composition more than dimensional lighting and perspective.

And while the Disney artists usually integrated flatly painted characters into highly detailed, dimensional backgrounds by way of empty spaces and pools of light, the slightly more two-dimensional world of SLEEPING BEAUTY required a more graphic approach based on shape and color to match the more evenly lit background patterns.
Storybook illustration from the prologue of SLEEPING BEAUTY (1959)

From the very beginning, Disney was very conscious of the cheapening effect of garish or random "circus poster" colors. Therefore, he urged his employees to more restricted palettes and concepts. By the time SLEEPING BEAUTY was conceived in Technirama70, his artists were able to use the whole spectrum of hues in a shot without making it look all over the place.
If you squint at those screenshots above (or look at them without your glasses), you will still be able to distinguish different characters and groups of characters because their costumes are mostly painted in analogous colors. This is even the case in the crowded image at the bottom. Even though these aristocrats are drawn so richly detailed we can easily see their overall shapes thanks to the unifying coloring approach.
FLOWERS AND TREES (1932): the first 3-strip Technicolor cartoon ever, already restricted palettes: depending on their personalities, tree trunks were painted warmer or colder (the villain even gray!).

Basic Color Concepts and Methods
This costume concept, more than anything else, embodies the marriage of two key concepts the Disney artists were after ever since the adoption of color in FLOWERS AND TREES (1932): clarity and opulence. The early Silly Symphonies already sported carefully limited palettes of subdued colors with sparingly applied primaries. Even a rainbow-happy Easter film like FUNNY LITTLE BUNNIES (1934) was thoughtfully color coordinated.
Color for the bunnies' costumes were carefully chosen: primaries for the workers, pastels for the artists, some even wear suits.

Mickey. Donald and Goofy, however, had very simple costumes based on the primary colors red, blue and yellow, as can be seen in these screenshots from MOVING DAY (1936). Mickey being the oldest and "strongest" is wearing primary red and yellow (toned down considerably for this film). Donald's sailor suit is blue (considered a weaker hue for centuries because it is receding compared to red) and contrasts strongly with his yellow/orange beak and feet. Unpredictable Goofy is dressed in complementary colors orange and blue.
Mickey, Donald and Goofy vs Pete in MOVING DAY.

Simple color schemes for the protagonists. Mickey's yellow shoes are subdued. "White" is slightly blue (not necessarily on purpose, though).
Behind these colors is a color scheme based on the primary colors red, blue and yellow according to Johannes Itten's color wheel. From these three colors, a painter could mix any color of the rainbow - in theory, that is. It never works completely with real pigments. Those primary colors were also available early on for comic strips in newspapers. In the movies, both Donald's (I count his beak as a costume color because it is so dominant) and Goofy's colors are opposites on the color wheel which means that they are complementary colors. Complementary colors are often used to reinforce each other.
How these colors are arranged on Johannes Itten's influential color wheel.

In contrast, Pete's old west sheriff costume (no peg-leg here) is made up of natural browns and beiges. Because their basic hues are very close to each other on the color wheel, they are called analogous colors.
Top row: actual colors of Pete's sheriff costume; bottom row: corresponding hues.

As can be checked in photoshop (see below), the hues of these five colors (top row) are all within a range from 22 to 49ø which is pretty narrow. If I adjust brightness and saturation to 100% we can easily see, how close they are to each other (bottom row of color swatches) on the color wheel.
top: the actual color; bottom: adjusted saturation (S) and brightness (B) show what the hue of 24° looks like.

This might be easier to recognize in this multidimensional color wheel that adds saturation and brightness (but not their combination). Although all of these colors are basically muted variations of yellow-orange, strong contrasts are achieved by varying saturation and brightness.
Left: the range of hues on the color wheel; right: on a multidimensional color wheel.

To compare different costume color schemes I have created a template based on Keira Knightley's ANNA KARENINA gown that has all the fancy parts including a lush waistbelt and a feather on the hat. If we apply Pete's costume colors to her, she looks like a stepping out of a western saloon. Against this background, the "white" taken from Pete's shirt looks deliberately bluer than in the original screenshot* and (accidentally?) serves as a spot of complementary contrast.
The generic "Anna" template for color comparisons. Right: in Pete's colors.

Because we always perceive color in context, I have limited the surroundings in my template strictly to neutral grays and browns. This makes it easier to see hues without color constancy interfering. In fact, Disney often balanced spots of saturated primary, secondary and even pastel colors with neutral grays, browns and natural greens. Those backgrounds and props appear matter-of-factly and boringly devoid of any artistic decisions. But such unobtrusive background colors seldom seem to be randomly chosen.
Left: Silly Symphony title card in primary colors; middle and right: to make the costume colors stand out, a lot of the image was kept in natural browns and grays.

However gorgeous Disney's first two features SNOW WHITE (1937) and PINOCCHIO (1940) may look, the studio's typical color styling only fully came into being with FANTASIA and BAMBI, in my opinion. Since animation did not seem to be worthy of Technicolor's "color director" Natalie Kalmus' close surveillance, Disney's color department experienced more freedom than most American live-action cinematographers and production designers.
Nevertheless "Walt's people" discovered many of the same principles probably because they were, on the one hand, cross-influenced by live-action films and, on the other hand, came from the same painting and illustration background as Kalmus**. In some areas, they had to solve technical issues unique/innate to cel animation.

In color, the jarring difference between the organic textures of gouache backgrounds and flatly painted objects on cels became more noticeable. To make the moving trees in FLOWERS AND TREES match the dimensionality of their background counterparts, the animators added two separate lighting layers, one for the shadow, one for the glow. As any animator knows from experience, this makes convincing animation more difficult and time-consuming.
Both trunks and tree tops had additional layers for shadows and light reflections.
With smaller shapes, a similar effect could be achieved without so many additional lines to the animation process. In order to create a sense of depth to a curtain of leaves in FUNNY LITTLE BUNNIES, the leaves were randomly painted in different shades of green.
Left: simulated depth by adding shadows and different shades of green; right: had all the leaves been painted in the same green, the hedge curtain would look pretty flat like in this digitally simplified mock-up.

With four shades of green and shadows on less than a third of the leaves, the hedge looks pleasingly detailed and textured than in the digitally simplified mock-up on the right. This concept was successfully applied to add natural variety to groups of animals in BAMBI (1942) or FUN AND FANCY FREE (1946) about which I have written some time ago.
Thumper's family in BAMBI (left), the wild bears in BONGO (FUN AND FANCY FREE).

As can be seen in the examples above, this same visual idea was employed to individual animals (just look at the many shades of gray on Thumper). The overall forms of birds and rodents had been divided up into brighter and darker areas that were far easier to animate than shadow layers. Starting with the more realistic tone-on-tone schemes of BAMBI, many forest animals were painted in more closely related analogous colors that are less contrasting but feel more sophisticated than those in earlier features. Even on birds that are made up of different hues the colors look much more unified than in the SNOW WHITE examples:
The 1930s: Clearly distinct colors in SNOW WHITE.
The 1940s: closely related tone-on-tone schemes, natural browns against grays... well as closely related soft colors in BAMBI.
This certainly increased the ink and paint budget, but that money showed up on the screen. While the subdivision into smaller shapes make the animals look more sophisticated or delicate, their overall shapes are easily readable because the parts are unified by one basic hue.
The 1950s: the animals are still very monochrome (different shades of one single hue) in more stylized colors and wearing hats and scarves in spot colors (CINDERELLA and PETER PAN).

In fact, the same concept can be observed in the much maligned Pastoral segment of FANTASIA, but only for mythological fantasy creatures like Pegasi or Centaur(ette)s in art deco colors:
FANTASIA: completely stylized art deco colors: tone-on-tone with spots on the Centaurette.
All this may have initially come to the artists intuitively. But gradually, they started to adapt the same concept to monochromatic fantasy characters and props. In THE GOLDEN TOUCH (1935) - Walt's failed return to the director's chair - everything that turned into gold was animated with both shadow and glow layers (below left). To distinguish golden props on a golden table, each shape was painted in a different shade of yellow (below right).

Eleven years later in FUN AND FANCY FREE, the golden surface of a magic harp was evoked simply by dividing the overall shape into many different shades of yellow/orange without relying on lighting effects at all.
Instead of animating shiny effects, the harp was divided into many different shades of yellow.
If we apply these two concepts to "Anna", the main difference becomes clear very easily: in the left version with shading and highlight, she looks like lifeless golden statue, in the right multi-colored version on the right (which would be less complicated to animate but more laborious to paint), we see a woman wearing yellow or golden clothes. The key difference, in my opinion, is the addition of a more or less organic skin color that is within the same range of hue as the shades of gold.

"Anna" as the golden harp: 1. only shadow and glow (MAGIC TOUCH version), 2./3. using the harp colors from FUN AND FANCY FREE screenshots above.
The importance of such natural skin tones to Disney realism will be further discussed in Part II: Rare Experiments.
* This may be due to the fact that gray tended to lean to the blue side in three-strip Technicolor. But in this shot, we cannot actually tell because it is not even clear if the DVD transfer came from an original Technicolor print. However, I prefer to leave images untinkered so as to analyze what we actually find on current DVDs. In any case, it is important to remember that these colors could have been looking quite differently due to several factors. This, however, does not change the basic concepts at work.

** According to screen credits, Natalie Kalmus expanded her rigid color dictate to the studio only in live-action segments starting with THE RELUCTANT DRAGON (1941).

Friday, March 18, 2016

Preview: Costume Colors

Although I personally like a wide variety of independent and studio animation and anime, I find myself analyzing Disney features more often than even Studio Ghibli. This is due to the fact that I am always interested in what we can learn from classic Disney (and golden age cartoon studios in general) for our own work.

Even if we neither endorse their conservative subject matter nor have their budgets, the thought processes and underlying principles - whether obvious or obscure - can strengthen our storytelling skills considerably and provide us with tools to bring out our very own artistic voices and hopefully advance animation as a medium.
SLEEPING BEAUTY (1959) is always worth a look when it comes to studying color concepts.

Among the fields we can learn most are layout, dimensional animation (animating forces, not lines), timing/rhythm and color. In my next few posts I will have a look at how to make costumes appear richly textured by way of analogous colors. Hopefully, I will also find some time to write about Cartoon Saloon's THE SECRET OF KELLS (2009).

Wednesday, March 9, 2016

ZOOTOPIA Trailer #4: Smooth Continuity Editing

If there is one thing Walt Disney Feature Animation continues to excel at, it is clarity of layout, i.e. staging and pre-editing. Their extremely smooth continuity editing even extends to trailers where shots from very different scenes are cut together effortlessly. In this post, I will analyze some striking examples of how continuity editing with an emphasis on match cuts provides flawless orientation in the latest ZOOTOPIA trailer.

Btw: it would be very interesting to know who assembled this trailer. So if you have any information on that, please let me know.

Try to look "through" this image to see it in 3D.

Successful 3D
Disney (and Pixar - just think of the otherwise forgettable THE GOOD DINOSAUR) have become experts on stereoscopic 3D that is easy on the eye and very effective as a storytelling device at the same time. For one thing, they usually do not go for obvious effects that remind us that we are watching a 3D movie. They even mostly refrain from placing objects too closely to the spectator. Additionally, with Disney's trademark flowing style of animation the head-ache inducing strobing of RealD projection is virtually absent from films like FROZEN or ZOOTOPIA.

Yet one of the central challenges of narrative 3D films lies in the smooth editing. Contemporary stereoscopic 3D films are composed of just as many shots as average action movies. Since there is one more parameter (convergence of our eyes to virtual layers of depth) for the viewer's eyes to keep track of from one shot to the next, smooth continuity editing is paramount for clear orientation.
[Note: Of course, I cannot (yet) examine it in 3D on this blog, so just keep this in mind]

Audiovisual continuity
If you are interested in seemingly invisible editing techniques for fast cut scenes, look no further than the formally unspectacular ZOOTOPIA trailer #4. Since a trailer's basic incentive is selling a product, it makes sense to make it as easily accessible as possible. While I personally prefer the more audacious trailers for films by the likes of Hitchcock (PSYCHO, ROPE) and Kubrick (DR. STRANGELOVE and THE SHINING in particular), this mainstream trailer below contains some amazing lessons in the classical technique of continuity editing:

Assembling a trailer is first and foremost a dramaturgical task: telling a story in two and a half minutes based on previously produced footage. In the video above, there are the usual structuring elements like fades to and from black, five of which are hard cuts to black on musical beats that then fade in (I marked all the cuts that match musical beats) as well as aural continuity by way of Shakira's "Try Everything" (and a Nino Rota pastiche in the GODFATHER epilogue).

Three dissolves serve to indicate more quiet moments even though the change of pace is merely simulated by overlapping two longer shots. The slow dissolve at 1:27:22 also gives us time to look from the left (previous composition) to the same character on the far right in the new composition.
The "sad moment" dissolve simulates slower timing by overlapping two shots.
With an average shot length (ASL) of less than 32 frames (1.3 seconds), the trailer does not leave us much time to linger on details or even adjust to where to look at. Therefore it is important that our eyes do not have to scan the whole frame for the main point of interest. Of course, the most obvious solution to guiding our perception are matching eyelines: we look at a character on the left and in consecutive shots automatically look to the right to see what they were looking at.

Reverse Angles in Widescreen
As you can see in the following images, even though the director could block the conversation partners far away from each other in a cinemascope frame, their eyes are rather close to each other from shot to shot (indicated with red lines). There are also visual similarities in the overall composition that soften the cuts (green lines):
Green lines: position of the backrest on the left and red object as well as lighter area on the right.

Green lines top: positions of Judy (object) and Bogo's fist. Bottom: the hand matches exactly, while Judy is slightly more to the center so that we do not have to move our eyes too much from Bogo's head to hers.
This last shot is especially interesting because it comes from a totally different scene (after Judy fails to produce the rabid Jaguar). However, it is cut at the exact moment when Bogo's fists match. It also helps that the high angle on Judy is the same as in the previous setup on the chair.

Characters are even closer in such shot - reverse shot conversations throughout the film as you can see in the following example that always places us closer to Judy. [On a side note: this is even more striking in 3D where we do not read the staging in the long shot as flat at all because Judy is much closer to our position in the room.]
Here Judy's position is even matched perfectly.
The usual "single" in the middle of the "two shot" setup, with a graphic match in the background.
A two shot vs a single that is basically a two-shot as well.
In terms of content, these setups (above) reinforce the theme of size relations that appears throughout the film.

Match Cuts Within a Setup
When two characters face the same direction, the filmmakers almost match the positions of their faces in consecutive shots. Since eyelines are maintained and there is enough visual information to understand who is on the left and on the right, this greatly helps us in keeping track of the main point of interest in a shot without moving our eyes!

The characters (and by using different lenses even the police banner) appear in the exact same spot in the middle of the frame while the lion's suit, tie and paw clarify their spatial relations. Although these are not consecutive shots in the film, their juxtaposition works perfectly.

The visual match does not necessarily have to be in the middle: in the "four shot" (24:03 below) our eyes are guided from the bunny (and the eyeline of tiger and wolf) to the Rhino's eyes by the timing of the action/movement within the shot. As we are looking at the Rhino, the cut places our bunny protagonist Judy at the exact same spot.
It is important to notice that the red lines only indicate what we are supposed to look at in the very last (top) and first (bottom) frame of a shot. Within a shot our eyes may be guided all around the frame from one character to the other, of course.

Cutting on action - in this case Judy reaching out to Nick - is made even less visible if the lateral position of the brightest part of what moves is matched:

With a characters face in the same position you can change the angle on the background almost any way you like without our noticing it:
During the following cut from a low angle to a far away high angle the whole background is changed. But since the fox and the buggy on which we focus are matched, this is almost indiscernible:
The same goes for the following scene where Judy is standing in the street vs sitting in her cart. No jump cut perceived here:
In the following scenes there is a lot of footage missing from the actual scene but the ellipsis works very well because Judy's lateral position is exactly matched. There even are sort of visual matches left and right of her, but those may be purely accidental.

My favorite moment is a very short cut-on-action sequence taken directly from the film that connects a near 180° reverse angle by matching jaguar paws so perfectly that even in 3D it felt like one single shot!
The eyeline match is from slightly right to slightly left of the frame center...
...while the paws on either side mask the cut almost completely (much brightened for this GIF).
In the scene within Mr Big's limo, the active part of the scene is almost exclusively within the right half of the frame with Nick the fox. Thus it makes sense that the driving car (from later in the movie) is cut when it approximately hits the lateral position of Nick in the next shot.

Within the shots, Judy is talking to Nick and our eyes wander back and forth. But overall, the interesting action is Nick searching the car. When the polar bears appear at the door, the one on the right is obviously more dominant since Nick looks slightly to the right and the right bear is larger and higher in the frame and slightly brighter than the one on the left.

When Judy realizes that "they're all Sloths", in the trailer we get a series of shots that all are framed like shot - reverse shot scenes but with all the moving or speaking characters on the left and the onlooker/overshoulder observer on the right:

It does not even matter that we are looking at different sloths as long as they are...

...changing position from right to left and back like in a reverse angle situation.

We start and end on Judy's reaction to their slowmotion behavior.
In these GIFs I only use the first and last frame of a shot (see timecode).

Match Cuts Across Scenes
The scenes we have seen so far are mostly working so tightly because they were planned that way in layout. After all, those shots came from within continuing scenes. Whoever cut this trailer, however, was able to match footage from different scenes in a completely seamless way.

The following match of two news anchors with a reporter and a cameraman may in fact come from the same scene, but if I remember correctly, those are not consecutive shots in the movie. Be that as it may, in both scenes we are focusing on the left character in both shots because it's only them who are talking.
Although both characters' positions match and we focus predominantly on the left half...
...their head sizes are reversed, making this look like reverse angles.
A less obvious but more elaborate version can be seen in a montage of animal behavior that uses shots from three different scenes (the nudist yak's lobby, the nudist's garden, the howling prison guards).

We cut from a left to right movement to a howling wolf slightly to the right (matching the panther's gaze).

Then the weight is shifted from the white to the gray wolf approaching from the right that shifts more to the center after the cut to the medium close-up.

Then we have the yak's naked rear more or less matched with the giraffe's and Judy's face turned away matching that of the giraffe.

Purely Visual Associations
Sometimes the position of a character is different. Yet we seem to recognize the same set because certain objects, colors or lighting appear to stay in the same position within the frame. Of course, the following is not a real match cut. But the GIF animation makes it clear how closely the beds match (the only part of the image that really is the same). The lighting source on the right and especially the radio that substitutes for the doormap are very interesting choices:
Up to now, the main point of interest in practically every shot was located somewhere in the middle section of the frame. That might well be the case because in 3D we do not like to be reminded of the left and right frame edges because occasionally this can destroy the illusion. But in the top shot above (Judy on the far right), we are looking at a rectangular room that hardly extends beyond the frame edges so there is less danger of breaking the illusion.

There are however some cuts that remind me of the opening shots of all the Indiana Jones Paramount logo matches:
Here, not only the castle is matched (after fading out and in, to be fair) but also the angle (look at the horizon) and the style and perspective of the clouds.

In the next shot the object in the distance is again matched while we cut to a much closer view of the same train top sight seeing window.

Visual matches not based on characters and eyelines are also visible in the following scenes:
Another one of my favorites and maybe just a coincidence. But since the train accentuates the diagonal line before the cut, I can't help but notice the same graphic element in the elephant scene. 

As I have stated above, size relations are a dominant theme which is perfectly illustrated by cutting from a food corner for Giraffes to a miniature car that occupies roughly the same space in the frame. The contrast is heightened by the green vs red background:

Matching Movement
And then there is one seamless match-on-action that not only works because of the positioning slightly to the right but because it completes a left to right movement that starts with Judy cleaning the plate and ends with the cart driving to the DMV (for motion check the trailer at the top of this post):
This approach certainly helps when mashing up two different action sequences into one dynamic swing. Again I have only used the first and last frames of each consecutive shot for this montage:
Starting with a diagonal composition triggered by a roaring lion (last frame of first shot) the trailer cuts to another diagonal line with the fox on the left that segways into a series of shots that have the characters swinging to the right and ends with a very kinetic shot-reverse shot upward diagonal sequence from two different scenes.

Although the point of interest is shifted slightly downward to the left we can see that...
...the overall movement is upwards and to the right.
Of course, most of what we see here is fundamental continuity 101 that we all know of. And by no means would I want to discourage you from finding more original or unique ways of staging and editing (for more interesting action staging across the whole widescreen frame see David Bordwell's examples).

But if your objective is to maintain screen orientation and geography without irritating the audience in fast cut scenes - we are looking at a trailer that is usually consumed only half consciously - then (near) match cuts are a very powerful tool. Especially fast cut action scenes benefit from clarity as can be witnessed in the center framed action sequences of George Miller's MAD MAX: FURY ROAD, where even shots of only 4 or 5 frames register because they are matched before and after.

If 3D, fast cut action films or intensified continuity are not on your mind, then by all means find ways of staging your characters in ways that make use of the full widescreen frame!

Of course, it is vital to cheat continuity on purpose for myriad reasons - and animation offers more tools (e.g. character sizes and proportions can change according to story or mood needs) than live-action - but there is absolutely NO excuse for accidental continuity errors in animation.