French Auteurs in CGI

Last week I have seen two French films that use computer generated 3D animation in explicitly non-photo-realistic ways. I finally managed to see Azur et Asmar (a synopsis can be found here) which has been around for almost two years now. In addition to that, I had the opportunity to see Skhizein – one of my Annecy favorites – again on the big screen. Both of them deal with delicate subject matter. As different as they may be stylistically, they display a few substantial similarities. For one thing, both of them rely heavily on straight front and side views as well as integration of two-dimensional elements.

Jérémy Clapin’s Skhizein tells the story of a lonely man who is exactly 91cm besides himself. This premise is taken seriously as we share the protagonist’s perspective. Although it allows for some amusing scenes, we always empathize with the misunderstood man. The escalating schizophrenic episodes feel neither ridiculous nor sentimental. Right from the beginning we are completely immersed in his distorted world. The tension then builds carefully with perfect timing and pacing.

A lot of shots are either head-on or bird’s eye. These strict setups reduce the three dimensional space to mere planes, so we can clearly see the 91cm distance between the protagonist and his world. Un-hasty camera moves and occasionally different angles support the atmospheric narrative flow gently.

Although the character looks quite different from the sets, we accept him as part of the 3D world without difficulty. His simple two-dimensional face is mounted on a 3D ball which works perfectly. Also his head is much bigger than the rest of his body due to the fact that this is all in his head.

I believed in this stylized character much more than in Michel Ozelot’s stiff “princes”. In fact, the awkward character animation prevented me from enjoying this modern fairy tale all that much. Nevertheless, I found it highly interesting and sometimes stunningly beautiful.

It is definitely worth a closer look because stiff animation may only be the most obvious reason why Azur et Asmar is less accessible to general audiences than traditional fairy tales. Contrary to films like Snow White (or even Skhizein, for that matter) we are not asked to identify with either of the characters. Ozelot escapes the pitfalls of psychological realism by adhering to abstract types rather than specific individuals. Relatively generic character design and stilted voice acting are further emphasizing this distancing effect.

Most of the time, the movie feels as emotionally distancing as an oriental version of Brecht’s epic theatre. The simple story is told in a manner that prevents immersion. We as spectators are required to critically look upon the characters’ action and draw our own conclusions. Simple as it looks on the surface, this is a fairy tale with a complex message at its heart. As a result, I found myself questioning my own (unexpected) reaction to ethnic stereotypes throughout the movie. In the beginning, I automatically thought of Azur as an arrogant brat, mainly based on his aristocratic origin and his racist father, although he looked and acted absolutely the same as Asmar (his Arab foster brother). The downside of this is that the plot is becoming increasingly uninteresting.

There are other things to marvel at, though. While the characters’ plastic doll faces have not been working for me, I still liked the fact that their clothes were mainly flat color planes without any shading or outlines. It is interesting, how Michel Ozelot sticks to his trademark side, front and sometimes ¾ views even in 3D. I couldn’t help thinking how wonderful this film could have been, had it been made with cut-out figures. Often, Lotte Reiniger’s Prinz Achmed comes to mind, which certainly is no coincidence given Ozelot’s past.

Being a movie about tolerance, it has to address questions of belief. It does so by contrasting (fictitious) superstitions rather than religion. Although we see the Arab world from a distinctly Western point of view, we don’t share Azur’s perspective for two reasons: first of all, he speaks French and Arab, yet we never get subtitles for the Arab dialogue. Secondly, Azur chooses to be blind while exploring the Arab world, because his blue eyes would offend the locals.

This is exploited by the talkative Crapoux, a character who is visually fairly different from the rest. He looks at everything from a frustrated Frenchman’s perspective, his whole life is based on lies. So he actually has a more specific face, with eyes hidden behind heavy glasses and clothes textured with dirt. I’d even say, he was animated differently, more cartoony.

It’s symptomatic that he complains that “they have no grey” while being carried through a vibrantly colorful marketplace. Sometimes there are more saturated primary and secondary colors competing for attention than I have ever seen on a movie screen before. Somehow this is in-keeping with the basic theme that everybody may be different yet no one (color) should be favoured. Although at times I was hopelessly lost in terms of where to look at, there are some beautiful compositions involving geometric shapes and a wealth of details. The more muted settings seem to be associated with poor people.

In the most stylized scenes there are touching and strikingly beautiful moments, for example, when Azur opens his eyes to prove his identity to his foster mother or when the fairy of the Djinns illuminates the hall of obscurity.

Both of these films show us what might be possible in CG animation (without even attempting subtle acting) with the necessary restrictions and a strong personal style.

Jérémy Clapin’s website

Skhizein: official site

Clapin’s Une histoire vertebrale (backbone story) on youtube

Azur et Asmar: English official site

While both of these films are clearly targeted at adults, the Arabian Nights feel of Azur et Asmar might also attract younger children. In fact, if you look at the English trailer, it comes across more like a low-budget Prince of Egypt than a poetic tale in the tradition of Cocteau’s la Belle et la Bête (1946).

Ozelot’s Prince et Princesse on youtube

Complementary colors

A friend showed me a link to this online applet: http://www.wackerart.de/mixer.html

As we all know, two colors are complementary if they cancel each other out, i.e. they produce grey when mixed equally. If you play around with it you’ll notice that if, for example, you enter the RGB code for Anita’s shirt, you’ll see that its RGB complementary color is blue. Obviously, you get the same effect, if you invert the picture in Photoshop (ctrl + i). But, as you may know, complementary colors are different for each basic color model. If you invert the same picture in the CMYK mode, you’ll get completely different complementary colors.

To make sure we are talking about the same thing, I’d like to define how I normally use the term “complementary color”.

When we refer to complementary colors, traditional artists usually think of pairs like red – green, blue – orange, yellow – purple. These are based on the subtractive RYB model of primary colors red, yellow, blue. As you can see in the color wheel, a primary color is opposite to the secondary color mixed by the other two primaries.


This same principle is true for the additive RGB model (red, green, blue):
red vs. cyan (mix of blue and green)
green vs. magenta (mix of blue and red)
blue vs. yellow (mix of red and green)

Sometimes red and green (RYB) are called complementary while red and cyan (RGB) are called negative colors. In order to avoid confusion I will adhere to this distinction. (So the wackerart-applet above gives you the negative colors.)

Mr Fun

Just a quick note: the address of Floyd Norman's Blog "Mr. Fun" has slightly changed. You'll find it here: http://web.mac.com/floydnorman/Site/Blog_/Blog_.html

Color in 101 Dalmatians: 2. In a different light

Although I’ve expressed my reservations about the way Disney digitally restores their masterpieces, my analysis is based exclusively on the 2008 restoration for consistency reasons (and because it positively looks superior to the previous DVD). In some cases I will consult the old edition just to remind you that the best looking version doesn’t necessarily have to be the most accurate.
Large parts of sequences 4, 5, 6 and 17 (the happy ending) take place within the kitchen and the living room (established in preceding sequences). Here, I explore the differences in lighting. As the story progresses, color schemes are not as clear-cut anymore.
So far I’ve been analyzing almost exclusively high key lighted scenes. So first I’d like to write about a few concepts involving value. Since the Dalmatians are black and white, they are predestined for studying some basic concepts that can be found in many animated films. To make it easier to focus on value I have converted all of these examples to grayscale pictures.

Character values

“It’s a matter of what the mood is supposed to be. For instance, if there’s a character working in the shadows, maybe it would be good to have a darker character. You know it’s the same one because the silhouette will tell you what he is, and the colors, though subdued, will still identify the character. And then in order to make it read, of course, you have to have the background a little bit lighter behind him. In a reverse situation, where we have a light value character, it will have to work against a darker background. A character is […] either generally a silhouette or generally lighter than the background they’re working on. And it depends upon what kind of mood you want, how you paint it.”

Al Dempster, 1972 (WP5, p.277)

Except for the church scene we haven’t seen silhouettes so far, because the first three sequences were basically done in high key lighting. Even so, the Dalmatians’ values change quite a bit from shot to shot. Normally, Pongo is slightly darker than Perdita. When they appear as parents though, they are usually equally treated.
Now look at Pongo crawling under the stove: whenever we see his backside (which faces the unseen light source), the whole dog is almost white (providing the lightest value in the picture, with the head and the black ears forming the area of highest contrast). Now the reverse angle (middle picture) shows us the shadow side, therefore the whole character is darker than the pool of light behind him. But in relation to the dark area under the stove we still perceive him as a white dog. The closer Pongo gets to Perdi (completely in the shadow), the darker he gets. This is only possible because we always cut to reverse angles in between. We’ll see this throughout the picture (and to a lesser degree in all the Xerox features). As we have seen earlier, due to the high key lighting the characters don’t need cast shadows to be tied to a background. So if the larger part of a character is in the shadow, the whole character is dark. This treatment of shadows replaced the tedious process of animating and superimposing cast shadows as can be seen in earlier features like Cinderella or Alice in Wonderland. In this film, cast shadows and double exposure are only used for two or three standout occasions where they are absolutely necessary.
Although we are again in a theatrical setting (not even windows), sequence 4 basically consists of a series of shots and reaction shots, which is a purely filmic technique. While the room was supposed to be illuminated by exterior daylight in sequence 2, the curtains are closed now and the TV set seems to be the only light source (look at the pool of light between TV and Dalmatians in group reaction shots). Thus, the room is darker, the colors more subdued. For the first time, there is no hint at outside lighting. We enter and exit on a dissolve to an up close TV screen.
Apart from shadows, this is a clear example of hierarchy of values: Even in a freeze frame we can’t help looking at the TV where the contrast of value is strongest. But the dogs themselves are pretty close in value to their surroundings and thus seem to blend in. In other words: they are passive, while the TV “actors” are active. It is interesting that the two puppies who actively look to the screen (“come on, Thunderbolt”) are a little lighter than the others. This may be unintentionally caused by cel overlay, but as we’ll see throughout the film, there are always deliberately different shades of white to prevent the puppy group from becoming flat.


This sequence feels like a contemporary equivalent of the familiar “dinner table scene”, where we get to know the relationships among individual family members. This whole sequence is only possible because the humans are absent. Normally the puppies are a little lighter than their parents, because they are much smaller.

Although we get to know some of them as (one-dimensional) individuals throughout this sequence, they are generally treated as a group character. Note, how Patch – the cheeky one – is always absent in group shots where his patch would stand out if re-used. For once generic design makes sense both economically and artistically.

Excursus on restorationI’d like to compare a few shots from the two available DVD versions in order to remind you of the fact that we have to be conscious of what we are analyzing. Keep in mind that the older inferior transfer looks like it was done from a somewhat faded re-release print and therefore everything is a little on the magenta side. In this example we see that the color of the cel overlay has been corrected, the matchlines are retained though. While I would agree that such minor cosmetics are true to the artists original intentions, it nevertheless makes it harder to look at the film from a historical perspective. But look at the following two shots near the beginning and the end of this sequence (2008 version):
The cels have obviously been painted differently, some of them have even been cleaned up differently (look at Perdi). But look at the ashtray on top of the TV. It has changed from turquoise to mauve for no apparent reason.
And to my surprise in the old version it stays turquoise throughout the sequence. Of course this may have no impact on the overall impression of the background, yet I’m curious why someone would do this on purpose.The important difference between these (virtually) identical shots is that in the second one (with the black TV) we notice Pongo and Perdi looking at each other as an anticipation of their telling the kids to go to bed. Even though we have seen the exact same animation earlier, we didn’t pay attention because the TV was much more attractive. Therefore, it makes sense, that the puppies aren’t brighter than their parents. I’m not sure, if the lighting difference was supposed to be as strong as in the old DVD version. Compared to sequence 2, the living room looks less tidy. We can see that this is the puppies’ playground (slippers, ball, paper on the floor). Like in Roger’s music room, all the objects are painted in their respective color even in the dark area around the TV as if to contrast the bright black and white serial on the screen.

SilhouettesIn the establishing shot for Sequence 5 the human characters are only seen in dark silhouette while the white dogs stand out a bit more. We then dissolve to a medium shot with the focus still on the dogs. But the dark clothed (in very subdued complementary colors) humans aren’t silhouettes anymore (they are not in the shadow). They then pass a grey car with silhouetted figures behind large newspapers. Again we dissolve to a closer shot of the same scene (as if our focus shifted) where the hands are “normally” skin colored. As the papers are lowered we discover two crooked figures watching Roger and Anita closely. Their overall brown appearance contrasts with the cold gray background.
This introductory scene shows the Baduns in their most normal color. Even despite their darkness they are highlighted against the background. As they approach the house, they are silhouetted again, because we have to recognize the now highlighted window (and mustn’t pay attention to the rotoscoped dummies inside the van). Like in previous sequences, window colors are used to connect parts of a façade to interior rooms. Surprisingly this time, the living room window seems to be connected directly to the kitchen.









Swapping colors
The layout tells us we are in the kitchen (actually we see the wall next to the where Cruella came in), but the room looks neither blue nor cold. Even though we have seen it at night in sequence 3, the supposedly artificial light is now rather ochre than white. This affects the tile floor and the brick walls alike, but not the characters. To prevent them from falling out of the picture, the backgrounds are balanced by desaturated areas of pine green. This provides the warmest environment for the puppies just before they are taken out into the cold. Again we see logical value changes from shot to shot. The most dramatic lighting (and framing) occurs in the middle picture when Nanny sees the puppies for the last time.

When Nanny discovers that the puppies have been stolen, the kitchen looks just a little less vivid. We also learn from the calendar behind her head, that Christmas is near and it’s going to be even colder outside. Here we have more “odd geography” that works perfectly from a filmic point of view: When the Baduns ring the bell, Nanny goes off to the right and enters the hallway from the left which looks totally natural (of course the orange window was on the left of the door, but inside it should be laterally reversed), yet when she discovers the theft, she leaves the kitchen again to the right and enters the living room from the left (like it was established in sequence 2). But both work perfectly because the hallway and the living room are held in the same color.
In sequence 5, the living room is also inverted. All the lights are off, because there is no one there (of course the characters could switch one of the numerous lights on, but they don’t want to spoil the mood). All the large shapes are rendered in tones of the same bluish gray hue. The arbitrary, faint pool of light that makes the silhouetted Jasper visible could be interpreted as coming from the window. Again, the details are in the line work but not painted distinctly, as it is harder to distinguish details in the dark. Although the almost neutrally brown floor prevents the BGs from becoming monochrome, there is absolutely no warmth left in the living room or Roger’s music room. Now all the parts of the house except the kitchen are unified by the absence of light.
We unexpectedly see the connecting corridor between living and music room as well as the window (the one that was always hinted at by the hanging binoculars) we know from the outside. Note how even the Baduns’ ghoulish skin color reads as brown against the bluish gray of the background. Nanny’s shadow side is considerably bluer than her neutral appearance.
Here again, it pays off to have the two rooms in such different lighting/color conditions. We don’t even have to recognize the door or see the movement in Horace’s bag to know where he’s coming from. While it might not be the most common thing to have a swinging door in your apartment, it certainly helps here. Whatever side a character comes from, he never has to break the dynamics of a run in order to pull the door open.
Outside vs InsideSince all the lights are off, the hallway now is even darker than the foggy dusk outside. So inside, Jasper is blending in better. This makes sense as he is a burglar whose goal is not to be seen in the dark. According to this, Nanny’s skin color is much paler when affected by bleak exterior light.

As we have seen in sequence 3, if a character looking/coming through a door/window is the important part of a shot, we hardly ever get to see, what’s behind it. A uni-colored area without visually distracting details serves the purpose much better. We see right away that: it’s dark inside (left), it’s foggy blue outside (middle), she’s coming out of the kitchen (right).Of course, it’s hardly possible to have just a pale blue, flat area behind Jasper, when there’s nothing else around him. So the foggy feeling is created by values that are close to each other. This looks downright dull (which makes sense, because it shouldn’t catch our eye) compared to what Nanny sees when she’s looking out of Roger’s window: By the time the Baduns’ getaway car drives off, it looks as if the streets were wet. If you look closely, there is nothing in the layout that suggests anything different from what’s seen before. While some live-action cinematographers (e.g. Janusz Kaminski) go to great lengths to water the pavement because this looks considerably more dramatic, Walt Peregoy’s background painters “simply” filled the backgrounds with vertical shapes in different blues and greens to suggest reflections of the buildings, while also serving as a pool of light for the dark van. First the saturated head and tail lights stand out, then we notice faint yellow streetlamps that haven’t been there before.
If you compare the first and last shot of sequence 5, you’ll notice that we practically see the same content, yet the final setup looks much more dramatic with the camera still head-on. While the establishing shot (left) is almost completely symmetrical and shows the streets in central perspective as we have learn to expect it, in the end (right), not only has the sky darkened, but the story has taken a new direction, so to speak. The dark houses are now almost overpowering the silhouetted Nanny, while yellow is provided more by the streetlamps than the windows. Note how the two framing trees are on different planes now. While up to now everything was flat and close to each other, this last shot already foreshadows the depth of the darkness to come. Nothing is safe anymore.

it's a dreary worldHowever, before the dogs finally leave their home, sequence 6 shows the human way of outside communication: through media and over the phone. First we see a black and white press photo of Nanny in the empty kitchen. At the end of the sequence we then discover that indeed all color (and therefore all life) has vanished from the kitchen. While normal colored humans would have stuck out inside the gray kitchen, the press photos maintain the dreary mood.
Hard white light illuminates the empty basket like a white hole. The overall grey background is achieved by colder and warmer tones according to importance and material. The dogs are almost black, yet we instantly recognize them as Dalmatians rather than Labradors, because of their darkened eye and collar colors. I’m deliberately leaving out Cruella’s and the Badun’s whereabouts, because at the moment I’m mainly interested in Roger and Anita’s home. While we have seen that almost all color is drained from the kitchen, the living room on this gray and rainy day is also less colourful. Again the lighting is influenced by the weather outside the window, again everything is lighted unobtrusively. We accept it as neutral, because the characters retain normal skin tones. Interestingly, blue is mainly absent. So it doesn’t come as a surprise that the (formerly blue) kitchen is gray now.
Although the orange and green pillows are as saturated as in sequence 2, the spread newspaper – it must be the same that Cruella and the Baduns are holding – gives the shot a more neutral overall color. While it fits that Roger and Anita both wear grey shirts at first, it comes as a surprise that Anita suddenly looks green. In the old DVD version, she doesn’t change that much, but the skin color is so off in this sequence that this can hardly pass for reference (you see the difference below. While the predominantly western rule of screen direction (left to right = forward, right to left = backwards) is maintained throughout the film, I think it’s safe to say: While the humans (especially Roger) have turned away from hope, Pongo is still going forward to see if the twilight bark works where Scotland Yard has failed.

The warmth returnsSequence 17 starts with a radio playing a popular rendition of Roger’s “Cruella DeVil” and ends with Roger actively playing the piano. Even if Anita hadn’t said that Roger’s first hit had made them quite rich, we would have immediately found out from the backgrounds alone. Not only does the fireplace look much more golden, but the whole rendering of the now neat place looks richer. Compared to sequence 2, some of the backgrounds are painted following the layout lines far more precisely. Also there are less arbitrary shapes, so it comes closer to a conventional Disney setting. Although the puppies haven’t returned yet, the living room looks warmer than before, because the orange curtain and furniture are stronger. On top of that, it’s Christmas eve. Yet the characters are still sad. While Roger’s lethargy hasn’t changed, Nanny is still emotional (serving tea like in sequence 2) and Anita again seems to be most down to earth. For her, everyday life has taken over, she even decorates the Christmas tree. Anita’s blue clothes make her stand out color wise like in the beginning.
Nanny still is the most “canine” – as Pongo puts it – as she is always black and white (with a red collar button). No wonder she’s the first to hear the dogs bark.







ConclusionSequence 4, 5 and 6 all ended on variations of their respective first shots. These final pictures add noticeable visual depth to initially flat stagings (4: TV characters vs Dalmatians in front of TV screen; 5: flat terrace vs. into the distance; 6: press photo vs. real situation).
Throughout, relatively flat high key lighting without visible light sources makes cast shadows obsolete. Even in the dark, the light is relatively diffuse. The large painted shapes are representational enough to separate shadow from light. In combination with rapid cutting, character’s values can be easily adjusted without changing them within a scene.

Sequence 17 is the only sequence where the living room seems to be lighted by artificial light, since there’s no light coming through the window. Note that the light never comes from one of the many lamps as strong light sources would draw attention to the inexistent cast shadows. Although light bulbs are somewhat warmer in tone than sunlight, this effect is achieved rather by accentuating the warmer colored furniture than by tinting the whole setup with yellow tones, like it is done so often nowadays. There are still balancing turquoise elements and neutral colors. It is only consequential that in full light even smallest details are discernible.
The pure opposite happens, when the Baduns steal the puppies and the living room is totally devoid of light. Therefore we can only identify large shapes. The almost desaturated background color is nevertheless perceived as blue because of the strong ochre tones of the kitchen. In sequence 4 the living room lit by the TV is completely unobtrusive although there are many details visible. It’s always clear that the black and white dogs are part of a varicolored world which is much darker than the TV screen. That leaves us with the telephone scene, where according to the dreary weather neutral and green are emphasized over the warmer aspects of the setting. This of course is subtle. Watching the movie, we constantly accept it as the same living room.
This is a little different with the kitchen: Even though it looked totally blue at night before, it is inverted in sequence 5 to balance the dark blue living room. Later, even all the color is drained from it, to match the bleak emptiness. The overall grey still consists of warm and cold neutral tones. Interestingly, I used to remember the kitchen as three different rooms because we also get to see it from a different angle every time.

Whatever inside lighting we perceive, it never affects characters skin tones. Of course it gets lighter or darker according to the concepts above, but the hue itself is hardly affected (the Baduns are a special case). This constant lets us perceive the background colors as completely natural even though they are not realistically painted. Speaking of neutral tones: black and white objects (Dalmatians, TV, papers) are always more or less painted in values of gray scale, only slightly affected by ambient hues.

An established color scheme is always consistent during a whole sequence, so it doesn’t attract our attention. It also has to fit all the mood changes within a sequence. The dramatic sequence 5 has a very strong but narrow color scheme (just dark blue and bright ochre). It disorientingly disrupts all the previously established color relations while plot-wise the puppies are stolen. It is framed by quiet sequences with large black and white parts and otherwise subdued normal colors. Obviously, color schemes were planned along the lines of narrative sequences.
Color reference (not scientifically checked): http://en.wikipedia.org/wiki/List_of_colors
All screenshots are from Platinum Edition DVD, RC: 2, 2008 unless otherwise stated. All the pictures are the property of Disney, used here for educational purposes.
Sequences labelled according to the final draft (posted by Hans Perk) and Mark Mayerson’s mosaics.

Inbetween

I must admit I was totally overwhelmed by all the positive feedback! Unfortunately, I was quite busy during the last few days, therefore I'm still in the middle of putting together the next Dalmatian installment which I'll probably post this weekend. So please be patient. A few weeks ago I did these value and color tests based on a sketch I did at the lakeside in Annecy. As I'm not overly familiar with pastels, I thought that a little exercise wouldn't hurt.