Some of the most memorable characters in Tim Burton’s ‘Alice in Wonderland’ are theAlice-in-Wonderland-Mad-Hatteranimated
CG creatures and personalities that emerged from Sony Pictures Imageworks
studios, from the hookah puffing Caterpillar to the fearsome Jabberwocky.
From Digital Media World Magazine out now

As Animation Director at Imageworks, David Schaub supervised the character builds and performances of each one of them with a seasoned team of animators.
“We created all of the non-human characters in the film,” David said. “We ended up with over 30 characters and produced more than 1100 shots over a course of nine months with a team of 60 animators divided into eight different teams.  We also had a small animation team at CafeFX that handled the book-end shots and some additional animation of Bayard and the Red Knights in Underland.  We got started with movement studies in April 2008, after Senior VFX Supervisor Ken Ralston had already spent about six months working with Tim Burton on character designs with a team of development artists.

Animation Style
“Since the designs were not completely locked down when I started on this project, there were no animatable rigs to work with. In order to get some feedback from the director, we hacked together a pseudo-rabbit using a guinea pig character from the ‘G-Force’ pipeline that the team had just finished working on, added some rabbit ears, spectacles and a top hat. At least it was something that could be animated in hopes of getting some feedback on the character animation style that Tim was looking for.
“I used a bit of White Rabbit dialogue from the original Disney animated ‘Alice’, to determine the animation and acting style Tim was looking for – just how ‘cartoony’, dramatic or expressive should these characters be? At first, Tim seemed interested in exploring the ‘cartoon’ style, with traditional exaggerated moves and comic timing, but quickly gravitated back to more subtle performances. Character design continued well into production.”

David’s role in the early stages of a film tends to involve working with modellers and riggers on a fairly tight development loop. They do a little modelling, then build some rigs and stop to test them with animation to see if the models will support the performance style they are looking for, and have the right proportions and an adequate level of control. This means that if something isn’t right they can turn back and work with it through further iterations on this loop, before committing to precise proportions in the design.
Models were created in Maya and sent to MudBox to sculpt the fine details. In some cases displacement maps were baked out for fine wrinkles to be applied further down the pipe.  The final step in modelling was to convert back to Maya for the rigging team. In addition to the native rigging tools in Maya, Sony Imageworks developed a system of ‘slick-skin/wire-skin’ deformers that help preserve volume and enable skin to slide over surfaces.  Maya’s C-Muscle technology was also used for skin dynamics, collisions and additional skin-over-bone effects.

Tim gave David’s team plenty of freedom to experiment with characters, resulting in one of the most rewarding and collaborative experiences that he’s had. Tim’s confidence with the crew meant that he could allow them to explore while keeping his eye on the bigger picture, and steering the creative process by providing guidance along the way. David noted that Tim handled all teams and the whole production in this way, and was a great collaborator on all levels, encouraging each team to find answers on their own.

Wacky to Traditional
A huge group of artists were working through initial design ideas for Tim and Ken Ralston,  but in the end it was Michael Kutsche that seemed to nail Tim’s vision, and Michael produced most of the final character designs. Michael was given liberties to explore many possibilities, from wacky to more traditional. Two major designs that, in fact, gravitated back toward the original descriptions in Lewis Carroll’s book and the engravings illustrator Sir John Tenniel made for its first edition, were the Jabberwocky and the Cheshire Cat. The book eventually influenced many looks, although as David said, perhaps what was novel in the movie’s characters was the way the team used the traditional designs.

By this time, script writing had been virtually finalized by Linda Woolverton, and Tim’s production team had produced storyboard panels and previz for some of the trickier sequences. David’s team were free to focus on characters and getting performances from them to realise this story.

Building Rigs
Animation production was underway from April to December 2008, a very short time for so many characters. David said, “The typical process on a show like this is to rig a character for every sort of action that might be required in the shots. We normally take into account all the muscles, tendons and other anatomical deformations to be sure the character can hold up under any scenario and look right from any angle. The reality is that no matter how much you engineer and prepare for what you expect, 20 per cent of the time, production shots present something the rig can’t handle.

So, they simply aimed to rig for about 80 per cent of possible cases, and didn’t worry too much about the extreme, ‘hero’ shots that were always going to present a challenge. “We don’t refine the character to the highest level of detail until we’re faced with a specific demand,” said David. “Then we can build controls for that particular case, and the character rig controls and features evolve as we work through the show.”

The animators had their rigging lead Aaron Pfau working with them through to end of production. Instead of simply rigging a character, handing it to the animators and moving on to some other project, he helped them add a ‘shot finaling’ step. Once the basic performance was approved by the director, Aaron’s team would add special touches like folds in the skin, additional tendons and muscles designed for the shot, as well as a skin dynamics and ear simulation for a character like the Bloodhound.

Tongue Twister
Another example is in the scene where the Bandersnatch licks Alice’s wounds. Those shots demanded close-up facial movements that they had never anticipated in such detail. While his tongue did have controls and a joint chain, it was only capable of a normal range of movement. So they built a more complex tongue rig specifically for that shot only, instead of putting it in permanently to do something that would never be needed again, which would have slowed down the performance of the rig in the rest of the shots.

This flexible approach represented a shift in thinking for a large production like this where over-engineering tends to be the typical approach. “It also opens the possibility of keeping a simple set of generic rigs and embellishing them as needed,” said David.
Despite appearances, all the animals in ‘Alice in Wonderland’ were animated. No live animals were used at all. The shooting schedule was so tight that Tim wanted to capture the shots quickly and not have to worry about animals on set. The Cheshire Cat, one of their earliest characters, helped set their approach to animals generally. “A typical first-pass for a character like this would be to do a walk cycle, which we did,” David explained. “Taking this first step forces the conversation with Tim, and we quickly learn more about what he has in mind. Yes, it is a cat, but this cat floats weightlessly. So if a cat was weightless, how would it move and negotiate its way through space? These are the kinds of riddles and challenges that Tim would leave us with.”

They decided they would avoid humanising any gestures to align with actor Stephen Fry’s dialogue. Instead of moving his hands, he would roll around and flick his tail, like a real cat. However, how was he going to fade in and out of view?

Cat Secrets
This character had to directly involve the visual effects department, where Ken Ralston requested further performance adjustments that would enhance the effects they were creating. That meant further iterations and evolutions to achieve the necessary expressiveness.
The Jabberwocky, of course, also needed input from the visual and special effects teams. Tim’s direction was to make this creature different from other dragons that everyone has seen before. Instead of blowing fire, he blows electrical effects from his mouth, and a misty haze rises from his skin. He has wings, not to fly around with, but to use like legs on a large insect, for balance as he climbs up the tower.

Poses and animation nuances were inspired by Ray Harryhausen’s classic fantasy creatures, which Tim and Ken Ralston both admire. The big ‘reveal’ moment of the Jabberwocky was a direct recreation of ‘Night on Bald Mountain’ from ‘Fantasia’, 1941.

Hybrid Animations
David is often asked about the hybrid characters in the film, Stayne who is the Knave of Hearts and the pair of twins, Tweedle Dum and Tweedle Dee, and whether their performances relied on motion capture or not. The team actually didn’t know much about the range of behaviour the characters would need to perform and so motion capture sessions were carried out in both cases with the actors on stilts. The data and performances were stored, but never used as the basis for scenes for either character. The kind of movement that Tim ultimately wanted was more stylised than what the motion capture produced.

Crispin Glover played Stayne, the Knave of Hearts, but the only photographic element of Glover that the character designers actually used from his performance was his head. “The rest of his body is completely CG. From the point where the skin met his collar was green all the way down, partly as a means of accelerating the shooting schedule and partly because they didn’t know what the costume would look like for quite a while. He was simply another strange character in this strange world,” David said, “Eight feet tall and lanky, that was really all we knew about him going into this, and details about his costume were figured out long after the shoot was complete.”

The photographic element of Crispin Glover’s head was extracted from the plate and stabilized about a point at the base of his neck.  That image was applied to a card, and the card attached to the animation rig and orient-constrained to the camera.  That means the head was no longer locked to the position that it was in the plate and gave the animators freedom to move the character around in the scene as needed.  Once the performance was approved, fine tuning of the tracking was done in Katana - Imageworks in-house compositing software - and in some extreme cases was stabilized and tracked in Nuke or Flame. 

Creepy Twins
In the case of Matt Lucas playing the Tweedles, the video recorded during the shoot became the animator’s visual reference. Matt would play one part alongside a stunt partner, and then swap roles and play the other part. Editorial then took each performance and cut them together, forming the master reference for the animators.” The photo elements from Matt were his eyes, nose and mouth, pasted back onto the faces for an extra touch of realism, making the twins unmistakably Matt Lucas. Katana was used to dial in the final tracking of these elements, and in some cases was done in Nuke or Flame.

Because the one actor handled the Tweedles single-handedly, they can be regarded as one, hybrid character. Otherwise, like Stayne they were totally animated CG characters. The pair’s creep factor particularly appealed to Tim, inspired by the nightmare twins from ‘The Shining’.

Off with Their Heads

The live action shoot was completed in only 45 days, and required Tim to move from his home in London to Los Angeles, which he’d been hoping to avoid by relying on video conferencing for the post-production process. But it would have been impractical for Tim and his editor Chris Lebenzon to wait until the end of each business day before receiving anything to edit or review, due to the time difference. Furthermore, having him nearby meant he was able to spend about two hours a day working with the animators and VFX supervisors Ken Ralston, Sean Phillips and Carey Villegas.

The animation of various inanimate objects in the sets, many of which turn out to be animals with the job of holding furniture in place is a distinctive feature of this film, and kept David’s team busy. Large contingents of monkeys holding candlesticks at the Red Queen’s court, for example, serve her silently in terror while frog footmen stand at attention in the background, frozen in fear of losing their heads. The frogs’ ability to express this fear with great subtlety was important.

Subtle Performance
Because the animated characters and live actors interacted so closely in the story, their acting style needed to match the direction Tim was giving the actors on set, which emphasised subtle, nuanced performance. Ultimately, this meant that enough subtlety had to be built into the models and rigs. “The green set was across the street from the animation team, and we could watch Tim in action and understand the expressive type of performance he demanded. Typically, animators tend to go for broader more obvious moves and expression, but in this case we had to pull our performances back to match the live actors’ style.”

In one sequence, one of the frogs undergoes an interrogation from the Red Queen. He is virtually frozen with fright, but the tiny flickering movements just visible in his throat as he gulps and of his bulbous shining eyeballs, portray a guilty conscience better than grand, dramatic gestures would have done. Very precise timing was critical.
Incidentally, Actor Helena Bonham Carter as the Red Queen performed her court scenes entirely alone on the green set with only green cardboard cut outs to establish eyelines and locations for her frogs. Furthermore, all the live action courtiers and extras participating in these scenes were captured at a separate shoot several  months later.

Geometry of a Smile
Tim liked the effect of animals popping out of the woodwork and enlivening every scene, using them to show the dominion of terror which the Red Queen held over her subjects. As production progressed he had the team add many such animals to several scenes. The crowd at the Hatter’s execution developed from a generic crowd to one that included a collection of animals. The scene was re-created either with the animals inserted as elements, or with separately produced shots inserted into the edit. Finalising this scene was one of David’s team’s very last tasks.

David described some of the special problems that certain characters posed. The Cheshire Cat’s smile spanned a challenging range of movement, from when the cat makes a narrow contracted ‘ooh’ sound to his wide ear-to-ear grin. “Technically, this represents a huge stretch of geometry,” explained David. “Fur thins out when it’s stretched and bunches up when compressed. “Also, when he smiles, we had to add controls to deform the internal palette so that the teeth would track along with it. We realised that smiling separates the mouth from the teeth, revealing the gumline and skull, and needs controls to curve the interior of the mouth to show teeth top and bottom.”

Mad Menagerie

The White Rabbit also demanded special subtlety. The final character, all CG, has a photorealistic quality with a beautiful face and detailed suit of clothes. “Although our earlier iterations were more cartoon-like, eventually Tim definitely decided the animals’ looks and actions should be firmly grounded in the natural world, with completely natural detail. The White Rabbit couldn’t walk like a person but had to be fully rigged to like a rabbit to lope along on all fours, but yet transition into a bipedal pose to deliver his dialog.

“The March Hare’s character was developed fairly late in production. The brief there was for a crazed, jittery creature, constantly twitching and moving. The Caterpillar, sitting on his mushroom, needed organically flowing movement and also had to interact with CG smoke that the FX department added to the shots, which meant we had to anticipate his eyeline for smoke that would be added later. His fluid facial movements were critical, but fortunately we could start with our generic facial rig that gave us about 80 per cent of what we needed, and then added controls to extend the range. The Bloodhound had loose baggy skin that required, as mentioned, a further simulation layer on top of the animation, and extra facial controls for his expressive eyes and brow area.

“This project really was demanding in terms of time frame and the sheer number of characters and the diversity of techniques applied to bring the imagery to the screen,” David said. The large, diverse team at SPI, and their ability to collect large amounts of data about characters and performances, is a valuable asset. We place witness cameras on set to collect our own version of the footage captured for the movie, but we also place other cameras around the stage to capture a scene from different angles. When it’s time to animate, the plates are scanned, the team looks at the shot and gets started. If they want to see it from another angle they can stop and look at that. All the data is catalogued to refer to later.

“It’s not often necessary but in Stayne’s case, because Glover’s head had to be tracked back in, they really needed to see all angles to accommodate the effect the 3D would have on the image. They had to register him in z-space as well as x and y.”

Third Dimension

The film was not shot with a stereographic rig. The 3D was applied to the photographic elements later as an effect while our CG worlds could simply be rendered with two cameras. Nevertheless, the animation team had to be aware of special demands that stereo 3D would place on their performances. For example, the usual ‘cheats’ that animators can rely on in 2D, such as when live action and animated characters touch  no longer works. When the Tweedles hold Alice’s hands, one on each side, their hands could be visually joined in 2D without too much trouble using light and shadow effects to perfect the connection. But in this production, that contact had to be ‘real’.

“Alice herself had to be dimensionalised, and her photography needed to be rotomated – that is, a CG version was made of her, animated to match her precise movements. The live photography was projected back onto the geometry of the CG ‘double’, and the Tweedles could then grab this double’s hand.

“Eyelines are complicated as well by the third dimension. When two characters are looking at each other, eyelines won’t work if one of them appears to be standing deeper into the set than the other. In post, the stereo could be applied to each character separately to correct the problem but obviously, this made the process more complicated. Also, character performances that looked right in 2D may appear tipped, leaning backward or forward, after 3D has been applied. We had to remember this while we worked, and really focus on balancing a character correctly on all three dimensions.”

Words: Adriene Hurst with David Shcaub
Images: courtesy of Walt Disney Studios
Featured in Digital Media World. Subscribe to the print edition of the magazine and receive the full story with all the images delivered to you.Only$77 per year.
PDF version only $27 per year