And while the story does not focus on furry animals, hair plays a key role, with a number of characters sporting a large number of hairs. Indeed, Santa has his trademark beard (which, when combined with the hair on his head, totals more than 900,000 hairs), and Grandsanta has nearly 400,000 hairs covering his head and beard. Furthermore, Santa’s beard is fully rigged with 22 controls, while Grandsanta’s contains 27 controls to shape and move his. Although Arthur does not have a beard, his bushy coif contains more than more than a half million hairs and ranges in length from 6.64 inches to .02 inches. The reindeer Dasher, however, tops them all with over a million hairs. And this is just a sampling in the CG hair/fur area.
Doug Ikeler, visual effects supervisor on the film Arthur Christmas, provides a closer look at these and other digital techniques and innovations required for this unique project in a Q&A with CGW chief editor Karen Moltenbrey.
How long as the movie been in production?
Was this a true Aardman/Sony collaboration?
It was a collaboration. Aardman came up with the concept and script. We worked together on the design and visual development, and our team at Sony created the digital imagery.
What look were you going for? It definitely has an “Aardman” look.
We knew quite early that we were not going to have a stop-motion look like Flushed Away had. Rather, what we wanted was a new and unique look that would ‘feel’ like Aardman. It was essential to the theme of the movie that we place the viewer in their world when we told you how Santa gets the job done in one night. Most Christmas movies place you in a ‘Christmas world.’ Our world needed to be a recognizable and familiar-feeling world. The ‘Aardman-ness’ of the movie came in the design and acting of the characters, and then we aimed toward a tactile quality to the world. Stop-motion sets have a very tactile nature to them due to the qualities of surfaces and textures on a miniature set.
What is new technically with this film?
There were plenty of technical hurdles and efficiencies that the Imageworks team worked through, but the scale and detail are the real milestones for this movie. Our film used our in-house Arnold renderer, which is a raytracer. Raytracers by nature give a much greater sense of detail just because of what they can accomplish. They are also by nature hungrier for processor time. If you couple that with the desire to have many large-scale and intricate sets, you see that very quickly you can bite off more than you can chew. The fact that this team was able to manage that, the ambitious goals of the film, and then hit the visual mark that we had set is what sets it apart.
Describe the pipeline you used.
We use a modified version of [Autodesk] Maya for modeling, rigging, and animation. We used [Side Effects’] Houdini for effects and Massive software for our elf crowds. We also have our lighting tool, Katana, and our renderer, Arnold. In addition, we used Maxon BodyPaint for texturing. The backbone is an in-house asset management pipeline that connects everything and a production-tracking pipe that references it all.
What about the modeling? Which characters were particularly difficult?
We were modeling pretty fast and furious. Much of the look and design of the sets was added during the modeling process—something we are not used to doing. That is a testament to creativity of our modeling team.
We started the character models early on, when we were still in Bristol, England. We did a ‘look of picture’ test in Bristol to help flesh this out. The models had an extra level of detail that helped sell the scale of our world. That detail was, of course, in the textures, as well.
Grandsanta probably had the most TLC put into his model. His design is heavily influenced by his beard and the shape of his poses, so we went back into his model a few times to get the all the parts fleshed out. The same goes for Santa; his beard was such a contributing factor, and we had to work back and forth between the model and the grooming of the hair.
Arthur posed a challenge with his sweater/jumper. He is very skinny, a stick-like figure. His silhouette is heavily influenced by the shape of his sweater after it gets relaxed onto his small body. We would approve the model in a T-pose and his sweater in a ‘balloon,’ pose and then animation would pose Arthur and cloth would relax the sweater on him. Once again, we would have to go back and forth through these steps before we were confident in the model and design of the character.
Did the humans or non-humans present the most difficulty?
Humans are typically more difficult because I feel the human eye is more critical due to the familiarity of the subject. In CG feature films, I find that human character design and animation fall down before any other character because of this. That said, I felt our character designs were stylized in such a way as to not make this a problem, and the style of animation was slightly more exaggerated.
What about the textures?
We relied on the standard array of textures, like color, spec, transparency, sub-surface, a couple of bumps, and displacement occasionally. But the real power of our textures was the skill of the artists and the way they worked with our look-development team. They iterated quickly and were always contributing to the creative conversation. I was frequently provided with options of looks because they worked so quickly.
There were a lot of character rigs. Why so many?
The characters in Arthur Christmas were some of the most complicated rigs we've ever built. This was attributed to the numerous costumes, accessories, hairstyles, and skin texture options each character possessed. For instance, each elf had three primary costumes, two styles of backpacks, and a vast array of accessories, such as eyeglasses, gloves, watches, patches, harnesses, hats, tools, hairstyles, and so on. A complex set of rules had to be written to ensure accepted combinations as designed by the art department. Arthur Christmas was also one of the most heavily prop-rigged shows we've done at Sony Pictures Imageworks, numbering in the hundreds. Nearly every shot had a character with something in his or her hands, whether it was elves and their tools, Mrs. Santa wrapping her presents, Steve and his Hoho 3000, Arthur and his slippers, or Grandsanta with his cane and dentures. Dozens of rigs were created just for wrapping gifts, which included various wrapping paper, bows, and ribbon rigs.
Was the sleigh, the S-1, the most complex object?
The sleigh was essentially a ‘super rig’ comprising 14 complex character rigs: eight reindeer, hitch, sleigh, Arthur, Dasher, Grandsanta, and Bryony. Each rig had a vast array of options for accessories, textures, and other details. Thousands of connections were made between each character rig, and the whole assembly had to hold together through a dizzying array of dynamic animation shots.
Did you use any crowd simulation?
We used Massive as our crowd software for our elf crowds. We also used a variation process to deliver many options when it came to building an elf for the crowd system. We designed the process to have the following choices: four male and four female models; skin color; skin texture; costumes, hair types, proportion adjustments, and gadgets to carry.
Can you give a quick snapshot of the lighting that was used?
Lighting leveraged off our HDRI sky domes and had the added realism of indirect diffuse calculations. The images were rendered in floating point, which allowed for a full, unclipped value range that gave us greater flexibility in compositing. Ice ended up being one of our larger challenges. Arnold gave us the ability to get close, physically accurate ways to render it. We made at least nine different types and looks of ice.
What would a Christmas story be without snow and ice? What other effects are there?
Among our effects are: magic dust, the effect that allowed the reindeer to fly (particle and fluid sims); water interaction from characters and the S-1 pole (particle and fluid sims); fireballs from the sleigh as it re-enters the earth’s atmosphere (fluid sim); footprints in both sand and snow (dynamic particles); ocean sims with procedurally generated surface foam; the aurora borealis; clouds, some that get sculpted into a snowman (volumetric renders in Arnold); smoke and flames from predator drone.
Why was the hair so important?
The hair on several of our characters was a defining attribute to their look and animation. Grandsanta’s hair was thin, wispy, and soft. It softened his look quite a bit. Dasher (what we call Grandsanta’s pet reindeer) had clumps of hair that gave him an aged look but made him cute at the same time. The hair around his neck specifically went a long way to making his mangy look more appealing. In fact, all the reindeer are such important characters, and their look is almost entirely defined by the hair. Arthur’s hair was particularly important to his look, too.
What did you use to generate the hair?
We have a proprietary hair-grooming tool at Imageworks that uses Maya as an interface for the placement and shaping of the hair-guide curves and a dso that interpolates the hair density as well as adds effects to refine the shape and clumping of the hairs. Grooming starts off with placing curves in Maya and shaping the general direction and flow of the hairstyle. From there we add procedural effects through our hair dso to generate clumps, wave, randomness, and so forth. Within Maya, we have a virtual barbershop with curve manipulation tools.
We created fairly elaborate hair rigs to give us control over the general hair volume as well as individual hair guide curves. Arthur's groom was challenging, as we went through multiple iterations to make the length of his bangs in his side profile integrate seamlessly with the artwork from his front profile. His hairstyle was integral to help define the characteristics of his personality.
How did the varying lengths of Arthur’s hair impact the process?
As mentioned above, the biggest challenge with Arthur's groom was to marry the front and side profiles together to make a seamless groom. A large emphasis was made on the mass of hair that was generated by his bangs, but it could not distract from Arthur's facial performance. For the approved version of the groom, we ended up with the longest bang hair measured at a length of 17.04 centimeters (6.64 inches) and the shortest measured at .51 centimeters long (.02 inches). Arthur's total head hair count is 511,716 hairs. The team engineered a hair rig to control the main hair volume as well as individual guide hairs, but at the same time, let the hair simulate naturally.
Was the process radically different for Santa’s beard?
The basic methodology for grooming remained the same; however, we expanded the skin under Santa's chin so that it wouldn’t have to be made entirely out of hair.
How many characters and creatures needed hair/fur?
We groomed around 41 different characters, props, and environments—this does not include the multiple variations of hairstyles. With hairstyle variations (different hairstyles) and look variations (dry, wet, hat grooms, and so forth), we ended up with more than 75 different grooms in total. There were 17 male elf styles, 13 female elf styles, eight generic male styles (with two different facial hair styles and two different eyebrow styles), and four generic female styles. We also groomed grass for the opening shot and the beach sequence, as well as groomed Christmas trees, wreaths, tinsel, and even the brushes on a Zamboni.
I assume there was also a great deal of cloth simulation required?
Yes, we set up 19 different costumes for the main characters as well as a few simulation props, such as the reindeer hitch and reins setup. We also had an additional 25 cloth-simulation setups (for a total of about 44 cloth simulation setups—and that does not include one-off setups in specific shots such as for the parachute, tape, airbag, and so forth, for the male and female elf variations, as well as the adult male, adult female, youth male, and youth females.
What were some of the more noteworthy cloth work?
For Arthur's cloth, we tailored two costumes: a bulky, oversized sweater and a red kagool (windbreaker) that has a Gortex-like quality. We have a proprietary cloth solver (called Tango) at Imageworks, where the engine was tailored specifically to give us the ability to art-direct the simulations. For Arthur's kagool, we had the software department implement new code to achieve a never-before simulated, plastic-like fabric. For Arthur's sweater, our biggest challenge was to create a bulky, unnaturally thick fabric that moved like wool but was artificially inflated. To achieve this, we again requested new features to implement into the cloth solver. We also had variations on Arthur's kagool setup, as he has a rope tied around him in one sequence and is taped to the sleigh in another. For the sequence of Arthur carrying the letters, we did a two-stage simulation process. The first stage was to get the dynamic motion of the letters using a commercially available dynamic solver; the second was to use those results to collide and simulate the sweater against using Tango. Our cloth lead had the very challenging task of setting up Santa's Gortex suit. Unfortunately, due to Santa's body size/shape as well as his movement, we were not able to leverage off the Arthur kagool setup. The costume had to simulate with a plastic-like quality but not let it slide around loosely, as the costume needed to retain the general shape/silhouette of Santa's body. We used a pose space deformation system for tight-fitting clothes that emulates the look of simulated clothing (wrinkles, folds, gathering). This system bakes out an actual simulation for a character's costume and rigs it into the animator's puppet. And it is joint driven. This technique was used for the elf clothing (including Bryony's costume) as well as Arthur's pants and Steve's camouflage suit.
Was there a special backdrop that presented a challenge?
The arctic environment was a particular challenge just due to its size and the way we needed to shoot it. The camera is always moving relative to the environment because we are flying with the sleigh at a high speed. That limited the amount of matte painting that we would have liked to have done. The set needed to go to the horizon all around you. Couple that with the need to move the ice formations around for composition reasons, and we had a lot to figure out. The sleigh was moving so fast that you could see parallax a long distance from camera. Layered matte paintings were projected onto a low-res version of the geometry when we could, but the set still ended up being huge.
Overall, what was the most technically challenging aspect on the film?
The complexity and size required to make the movie was our greatest challenge. There were numerous large-scale environments as mentioned, there was a large-scale crowd system, there was a high character count throughout, there was an ocean to create together with a beach break, volumetric and interactive clouds, large amounts of hair cloth simulation, and numerous effects like magic dust, water interaction, fireballs, footprints, an aurora borealis, and sculpted clouds. There was a huge amount of textures and matte paintings done for the movie. There was a tremendous amount of energy put in to the development of ice shaders for the many types and looks of ice in the movie. There were five artists whose sole task was the generation of ‘motion graphics.’ This is the imagery that you see in all the monitors and electronics of Mission Control, the bridge of the S-1, and the handheld devices throughout the movie. The generation of the design and the graphics themselves was a large undertaking all on its own.
For more on 'Arthur Christmas,' CLICK HERE to check out CGW's exclusive report.