'The Santa Clause 3': No Escaping These Yuletide VFX

In the first movie, Scott Calvin (Tim Allen) became Santa Claus. In the second, he found a new wife. In the third opus, The Santa Clause 3: The Escape Clause, opening from Walt Disney Pictures on Nov. 3, he must foil Jack Frosts (Martin Short) crafty scheme to take over Christmas. Directed by Michael Lembeck, The Escape Clause, turned out to be a greater technical challenge than the first two movies, as more than 90% of the action took place at the North Pole. It meant that the elf world had to be significantly extended, with many new magical locations being introduced to the audience. A large part of this universe was created via elaborate sets designed by Richard J. Holland, but its real magic was crafted by digital artists at Furious FX and Tippett Studio, the two facilities selected by vfx producer David Yrisarri.

Having just delivered 115 shots for Sky High for Yrisarri, Furious FX was a logical candidate to handle the bulk of the work on The Santa Clause 3. The team included vfx producers Tracy Takahashi and Tiffany A. Smith, CG supervisor Mark Shoaf and creative director Kevin Lingenfelser. We delivered just under 180 shots, which accounted for our largest project total to date and our most complex CG effects so far, notes co-founder and exec producer Scott Dougherty. We also had the notable distinction of being the first vfx house to alter the brand new Walt Disney Pictures CG logo (introduced with Pirates of the Caribbean: Dead Mans Chest) via an added opening segment where the logo is revealed in a snow globe, then picked up and shaken by Santa.

Altering a Landmark Logo

It was Lembecks idea to extend the camera pull back on the Disney logo to reveal the snow globe in which it is supposedly encased, and then introduce Santa Claus, creating a smooth transition with the opening shot. They wanted to get the audience into the movie right away, Co-founder and exec vfx supervisor David Lingenfelser explains. The only problem was that we got this assignment about three weeks before our deadline And since we were altering the companys logo, the approval process was more complex than it typically is. After heavy previsualization, we shot a locked off plate of Santa holding a base without any snow globe mounted onto it. Then, we tracked a CG snow globe to that base and projected that onto our CG camera move, so that the plate would move at the exact same rate as everything else in the shot. We used the 2K file of the Disney logo and tracked it to the snow globe, adding CG snow and a through-glass effect. All this was modeled, animated and rendered in Maya, before being composited in Shake, our main pipeline for this project.

Although 80% of the CG elements created for the movie were rendered in Maya, there were some specific shots for which the team felt that RenderMan or mental ray were more appropriate. Basically, whenever we had heavy effects particles that required motion blur, we used mental ray, Lingenfelser notes. That was the case for the magical gold dust that appears many times in the movie. And when we needed a lot of motion blur on a heavy CG object, like the toy mobile or the gift bag, we employed RenderMan.

Revealing the Great Hall of Snow Globes

Snow globes also played a key role in the Snow Globes Room sequence, an extremely important effect for the director. Lembeck wanted it to be the most magical scene of the entire movie. In this sequence, the Hall of Snow Globes is revealed in all its glory. Surrounded by huge carved ice windows with changing colors, about 60 snow globes float magically in the room, each encasing an image linked to a previous Santa. When we shot the plate, all that existed were the two actors and the frame of the room, Lingenfelser says. We added all the floor-to-ceiling windows, with CG frost animated to softly form and dissipate, the floating snow globes and their content, the snow falling down in the room, plus the artificial snow inside each globe, and all the interactive light on the environment. As a final step to give it a little bit more of a magical feeling, we added many highlights and glints on the globes that turned on and off.

All the elements for this sequence were created in Maya. Once the team had the live-action plate, CG artists started by layering in the background windows. We ran a lot of single frame tests playing with different shades and colors of background. We wanted to see what would give the best translucency on the frost layers covering the glass, without revealing any of the exteriors. We did about 60 iterations before finding the right balance. We then covered the windows with several layers of CG frost passes, and within Shake, we added little glints on the highlights of the glass that would twinkle quickly and go away. After that, we put in three layers of CG snow globes: background, mid ground and foreground. For the background globes, we decided to have some fun and included humorous objects, such as a snowman on a tropical island. Since the globes were all rotating, we had to use CGI to create every object. We even included a three-dimensional image of the director, for which we projected still photographs onto 3D geometry.

Each snow globe required at least 11 passes: a brass base with its own highlight, the 3D interior object, reflections, refractions, artificial snow, and finally, diffusion. Multiply this by about 60 snow globes and youll get an idea of what this sequence involved, Lingenfelser observes. To create a realistic refraction, we actually flopped the live action image upside down and tracked it to the globes. Seeing the actors being refracted upside down really sold the whole concept of the room. This sequence is visually so rich that you probably need to view it ten or twenty times to really see the amount of detail that we put into it

The advantage of breaking the shots down in many passes was that it gave the team much more control to tweak the images and get exactly what the director wanted. If Michael preferred to have, say, less reflections or less refractions, we could back them off in 2D without going back to the 3D stage and re-rendering. In the long run, it probably saved us time, because we had so much more control on the 2D side.

Giving Santa and Jack Frost Their Magic

Another major contribution from Furious FX was the creation of Jack Frosts powers. The character is able to instantly freeze anything or anybody with a single breath. The Frost breath was a particle animation created and rendered in Houdini through a collaboration with Martian Labs. CG artists used 3D animation to reveal Frosts diminishing powers, and to also show the character defrost on screen. Short was first shot twice, once performing the action with a blue jacket (as bad Jack), and then, with a white one (as good Jack). The key aspect of the sequence was the frosted blue jacket breaking up into hard pieces and falling off of Jack, revealing the pristine white cloth underneath. We used high-resolution photographs of the real jackets, and scans of the fabrics, to create realistic cloth textures on our CG replicas, Lingenfelser explains. We then replaced the real jackets in the shot with the CG versions on which we did all the animation within Maya: cracks appearing, more and more chunks falling off In the last shot, we transitioned to the live action Martin Short wearing the white jacket. The last tweak was to paint out Frosts frozen eyebrows to complete the transition from bad Jack to good Jack.

Artists at Furious FX also got to tackle Santa Claus and his many gadgets. The toy mobile was created in Maya and rendered in RenderMan, based on the practical prop that had been used on set. The colorful carousel is revealed in a unique manner, when Santa starts bouncing a simple chain up and down. At the beginning, its just a chain, but we replaced it with a CG version and animated CG particles running down the geometry, Lingenfelser says. The particles quickly draw the shape of the mobile, and, finally, the full toy appears out of the particles. We also created a nice sequence where Santa comes down a chimney. The plate was shot with a chimney set, but we replaced it with a CG model that we animated to stretch and open up for Santa. Tim Allen was shot separately. During plate photography, we had the practical effects team turn the fire down, and then back up. Later on, we timed Santa coming out of the chimney with the moment the fire went down.

Of Rudolph and Co.

Santa was also taken care of at Tippett Studio. Less than 10 shots were realized at the facility, but all were highly complex. The project was produced by Tim de Pala and supervised by Brennan Doyle, with Eric Leven taking over as supervisor for the last 10 weeks due to a scheduling conflict. We created Santas reindeers, as we had done on the second movie, Doyle says. We also built several highly detailed large-scale environments such as Elfburg for the beginning of the film, and the frozen tundra of the North Pole as seen from the air. This last shot also required a CG plane to be created as well as an aurora borealis. For the reindeers, we re-used the CG models that we had built for the second movie, but ran them through a completely different pipeline. We have a proprietary new fur tool called Furocious that generates a much more realistic fur than what we used on The Santa Clause 2. The shots were animated in Maya, rendered in RenderMan, and composited in Shake.

The trickiest part of the reindeer shots was linking the CG reindeers to the live-action sleigh with Allen and the rest of the actors sitting in it. The team started by having on-set matchmove supervisor Devin Breese previs the shots with motion control specifications provided by General Lifts Joe Lewis. The previs was then recreated by the motion control team on the greenscreen set. General Lifts team was able to drive the camera, but the sleigh was mounted on a hand controlled hydraulic gimbal rig. Much practicing was needed to synchronize the camera to the sleighs motion. Eventually a select was made and the camera move was reverse engineered to make it appear that the sleigh was moving through space, instead of the camera. They also locked the CG reindeers to the sleigh to give it the feeling that deer were initiating directional changes and the sleigh was following, rather then the other way around.

Bringing Elfburg to Life

The exteriors and the interiors of the ice cave in which Elfburg is hidden were created under the supervision of painter Ben Von Zastrow. We used World-Machine, a node based terrain generator, to create the basic terrain, Von Zastrow explains. I painted a series of masks in Photoshop to control everything from height to erosion and fed those into WM to create the terrain. Unfortunately, WM is a Windows-only application and, as such, suffers from Windows 2GB memory limit. As a result, I had to create it in four tiles, each of which was 4Kx4K. WM output a grayscale height field that I imported into Photoshop under OS X, and combined the four tiles into a single 8K map. Once I had an 8K height field, the whole thing was taken into Terragen via a plug-in called F.E.O. (which stands for "For Export Only"). The same tool then exported a low-res obj that I could import into Maya.

I also created a simple plane that lived in the exact same location. I used an in-house shader in RenderMan to apply the original 8K height field as a displacement map on this simple plane and generate the actual terrain. This terrain was lit with some basic lighting, but had no shading other than a single, icy blue color. I then rendered a series of still frames of this basic, lit displacement and painted them up in Photoshop with all the details other than shadowing. These images were then projected from the same cameras used to originally render the base images onto the low-res obj imported from F.E.O. Finally, a ray-traced reflection pass was rendered and perturbed using a projected normal map to generate reflections of the aurora (which was originally created in 2D in comp).

The elf village was modeled by Paul Zinnes, including 1,000 cards on which painted trees would be projected. Von Zastrow painted each of the houses in basic UV space and then rendered a single frame of the entire image in RenderMan with some basic lighting. He also rendered a single frame, from the same camera, of streetlights illuminating white buildings using mental ray and Final Gather. These images were combined in Photoshop and then painted on top of to give the final image of the elf village.

As for cave, it was created in two stages. The first, the cave walls were done by painting projected displacement maps and rendering out still frames with basic lighting. Then, icy textures and details were painted to suggest massive ice walls. The mountains and terrain were accomplished more or less the same way as the exterior tundra shot.

Although the project was full of challenges, both technical and artistic, it ended up being one the most enjoyable films for Lingenfelser and the Furious FX team. We were given the freedom by the director to shoot the way we needed in order to achieve the best effects. I couldnt have dreamt of a more collaborative effort.

Alain Bielik is the founder and editor of renowned effects magazine S.F.X, published in France since 1991. He also contributes to various French publications and occasionally to Cinefex. Last year, he organized a major special effects exhibition at the Musée International de la Miniature in Lyon, France.