The building in the climactic scene was actually 1/7-scale miniature built by Robbie Scott’s Cutting Edge Models.

More complex was a climactic sequence in which a huge tower plays a central part. “The building was a 1/7-scale miniature that was built by Robbie Scott’s Cutting Edge Models,” Docherty comments. “We first shot the actors on partial sets or on bluescreen and then used SynthEyes to track the camera moves and import them into a Milo motion control system. The creator of SynthEyes, Scott Andersson, actually rewrote the program twice to help us do this. The original camera files were scaled down to 1/7-scale, which allowed model unit director Steve Begg to shoot the model with the corresponding moves. The plates were then combined in Inferno or Shake, our main compositing tools, with some shots handled in Digital Fusion. There are more than 70 shots in which what appears to be a set around the actors is, at least partially (and in some cases completely), our model. Richard Bain, Paul Round, John Swinnerton and Janet Quen were our senior compositors on the show.”

The creatures in this grim tale also included this CG mud creature.

Of Wolves and Teddy Bears
Although the set extensions were very demanding, the most challenging aspect of the project for Peerless definitely was character animation. Four different creatures were required: a wolfman, a mud creature, the mirror queen and a living tree. Each necessitated extensive research and development from the CG unit headed by Ditch Doy. When Houston and Docherty found out that they had to tackle their very first CG wolfman animation, their first move was to study previous attempts in other movies. “In all honesty, we were not all that impressed by some of the current CG wolves in films,” Docherty notes. “On the other hand, I really liked the wolves that ILM created for The Day After Tomorrow. This sequence became our reference. We also looked at the animation of the CG Morlocks in the Time Machine as a nice mix between human and animal behaviors.”

The wolf sequences presented two main challenges. First, the wolfman needed to be lit in the most believable way, but still had to be rendered in an acceptable time. The creature was modeled and animated in SOFTIMAGE|XSI, Peerless’ software of choice for character animation. Maya was used for everything else from falling rocks to explosions to particles to webbing. Since the wolfman actually is a lead human character, the CG model had to retain some of the actor’s features. To this purpose, Eyetronics scanned the performer’s head and the resulting mesh then became the basis for the skull of the wolf.

As for most of the CG animation, the render was carried out in mental ray. “Developing the fur was one of the trickiest parts of the project,” Docherty explains. “We actually had a [mental ray] R&D technician on staff at Peerless to refine the hair renderer and also to write an interface with XSI. The problem was that it was incredibly easy to make our wolf look cute, but very complicated to make it look scary… We found out with lead technical director Raffaele Fragapane that the best way to light it was to use high contrast rim lighting. However, this meant that the hair had to really work because, otherwise, you would get crackle and break-ups in the highlights. We ended up rendering seven to 10 passes per wolf. The sequence’s lead compositor Dennis Jones did a fantastic job of making the lighting work with the layers that we gave him. Every shot required its own approach. We would develop a lighting technique for one shot, but when we applied it to the next one, our wolf turned into a Teddy Bear! If one light was off or if the camera angle was different, it looked fake… We basically had to start from scratch for every single shot and all of them required endless revisions. It was very frustrating.”