Sony Pictures Imageworks animation supervisor David Schaub talks to vfxblog about his work for The Chronicles of Narnia: The Lion, the Witch and the Wardrobe.
The characters are quite amazing. Why do you think the characters are so successful in the film?
Well, I’d like to think they are the successful. I think it goes back to the idea that if they’re talking animals you make sure they don’t look caricatured. We’ve gone out on a limb and tried to make them as real as possible. The wolves, for example, need to have all the nuances and mannerisms that you see in real wolves, but timed in a certain way to support the dialogue. As soon as you start going too humanised, you’ve got a cartoon on your hands.
Then with the beavers – they are a little more stylised and human-like. They’re not really built like a beaver. If you look at real beaver anatomy, they’ve got very short arms and their bodies are essentially just a bag of fat, but we needed to do more with them. If you look at a beaver standing on his hind legs, it looks pretty strange. They’re got the short arms and balance on the ground with their tail, which means their upper body tetters in a very strange way. We took those liberties in design to make the beavers more human-like in their performance.
What kind of toolsets did you have to develop for this show?
Our facial system was fairly new for Narnia. In the past we’ve always used a blend shape system, in other words, we used model shapes that we just morphed from one shape to the next. We used to have a selection or list of pre-modelled face shapes to choose from to go back and forth between. This time we used a muscle-based facial animation system, which we had used for the first time on The Polar Express. The Smokey and Steamer characters were our keyframed characters in that show, with all the others motion-captured. The idea of that show was to have those guys mo-capped as well, but of course Michael Jeter died during the show. So we didn’t have the data from the motion-capture to finish the animation. So the whole thing was stripped down and we employed the muscle system being built for Sony Pictures Animation (for Open Season and Surf’s Up). It was given some great results and you could get some really expressive movements out of it. You weren’t limited to the poses that had been pre-built. You had this intelligent muscle control system that let you pull things and stretch and inflate while retaining the volume. On top of that you had the additional controls to make fine adjustments, whereas with blend shapes you’re pretty much stuck with the shapes that had been pre-modelled. With the muscle system you can have main shapes in a library but still have the ability to nudge muscles as needed.
The muscle system was used on all the characters. It was a little tricky on the wolves because of the muzzle. You’ve got this long snout which, when we did it on The Polar Express, it worked better for a human face, or was at least built with a human-type face in mind. When we put it onto the wolves, it had to be re-built a little, but the principle is all the same.
One thing I liked about the wolves was that real ones interacted and appeared with digital characters in the same shot.
I think the idea was to do a lot more of that. They spent a fair amount getting the wolves down there and training them up. But, like always, there are ones along the way that didn’t do what they needed – either they just weren’t angry enough or didn’t jump in the right away. As we showed them what we could achieve digitally, we basically stopped using the real wolves. Basically after the attack on the Beaver’s hut the wolves are all digital. The introduction of Maugrim in the White Witch’s castle is all digital. When the wolves run out of the castle those are live action with a couple of CG ones inserted. The Beaver’s hut is a mixing and matching of live wolves with CG ones, and sometimes live wolves with CG parts like augmented tails. Beyond that, the CG wolves worked well enough that it made sense to use them for the rest of the film. It’s always a compromise with real animals to get them to do what you want – you do get happy surprises sometimes. Andrew Adamson is big on performance and has really sharp instincts and desires in that area. When you can design a performance and make it work as you see it in your mind, I think that’s the way to go. Of course, this meant our shot counts just kept increasing all the time.
Can you talk about the creation of Mr Tumnus?
The actor wore greenscreen stockings. There was some motion capture data from the character on set. We used the motion capture for that character to help us get the hips where they needed to be. When there was a lack of survey data, the motion capture came in handy to work out where the hips were in 3D space, and it helped us get about 50 or 70 per cent there. Then our matchmovers took over and lined up the hips. They needed to be aligned with a really high degree of accuracy, because obviously any swimmyness or sliding in there would really show up with the mix of practical and digital fur. Once the hips were aligned it was a task of actually animating the legs. Even though there was motion capture for the leg, it really didn’t apply because it was a goat leg that we were after. Once we got that done, it was pretty simple process. Once the hips are where they needed to be, you just place the legs underneath them, using the proper compression between the heels and the hoof, so that you get the proper weight.
We accomplished Mr Tumnus on a shot-by-shot basis. Every shot kind of had its own problems. There were only 30 or so shots, but we worked on Mr Tumnus throughout the entire duration of the show. We started with him and ended with him – there was no mad rush because we knew we could do it – we just chipped away at it over time.
I liked how the fox emotes through his tail and ears and eyes. How did you approach the animation for him?
I think the fox is one of the more interesting canine characters because there is a lot more going on there. I should mention that 90 per cent of the fox shots were animated by one guy – Patrick Osborne. He came up with some really nice ideas. The wolves tend to be the hoodlums. We just needed to make them mean for their personality. The fox is more calculating. You see so much of what’s going on in his thought process through his eyes. There’s a lot of eye adjustment when he’s thinking. When Maugrim grabs him and asks where the kids are, you can really see that thought process. He doesn’t say anything, but you can just see the attitude through the ears. Is he going to give them up or isn’t he? The ears were tricky to rig because they can physically rotate 180 degrees. To get that to work without breaking the rig was a bit of a challenge. We did try to get the fox to emote with his paws but it didn’t look right, so we used the tail a lot. It’s so big – it’s a huge fluffy tail – which makes it hard not to use. Using it to make little flourishes does kind of replace hand gestures. Trying to get that reality factor in there was important. The body is constantly in motion, the breathing is tied to the dialogue and you feel it throughout the performance.
What were some of the other shots Imageworks completed for the film?
There was the whole waterfall sequence. That was made up of many, many different elements. If you look at what we started with – in raw form it’s bits and pieces of the kids, greenscreen all over the place, miniatures, CG water, CG ice, crashing waterfall and other stuff. We also did the opening bombing of London, which is entirely digital except for some shots of the German pilots. The professor’s house is a digital house that doesn’t exist. We did the ice castle and the shots around it as the kids are watching Edmond go into it. It was all kids in bright sunlight against greenscreen, so all the backgrounds and snow had to be added digitally.
The interaction between studios was pretty straightforward. We had to add the beavers in some shots there. The shots were rendered in layers, so the studio delivering the shot would be the studio that had the foreground characters. For example, ILM might have done the backgrounds, and then R+H had the next layer and then we’d finish it up with a beaver in the foreground, using their plate. The only real physical interaction we had was when Aslan pins down one of the wolves.
It was all such a huge undertaking with so much being digital. I thought at the beginning how on earth were we going to do this? And, just like always, we got it done.
[VFXblog]