How many times does an editor watch the movie they’ve made? How many times have you seen Life of Pi?

I’ve probably seen the finished film 30 or 40 or 50 times. The thing about movies nowadays is that there are a lot of versions you have to make and a lot of time required to go through them. We have 3D, 2D, IMAX, film, 2D digital, 2D film, and for audio there are 7.1, 5.1, Dolby Atmos, near-field for TV speakers, and more.

So it’s up to you to pick the best shots, and decide how long to hold on Pi’s face as he reacts to something?

There are hundreds of little decisions you make per day. You just do it. You do what feels right. You can think exhaustively or you can go by feel. Ultimately, you develop instincts about what things work. When you first see the footage, if something is really striking you have to really trust that. It’s not a science. Sometimes things work in ways you can’t really quantify. Sometimes I show it to someone and they say “wow” even if they can’t say why. It’s important to try a lot of things and find the relationship between this look and this reaction, and that line and this hesitation. My assembly process isn’t just about getting one version. It’s about exploring what’s in the footage and trying different things.

“WHEN YOU FIRST SEE THE FOOTAGE, IF SOMETHING IS REALLY STRIKING YOU HAVE TO REALLY TRUST THAT. IT’S NOT A SCIENCE.”

What was the hardest scene to cut in Pi?

Very often in a film, one of the scenes you spend the most time on has no visual effects at all. For me it’s the scene where Pi’s family is having dinner and talking about religion versus reason when Pi is 11. That scene we worked on a lot — changed it, rewrote it, reshuffled it, and shortened it. We decided early on that we were going to drop some of it, but what lines do we drop, and what lines do we put back?

Every version of that scene was really good, but it just took a long time for it to reach its final form. Some visual effects scenes, like the shipwreck scene — which is a big long scene, and took them a long time to do that in bits and pieces over a month — basically didn’t change. We got that right, and got it into its final form really early. There were enormous technical challenges in a scene like that, and in the flying fish scene, but the work that we did was about visual effects, and not re-cutting very much. I swapped the order of two shots in the flying fish scene, and we did that very early on. It was enormously challenging, but not in the normal editorial kinds of ways.

The hard part in Pi was the large scale stuff. The hard part was that the film doesn’t have this strong narrative pulling you through. It’s not a caper movie. You don’t want your audience to feel like they’re lost at sea. You want to feel a plot pulling you through, which there is, but not in the way of a lot of movies.

What type of challenges does 3D present for creating special effects?

Visual effects and 3D are really hard. Everything’s much harder. 2D is a lot more forgiving — even just with simple blue screens. Pi’s on a boat that’s moving around, and he’s got wild hair. Often when you’re working with a blue screen, if you don’t get every hair it doesn’t matter much, but in 3D his two eyes have to match and they have to match perfectly. If they don’t, you’ll notice. In order to avoid motion blur you might shoot with a wider angle, defining the edges on both eyes and making them match. If there’s something you have to paint out in 2D it’s nothing. In 3D, you have to have the whole space mapped and track it equally in both eyes, so it takes 20 times as long. Things that are relatively simple in 2D are a nightmare in 3D.

“THINGS THAT ARE RELATIVELY SIMPLE IN 2D ARE A NIGHTMARE IN 3D.”In our film, the goal was to make everything look real. Our tiger is not a cartoon tiger or a fantasy tiger. It has to look real, and that was one of the problems with getting this film made. You can’t make this movie until you can do a photorealistic CG tiger, and that’s really hard. In the film there are 23 shots with a real tiger. In some cases it’s intercut with a CG tiger, which set the bar really high.

We used that for scene when Pi was training Richard Parker with a stick. We shot the real tiger first for 4.5 hours. I went through that footage myself, sat down with the tiger trainer in Taiwan, and he told me what everything means in psych terms told me about what the tiger is thinking. Then we worked out what the scene could be, and how the scene could be structured given what we had.

At one point the tiger sits down scratches his claws on a hatch below him. The tiger decided to do that. It decided that’s where it would scratch its claws — that’s a real tiger shot. Our tiger trainer explained that that’s a nervous tiger trying to pretend it’s not nervous. It’s saying, “See, you’re not intimidating me — I’m gonna scratch my claws like I don’t care.” We included that because it was just stunning. Three weeks after we shot that part, we shot the same part with Suraj. He was there with our animation supervisor giving him an eyeline and swatting at the stick, while I was in the back of the boat narrating the scene.

Will all movies be 3D in 20 years?

3D is tough. Back when color first came around, some people said, “Well, color makes sense for some kinds of movies,” and then it became standard. You only did black and white if you had a reason to do it. 3D is a bit different since it requires more in production — quite a bit more. Cameras are harder to work with, and it’s a lot fussier. Lens changes takes three times as long as it would in 2D. There are expenses and complications involved in shooting 3D, whereas with [the rise of] color you just put color film instead of black and white film in the camera. The price you pay as a filmmaker is greater in 3D, and that’s one of the things that will limit its adoption for some time.

Is it more difficult to distribute 3D films?

A big issue with 3D is most places can’t get enough light on screen. You’re projecting through something that takes half a film’s light away, and then through glasses which takes more light away. That’s a necessary part of the technology, but you need to run bulbs bright — and some theaters run bulbs darker to make them last longer.

“YOU’RE ASKING YOUR EYES TO DO THINGS THEY WEREN’T EVOLVED TO DO.”What about 48 FPS, the new medium Peter Jackson’s experimenting with on The Hobbit?

What some people are saying is that it looks too real. You see guys in costumes instead of elves and dwarves. Film benefits from a certain level of abstraction. 3D is a way of taking away abstraction, and it’s unclear whether audiences will warm to that or not. Americans are among the slowest to adopt 3D, actually. The Chinese don’t want to see anything else. There’s a cynicism about 3D which is partly the result of films where 3D was done poorly, and that 3D has been done on action movies. And studios shot themselves in the foot by raising prices on 3D. A lot of people will choose not to spend the extra money, which is a shame when we put so much effort into it, and they just see it in 2D.

What’s the toughest part of creating a CG tiger — a tiger that so much of the film focuses on?

Even people who haven’t studied tigers understand physics. You just know how something that big moves. If the tiger hops down, how does its weight settle down in a way that looks real? When it steps how do its feet compress as weight comes down on them.? If it looks wrong, you know it looks wrong, but you might not know why. When you start with animation you begin with the anatomy of a real tiger, with real bones, real muscles, and the way skin hangs on its fur. You do it through simulations. If you had to animate everything, there isn’t enough time left in the life of the universe to do it right. Some of the animators are geniuses in stuff they came up with, and a lot of it was driven by footage of real tigers, like when the tiger is uneasy and turns its head, what does it do with its ears and tail? When irritated, how do tigers hold their tails when they’re walking?

“IF THE TIGER HOPS DOWN, HOW DOES ITS WEIGHT SETTLE DOWN IN A WAY THAT LOOKS REAL?”Will digital actors ever replace human actors?

It’s doable, but the hardest part is the face. Body motion we can do fairly well, because to some extent with body movement you can actually just use motion capture data or just photograph people. There are so many muscles in the face, and humans are so adept at interpreting those movements. As social animals we are acutely sensitive to every nuance and muscle in the face — especially around the eyes and mouth. With a tiger we can get away with it is since they’re covered in fur, and don’t have the same degree of expressiveness. Without the fur it’s really hard, but it’s not impossible.