‘Avatar’s’ Effects Whiz Talks Shop

Oscar-winner Joe Letteri on creating new worlds from scratch, and dealing with James Cameron

Last Updated: December 31, 2009 @ 12:17 AM

He helped create Gollum and King Kong and the dinosaurs in “Jurassic Park,” winning four Academy Awards along the way. But nothing Joe Letteri had done quite prepared him for the scale and scope of “Avatar,” in which the Weta Workshop partner led a team that created the alien landscape of Pandora and then filled it with strange flora, fauna and a race of 10-foot blue people called the Na’vi.

Director James Cameron’s blockbuster hit is a groundbreaker visually and technologically, with Letteri and Weta exploring new techniques during three years of painstaking work. Along the way, Letteri also found time to check in with a couple of other Weta projects, including the summer hit “District 9” and Weta founder Peter Jackson’s new “The Lovely Bones.”

Letteri sat down with theWrap the week he was in town for the “Avatar” premiere. It marked the first time he’d actually seen the film all the way through. (Letteri photo by Marty Melville/Getty Images Entertainment)

So what was it like to finally see the finished movie?
It was great. The thing I thought was most successful was the editing. I thought it really moved along and kept you engaged the whole time. I saw a rough cut about four months ago, and that was a good 45 minutes longer. And we hadn’t really even started working on the third-act battle scene.

That’s an awfully big piece of the film to leave until the last minute.
Oh, it’s typical for these movies. No matter how you try to plan, it all happens at the end. “Lord of the Rings” was the same way. “Kong” was the same way.

Was it a no-brainer when Cameron asked you to do “Avatar”?
In terms of wanting to do it, yes. Jim had this idea that was almost like “The Wizard of Oz” – what if we just go to this whole other magical place, where the people are not the people but we still recognize things about them? But when it came to sitting down and figuring out if we could actually do everything we had to do to make this movie, that took a little more brain power.

The line I keep hearing is that Cameron had to wait until technology advanced to the point where it was even possible to do what he envisioned.
That’s fairly true. It took, probably, another year beyond the point when we started making the movie to where we actually could make the movie. We’ve been laughing about that for the last few weeks: “Okay, now we know how to make this movie … but we’re almost done!”

What were the problems that you hadn’t solved when you started?
Everything. We started completely from scratch. We just ripped the whole engine apart, basically, and said, “Okay, everything we’ve been doing for the past 20 years, just assume its wrong and find where the flaws are.” Because we’ve got to be able to do more of it, it’s got to be better and faster and more flexible than we could ever do before.

Dealing specifically with the character animation, and the things that Weta did with Gollum in the "Lord of the Rings" movies — what parts of that did you have to throw out?
We kept the same basic idea, of trying to figure out from the expression that an actor is making what their face is actually doing, what the underlying muscles are doing. When we animate, we’re not trying to say “This point on the face moves this far, so we’ll move the spot on Neytiri’s face this far.” We’re trying to figure out, if this dot, this dot and this dot moved this much in that frame, what did the muscles underneath the skin do, and what does that tell us about the emotion at that moment?

That was a process we started with Gollum, and carried through with Kong. Gollum was all key-frame animated, because we had no way to do motion capture on the face. With Kong, we came up with a way to glue markers on the face to track the motion, so we had a more accurate representation and we were able to write software to solve what the muscles were doing.

But “Avatar” didn’t use dots glued on the actors’ faces.
No. For “Avatar,” we needed something that was quicker than all the markers on the face, so Jim came up with the idea of having a helmet-mounted video camera. But we needed a lot of software, then, to figure out what the face is doing, and how you translate that back to what the muscles are doing. And we applied that not just to one character, but to seven main characters with speaking parts, and other 14 secondary characters that also had some dialogue, and then another couple hundred for the tribe. They’re all singing and chanting, and we never knew where Jim was going to move the camera.

How did you make sure the avatars, the human/Na’vi hybrids, kept the actors’ performances intact?
One of the design elements was that we tried to use the mouth area of each of the actors who was playing the Na’vi, and then we blended that into the elongated face and the big eyes. But the eyes still maintained the character of each of the actors as well. We just worked the idea of lions’ eyes into their eyes, so you still had the resemblance. And that gave us something that we could really key off of as we were looking at the performances. You could feel what the actor was doing.

Did the actors always wear helmets with video cameras aimed at their faces?
Pretty much all the scenes where the actors were performing as Na’vi or avatars, they had helmets on. There were a few where they didn’t, like the kissing scene, where we had to take the helmets off and just key-frame that one with lots of reference cameras all around. And obviously for a lot of stunt work you couldn’t have the helmets on. But our animation system is set up where, whether the information is coming from a motion capture or an analyzed face, or whether it’s being done completely by the animator, or anywhere in between, it’s all done by the same set of controls. We can get a performance from anywhere along the spectrum, analyze it, and solve it, or animate parts if the solver’s not working.

Cameron is known as being a very demanding director. Was he constantly asking you to redo things?
Yes, absolutely. We did a lot of trying things to see if they’d work, and we’d even get some things pretty far along when Jim would say “Sorry, it’s really not working.” There were times when we almost started all over again. And mostly, those were creative and aesthetic choices more than technical ones.

When you’re working in a completely virtual environment, you’ve got nothing to peg your shot on. If you’ve got a plate, you can say, “Okay, I know what the time of day is, I know what the sun is doing, here’s my sky, I’m going to put a character into it.” Whereas if you’re starting from a blank slate, everything is completely open. What time of day is it? Is it three o’clock, or 4:30? Is there a cloud in front of the sun at this moment, or not? What does that do to the light coming through the leaves and the branches? If you’re shooting, you’re stuck with what you’ve got. But if you’re not shooting, you have to figure it all out yourself.

Since the film is set in an alien world, I’d think not only would you have to decide where the sun was, you’d have to decide how big and bright a sun it is.
We had that discussion early on, and we decided to treat it as if it were about the same as the sun on earth. And we treated the sky as blue and the clouds as white. The problem is that if you go too far with making things alien, the audience doesn’t have a point of entry. If the sky is purple, is it just going to look wrong at some point? Is anyone going to understand that there’s different colored gas in the atmosphere? It begs too many questions and gives you too many things to think about.
So what we tried to do is to create really natural touchstones wherever we could, and then add the alien elements. The idea was to make sure that people had enough to know that they’re on an alien world, but not enough to pull their attention away from the characters and the story.

Some people have speculated that if Fox had wanted to enter it, “Avatar” probably could have qualified for the Oscars as an animated film.
Yeah, we were just talking about that. I don’t know what the rules are.

I think it’s that a significant number of major characters need to be animated, which obviously they are in “Avatar.”  And that something like 75% of the shots in the film must have an animated component.
I don’t know if it would qualify or not, given the running time. I’m not speaking for Fox here, and I don’t know what they’re thinking, but I never got a sense from talking to anybody that they were interested in that. This was always a drama. You’re in a world that you have to create visually, but the intent is the same as if it were photographed.

How did you coordinate with the other effects houses working on the film?
Weta did all the main character work, and we were hoping to do everything else. But we always left as a safety valve, other things that could be split out if they needed to be, especially things not involving the Na’vi. And there was a point where the story got bigger and longer we just weren’t able to gear up as fast as new scenes were coming in.

So ILM took a big chunk of the floating mountain work. In the big battle, if you’re looking at the armada, ILM did most of those shots. If you’re looking at the Na’vi, we did those shots. It bounces back and forth, but I thought it worked really well. And if I understand right, there were about nine houses working on it doing various things.

Did you also have a hand in “District 9” (above) and “The Lovely Bones,” which were in production at the same time?
We did have supervisors who oversee everything on a day-to-day basis. But because I’m one of the partners in the company now, I reviewed everything on a regular basis, just to keep an eye on things.

So you had the time to do that?
No. (laughs) But I did it. Funny how that works.

On “District 9,” did it complicate matters that many of the effects shots are seen in bright daylight?
Not really. Because you had plates there, it was pretty straightforward. Artistically it’s still challenging, but because you have a plate, you understand what you need to do. As you said, it’s brightly lit, so you just have to make it look like you’re seeing the spaceship in that same lighting.

How about “The Lovely Bones" (below), where you’re creating fantastic backgrounds that are a key to the emotional state of the main character?
“Lovely Bones” was surprisingly similar to “Avatar,” in that you’re kind of starting with a blank canvas. Even though you had Susie (Saiorse Ronan) there, a lot of times she was just shot on a blue screen or a green screen.

It was really interesting conceptually, because the world she’s in is kind of her projection. She’s building this world around herself where none exists, and yet she’s not really building it – she’s not in control of it, but it’s coming from things that she knows. It’s almost as if this world is kind of reaching into her brain and trying to construct itself to fit her emotions in the moment. So we were always looking for what was actually happening in the story at each point.

After “Avatar,” where do you go from here?
We’ve got a couple of films we’re working on next, “Tintin” and “The Hobbit,” where we’ll use a lot of the same technique. I don’t know if we’ll have actors running around wearing head rigs or not, but we’ll be using a lot of what we’ve developed in terms of how to do big environments. We’ve got a real strong physical foundation for how we’re going to do creatures and characters and lighting and musculature and facial animation.

Is it safe to say they won’t be as big as “Avatar”?
You never know. (laughs) We’re still in the early stages, but they very well could be. In both of them, we’ve got big fantastic worlds we have to create.

CORRECTION: The original version of this story misspelled the name of Zoe Saldana’s character, Neytiri.