Search form

John Knoll Talks 'Mission: Impossible - Ghost Protocol'

ILM's VFX vet tells us about returning to M:I and tackling Brad Bird's first foray into live action.

Brad Bird gets his shot at live-action action with Mission: Impossible - Ghost Protocol. All images © 2011 Paramount Pictures. All Rights Reserved.

Ordinarily, John Knoll, ILM's senior visual effect supervisor, wouldn't have returned for the fourth installment of the M:I franchise (he worked on the first back in 1996). There are only 700 VFX shots and most are invisible. But this offered the opportunity to work with Pixar's Brad Bird, who was making his first live-action movie. I spoke with Knoll about taking this mission.

Bill Desowitz: You've known Brad Bird for quite some time. But what was it like working with him?

John Knoll: He has all of the important characteristics of a good director: strong story sense, able to get good performances from your actors, and, at least what's' important from my point of view, he's got a good eye for shot composition and visual storytelling. And so, all of those features were on good display on this project. I figured that he would be surrounded by various folks with live-action experience that could get him through the things that are new to him. But the things that are hard to teach he already has very well mastered.

The Burj sequence makes Tom Cruise's Ethan Hunt look like Spider-Man.

BD: What was the hardest learning curve for him?

JK: Probably the biggest part of the learning curve for him was the harsh reality of having to commit to things right then and there when you're shooting live action. And in animation you can often defer decisions or make changes later. After you look at the first version of a scene and make a whole variety of tweaks and changes to iterate on it to get it to your ideal state. And when you're shooting and you only have what you shot to work with and that can be a stressful thing to carefully think everything through.

BD: The Burj sequence is probably the most publicized because of Tom Cruise being such a daredevil up there, but what kind of support work did you provide?

JK: The work there was removing safety equipment, so they erased the cable and the reflection of the cable and IMAX camera. We also shot a glimpse of the Burj ground level for safety and logistical reasons and did CG extensions and the background environment.

BD: What about the approaching sandstorm?

JK: That was done in the Singapore office using our Plume GPU-accelerated simulation rendering application, which is a fantastic tool. It lets you iterate quickly and then generates spectacularly good results. Plume has dramatically changed how we do those fire and dusty things.

The sandstorm sequence posed a challenge for the vfx crew.

BD: And what about during the actual chase in the sandstorm?

JK: Once we are inside the sandstorm, most of the work is evening out the density of dust. They did try and have as much practical in-camera as possible, but there are times when you see blue sky on top of it and some grip equipment in the background, so we're augmenting dust to hide that sort of thing and make it appear to be a consistent level. And then about half-way through, it switches to a car chase, where Brad really loved the idea of the contradiction of a car chase in dense sandstorm where it's too dangerous to drive fast. And you think about the logistics of shooting that. Again, they tried to get as much practical dust, but it's a little too dangerous to do a huge amount for the same reason it would be in reality. So there was lighter dust for the driving scenes and so we had a little more augmentation of the dust level for that last half of the scene. They used ground paper pulp. We also used a lot of dust elements in our stock library of blowing dust. And in some cases it's just noise patterns in the comp. If you think about it, once you have really dense dust, it's everything converging on a flat color and you just want to be able to see a little bit of movement. It's actually low-contrast and you can get away with a lot.

BD: Talk about using the prototype of BMW's new Vision hybrid.

JK: We all decided that since there is a drivable prototype of this car, we would try and shoot it with a real car. And we were warned by BMW to be careful with this and not take it over 40 miles an hour; treat it gingerly. But it's meant to be a high-speed chase and we try to give the impression of speed with long lenses and tight framing and lots of quick cuts. But, despite the best of intentions, after the first evening of shooting, we catastrophically broke the transmission and it couldn't be repaired in time for us to continue the shoot. And this is the only one that existed so it had to get shipped back to BMW. And, meanwhile, we proceeded to shoot the scene with a target car and it ended up being computer-generated in a little over half the sequence. Ideally, you will never know that you're seeing a computer-generated car. It was animated in Maya. We started by matchmoving our target car and that was the basis of the car, and then we had an animator go in and put suspension squish and tire rotations and steering and minimal animation on digital doubles who you can see in the windshield.

The film concludes with a thrilling chase through robotic parking garage.

BD: And the thrilling climax inside the robotic parking garage?

JK: It's modeled after the Autostadt in Wolfsburg, a big cylindrical building, 20 stories, with 18 bays around, and then a central column with two robotic paddles that can come up and park cars in those bays or retrieve them. And so we built a set [in Vancouver] that was two stories high on one side and five stories high on the other and tried to get as much of that in camera as possible, but there were a fairly large number of shots where we sawed off the set in one way or another. It was a series of computer-generated extensions for that.

It was our first foray into production with Katana (the scene management and lighting tool), rendered in Arnold [both developed by Sony Pictures Imageworks]. Lighting in Katana is a different way of thinking because it's all about deferred loading and it's a big node graph-based application that is like Nuke in a lot of ways but for 3D objects. And so you have a lot of operators that you can throw into your tree to do various things. There's a lot to like about that approach and the guys that were using it were super enthusiastic about it. We're kind of moving that way as a company. But we limited our exposure to Katana to this one sequence. I wanted to use a ray tracer as well because it has all these cars, which are computer-generated vehicles, and I wanted to do it with a tool that was primarily a ray tracer designed to do this sort of thing. So it was a dual task: it was Katana driving Arnold.

Bill Desowitz is former senior editor of AWN and editor of VFXWorld. He has a new blog, Immersed in Movies (www.billdesowitz.com), and is currently writing a book about the evolution of James Bond from Connery to Craig, scheduled for publication next year, which is the 50th anniversary of the franchise.