Listening to Matt Johnson describe Cinesite’s efforts in Marc Forster’s World War Z, your first thought is, “Jeez, I wish I owned the coffee concession on that shoot!” Keeping the look of the film “photographed” rather than “rendered” required multiple crews working around the clock, complicated onset logistics, large amounts of data capture and ultimately, attention to details that no one in the audience would probably ever know or appreciate. The stated goal was always to make the visual effects seamless and invisible, the action grounded in real-world physics with a documentary-like reality. No simple task. Who knew creating a traffic jam on the streets of Philadelphia could be so complicated?
With over 430 shots primarily comprising the first half of the film, Cinesite’s greatest effort involved the recreation of various environments and action in and around Philadelphia. As the zombie infection hits the city, our hero, Jerry Lane, played by Brad Pitt, struggles to escape to safety, climaxing with his frantic efforts to get his family to a rooftop rendezvous with a rescue helicopter. Cinesite’s work included numerous aerial shots of the city in chaos, blocks of traffic jams, huge crowds running in panic, zombie attacks, scores of digital doubles, CG vehicles, a runaway garbage truck and other elements of mayhem and destruction often found with a zombie apocalypse and the explosive destruction of a city. Somehow, that brought Matt and his team to Glasgow.
I recently had a chance to talk to Matt about Glasgow, other locations and digital environments real and imagined, the enormity of the challenges faced as well as his desire to keep the film’s look based in the real world.
Dan Sarto: With the exception of the zombies, you’re recreating real cities, real environments. That gives you a great opportunity to create visual effects that aren’t obvious to the audience. If you’ve done your job, the audience isn’t thinking about the visual effects, they’re thinking about the characters, the story, the action.
Matt Johnson: My favorite visual effects movies are the ones where you don’t know there are visual effects all the time. My mantra on WWZ was I wanted it to look photographed, not rendered, trying to keep everything subtle and real. We wanted it so real that the audience would think they’re in Philadelphia, or a rooftop in New Jersey, not in Glasgow or in a green box. I didn’t want the audience to focus on the backgrounds. I wanted them to focus on Brad [Pitt] and Mireille [Enos] and what was going on in the story.
DS: What attracted you to this project?
MJ: What first attracted me to the project was talking to Marc [Foster] and the other creative guys about how stylistically, they wanted the film to be more like The Parallax View or All the President’s Men. I thought that was a rather interesting way to do a big sci-fi zombie genre movie.
DS: How early were you involved?
MJ: We got involved early and I was involved in most of the film. I was there from the early days of pre-production, well before shooting started, in terms of designing the zombies, in conversations on how they were going to move. There was a lot of thought going into the zombies, because, obviously, they’re a key part of the movie. Typically, in zombie movies, with George Romero, or in The Walking Dead, the zombies have a certain look, a certain way they move. That’s been established over the years in the making of these genre movies. But if you’re in a city, like Philadelphia, and you have vast areas of ground to cover, the traditional slow moving zombie wouldn’t work. We had to come up with something new and different. Everyone says that about their film, but I think we really did.
One of the most interesting things in pre-production was the involvement of experimental dancers. Avant-garde ballerinas. They can do the most astonishing things with their bodies. One guy, Ryan, is actually featured quite heavily in the movie, the guy who gets dragged out of his RV and transforms in the middle of a Philadelphia street, writhing his body around. Just looking at the way the dancers moved, the way they ran, we tried to design a new kind of language based on what these people were capable of doing. The one thing we found they couldn’t do in the way we wanted was lunge and bite. One of the key things in zombie movies is the bite, passing on the infection. One thing we looked at was, for example, if I’m leaping and trying to bite your neck, humans always put out their arms. It’s human nature, the reflex to brace for impact…
DS: …to break the fall…
MJ: Right. Of course, the key phrase there is “human nature.” These are zombies. They aren’t human. We saw videos online of attack dogs, German Shepherds and Alsatians that just leap, teeth first, at their target. I thought, “Wouldn’t’ that be cool for zombies to leap, to do that bite-first lunge.” Well, humans can’t do it. So throughout the movie, there are an awful lot of bites and leaps where you go from a human performer, whether a stunt person or experimental dancer, to a fully CG character for that moment of leap. We were trying to keep it looking realistic. You have to impart the correct sense of weight, and physics. This movie if firmly set in the real world. The laws of physics have to apply. We needed to make sure that in our animation, we were able to do that in a realistic way. We had to impart real world believability in that motion, even though for humans, it’s physically impossible.
DS: I’d agree. I was quite impressed with how real everything “felt.” Even the zombie attacks. Not that I know what that actually looks like. The first thing that makes me raise an eyebrow in a vfx-driven film is when the physics obviously don’t work. Tell me a little about how you animated and made the zombies?
MJ: There are some rather unforgiving scenes we had to do. For example, where the girl leaps and smashes her head through the windshield of the RV and drags the guy out. When the guys is being savaged on the ground, that’s a fully CG zombie. We had several scenarios where we had to cut between a real performer and a close-up of a fully CG character that’s supposed to look like them, cut back to a real performer, then cut back to CG. It was almost a visual effects nightmare scenario. We were so keen to not have them look like CG “things” that we had to go to town to make sure the digi-doubles were as realistic as we could get them. We had a whole scanning pipeline and an onset photography pipeline. If I thought a zombie looked “cool,” we had a system using color-coded pieces of paper that I would give to the performer. If it was yellow, that meant that zombie should be recorded for reference. If it was green, they needed to be photographed and scanned. We had this whole separate village setup where we were scanning 50-60 people a day. We needed that because we just didn’t know what was going to play later in the movie. While I was onset making the movie, there was this whole other team scanning and photographing and moving people on turntables.
Antony and his team had spent a lot of time working on flesh and hair sims. For example, the girl who is biting on Ryan, she has long dark hair which is flying and flopping around. You want that sim to be realistic and natural, obviously, but you can’t have her hair flopping over the point where the director wants to see the teeth biting the guy’s neck. I can’t tell you how many versions of that sim I saw.
DS: It sounds like a ridiculously complicated and involved production.
MJ: It was. But I was never aware of it because it worked so well. Onset we had two units working. Each had a dedicated group of trackers, match movers, things like that. The movie style is all hand-held. We’re zooming all over the place. All things that make our lives a little bit tricky. Each camera had to have zoom encoders built onto it. All the data from that had to be logged. We had teams running witness cameras. It was a 24x7 operation. When we finished shooting, the scanning team rolled in and worked through the night, scanning locations everywhere from the town center of Glasgow to the old streets of Malta. It was pretty busy.
In post, I spent my entire life locked in a darkroom [laughs] so I never got to see just how many people we had working on everything. Key people were Andy Robinson, who was my digital effects supervisor, Thomas Dyg who was my environment supervisor, Antony Zwartouw who was the CG supervisor. We were unified in our desire to keep it as real and subtle and photographed as possible.
DS: As far as the environments, tell us about the scenes in Philadelphia and the New Jersey rooftop. Why would you choose Glasgow of all places to transform into Philadelphia?
MJ: Essentially, Glasgow stood in for Philadelphia. Glasgow is similar to Philadelphia in some respects. It works very well on a “first floor” level. You go anything above that, it’s not America anymore. You don’t have the tall buildings or the skyscrapers. I was trying to create a weird fusion between the two places that kept some sense of reality. When I got to Philadelphia, I had in my mind a logic of how the two locations would work together. This is the square in Glasgow…this square in Philadelphia looks very similar. We made sure that when we were adding in buildings, we were adding in Union Station and other buildings that in reality would be in Philadelphia. The Comcast Tower and things like that are all geographically in the place they would be if it were Philadelphia. Hopefully, if you’re familiar with the city, it will kind of make sense. Nobody would probably ever care, but there was a lot of work making sure it all properly blended together.
Aviv [Yaron] and I went to Philadelphia and spent a great week on various rooftops. We’d be on the roof of the town hall, office buildings, places like that, texture photographing buildings, building facades, skycrapers. Aviv was on an endless series of scissor-lifts. We were always closing down sections of the road on Sunday morning, driving up and down working at various heights. Originally, there was the thought that we’d do all these buildings in CG and we’ll do textures, things like that. But when I was in Philadelphia, I said to Aviv, “That building is right here. It’s real. Let’s just photograph it and use that.” The environments team did something incredibly cool, which worked out really well. Rather than do everything in Maya, the traditional 3D way, all the environments, from Philadelphia to the New Jersey rooftop, are actually created in Nuke, using the 3D capabilities of what is essentially a 2D package. Again, my big thing is to make the film look photographed. So, we’ll start with photographs! We’ll use reality as much as we can. Real buildings and real photographic textures. Then we’ll put them onto geometry that we create within a 2D package. It looks like a real building because it is a real building.
There was a lot of work done by the matte painting team, a lot of re-projection work onto geometry. So, you have the ground floor of a building in Glasgow, but it becomes a real building in Philadelphia. Of course, there’s a lot of work done to make that blend seamlessly together. We got great feedback even from early tests. The filmmakers forgot that that what they were looking at was fake, which is a good compliment. There’s a shot where we’re looking down a street which is a canyon. Even the buildings right close up to the camera, they don’t exist. People said, “Where was that building?” Well, that building doesn’t exist. It’s in the computer. It’s not real.
We could create the canyons of Philadelphia and then in CG add in cars and cabs and fire trucks, crowds, things like that, to give that sense of scale. We had scenes where 50,000 people would run across the square. We used massive crowd sims. For example, there is a shot where you’re looking out the back of a helicopter as you’re flying over the city. So, for a shot like that, which no one would ever know, every single road in the live-action helicopter plate was digitally replaced. All the real moving vehicles that were shot at rush hour were removed. All the streets were painted clean, and then re-projected back onto geometry. Then we added CG cars jammed in traffic. We wanted to show the city gridlocked. The whole idea was the city was at a standstill. We didn’t want to see even in the background any cars moving. We had to remove all the traffic from the Ben Franklin Bridge and replace it with CG objects we’d created.
DS: I understand you used a lot of LiDAR scans.
MJ: Yes. Basically, we pretty much LiDAR scanned Glasgow by the time we finished with it [laughs]. We had a lot of reference of what the Glasgow buildings were. Aviv had a huge setup, like a robo-head, with a digital panorama camera, doing HDRI, multiple exposures on everything so you got the full dynamic range of the image. Aviv and I would go off and do these huge panoramas of the environments, detailed textures of buildings right down to the window level.
One of the environments I’m most proud of in the movie, skipping ahead from Philadelphia a bit, is when Brad and his family are being airlifted to safety and they have to make their way to the roof of an apartment building, which is in downtown New Jersey. That was the visual effects nightmare scenario. That essentially was done with a green box. It was a small roof-set piece, with just the top level they were standing on finished. Everything else was a complete 360 degree digital environment. Everywhere the camera looked, it was looking into a greenscreen. We had to create a world that doesn’t exist. There aren’t any housing projects like that in New Jersey.
So, again, what Aviv and I did, when we finished in Philadelphia, was spend a week in New York, creating a hybrid world that doesn’t exist. The building that Brad and family are on and the buildings immediately surrounding them are based on a housing project in Harlem. Those had interesting architecture. We were on various rooftops in Harlem, photographing different buildings. We used scissor-lifts, things like that, doing flat on camera textures of building facades, air conditioning vents, looking over, looking down. Other buildings in the shot are actually from the lower east side of Manhattan. We photographed a housing project there because it had some interesting shapes. Then, we did go to New Jersey. Our location scout got us onto the top of a tall building, so we were able to take these huge panoramas of the real New Jersey. So, in our shots, we did kind of a sandwich. In the background, it’s really New Jersey. The skyscrapers and various landmarks. In the midground, it’s the lower east side of Manhattan and those buildings. Then, immediately surrounding Brad and his family are buildings from Harlem.
Andy Robinson and the 2D guys were always thinking very hard about depth of field, the focus. He was completely obsessed with making sure the depth of field and stuff like that was right, to make it feel that it was photographed. We even swiped chromatic fringing onto things to make them look as if they’d been shot with a lens. Lens flares, things like that, we set it up so that we could have some cool shots looking into the sun. It was our world. We could do what we wanted with it. We photographed real lens flares rather than use digital ones and then manipulated those to get the nice striations and color balancing. And when they were occluded by the objects we would dim them in comp. All those things, which again, no one will ever know, we did just to try and keep it grounded in the real world.
DS: This was an expensive production. There were a large number of locations. You had different crews running about. You’re there to capture what’s needed for the visual effects and hopefully do so in a manner that minimally impacts the rest of the crew. Tell me about the onset dynamic. How did you manage such a complicated onset production?
MJ: That’s a key and crucial element. I did a lot of work with Simon Crane, the second unit director, working on the garbage truck and big action stuff. In the old days, if you went in there as a visual effect supervisor, it would be like, “OK, it’s a visual effect shot. Everyone back away from the camera now! You can all just stop. We’re going to take over now.” If you did that today, you’d be fired in less than five minutes. On this film, it was nice being part of the creative team, working with Marc and Simon, to contribute and obviously not get in the way. On a movie like this, you need to know which battles to pick. For the visual effects, there are some things we need to do and make sure that they happen. There are other things where you just roll with it.
For example, in Glasgow, in order to allow the ease of tracking later, I had a team of people in different hotel rooms and offices around the square, shooting with witness cameras. I’d be onset talking with the director or dealing with the ADs, but there were all these other people with their witness cameras making sure they had multiple lines of sight on where the camera crews were located. One of the key things on this movie, because of the hand-held nature, we had physically mounted wide-angle lens reference cameras fixed onto the production cameras. For example, if the camera had a 200mm lens or was zooming around quite tight, our digital video camera always provided us a fixed wide reference point which we could synch with the live action. So even though the live action footage showed a close-up of someone’s face, the reference camera would be showing the doorframe or the wall or something behind it. Those types of setups were key in allowing the movie to keep shooting, be fluid and have the hand-held style that everyone wanted. But, we knew we had all this information in our back pocket.
DS: Ultimately, what was your greatest challenge on this film?
MJ: For me and for the team, it was just making sure people didn’t have a clue what we were doing. To be true and faithful to the style of the book, to make sure our work fit in within the reality of the movie, so it didn’t draw attention to itself. Yes, there is some very cutting edge, very sexy visual effects work in it. But it’s not a film about visual effects so much. You can do big flashy visual effects that work beautifully on their own. What’s harder in some respects, and even more rewarding to me personally, is to do stuff like that but keep it completely stylistically locked within the style of the film you’re making. In visual effects, we can do anything, as you well know. But, it’s not a space movie, it’s not a superhero movie. It’s more grounded than that. We were all just focused on trying to make it look real.
Dan Sarto is editor-in-chief and publisher of Animation World Network.