Exclusive Interview: 2012’s Visual Effects Wizards Tell Us How They Destroyed the World in 1,315 Simple Steps (Part 1)

Share on Facebook posted 11-09-09 by John Freeman Gill

It’s the end of the world as we know it. Again. Serial disaster fetishist Roland Emmerich, who blew up the White House in Independence Day, stomped the Empire State Building in Godzilla, and trashed Tokyo with football-size hailstones in The Day After Tomorrow, is back for another round of planet-savaging mayhem in 2012. And this time he means business. Apparently convinced that the past few years’ terrorist attacks and global economic crisis haven’t made us anxious enough, the German impresario of destruction has taken it upon himself to show us what the apocalypse will look like in three years if an ancient Mayan end-of-world prophecy proves accurate. Mark your calendar.

2012, like Emmerich’s earlier cataclysmic visions, will feature some pleasant and reassuring faces, in this case those of John Cusack and Amanda Peet. But let’s be honest: The real stars are the visual effects. However thin the characters in 2012 may turn out to be, the film provides an eye-popping showcase for the latest advances in VFX technology.

So just what does it take to flood the Himalayas, topple the Vatican, and flatten the White House with an aircraft carrier borne by a mile-high tidal wave? Volker Engel and Marc Weigert, co-producers and visual effects supervisors of 2012, give John Freeman Gill the lowdown.

Blowing Up Models Is So 1996

30 NINJAS: You both worked on Independence Day, which when it came out in 1996 was probably considered the cutting edge of both visual effects and disaster films. How has visual effects technology evolved since then, and what did these technological improvements allow you to achieve in 2012 that you couldn’t in Independence Day?

VOLKER ENGEL: To start with a statistic that is very interesting, as Independence Day is always deemed a really big visual effects movie, we had about 400 visual effects shots on Independence Day, and on 2012 we have 1,315. Which comes down to a screen time that is half the length of the movie. So there’s so many more possibilities that we have these days, and just technology-wise, the biggest difference is that Independence Day was one of the last really big model-miniature films, where we physically built miniatures and had these destruction sequences, where we just blew up models and destroyed them on stage. That’s definitely one of the biggest changes, because I literally spent nine months shooting just the miniatures on Independence Day. On 2012 I think I was involved in only two days of shooting some insert pieces for miniature bits, and everything else was created digitally.

MARC WEIGERT: What we’ve done on 2012 wouldn’t have been possible about five years ago. And, of course, the development of computer hardware and software plays a huge role. The reason we literally had to push miniatures out of the way, although both of us love it — and really, nobody likes to sit in front of a computer monitor for the whole day for 12 months, so obviously it’s so much more sexy to actually go out there and blow something up. But we didn’t have the choice, really, for two reasons. One is the fiscal reason that miniatures now in most instances are becoming more expensive than Computer Generated Imagery. And the second thing is for physical limitations.

For 2012, you probably have seen the sequence that we’ve done in-house where we’re driving with the limousine through Los Angeles while literally the whole of the city is breaking left and right by a 10.5 earthquake. If you look at the sheer amount of stuff that has to be built there, it’s literally dozens, possibly even hundreds of houses in a residential neighborhood: trees, bushes, anything that moves (which is everything, by the way), the entire environment is 100 percent computer generated. Sometimes for the full shot including the limousine, we had a real limousine — very rarely — and so everything has to be built. Now, each miniature that you do costs money — miniatures are not cheap — then you have to rig, in this case, the miniature to actually break apart. Now, with a computer it’s easy to multiply things and so on. And if you’re doing an entire neighborhood it becomes totally cost-prohibitive [to use models], in two ways: not just building it but also shooting it.

The second thing is physical limitations. The big [improvement] from Independence Day and even other movies that [director] Roland [Emmerich] did up to The Day After Tomorrow is that in 2012 we’re really in the midst of the action, so there’s not many wide shots — you know, God’s eye view, where we’re looking down on stuff. We really are straight in it, with a Cessna plane, with a limousine, and so on. That means we’re driving by and through all these things, and you know, with miniatures you have to shoot high speed, for instance, and that means also the speed you’re driving with has to be then, whatever, four times normal speed, and there you get into physical limitations of the camera rigs. So some of the stuff, even if we could have done it with miniatures, we couldn’t have done it physically, in terms of the camera rigs.

Shake, Rattle, and Roll

30 NINJAS: Can you walk me through, blow by blow, from when you got the script, what the steps were to bringing that sequence to the screen?

VE: Sure. That limousine earthquake sequence was very interesting, because that was slightly different than normal. That one actually was developed during the previsualization [the lower-quality animation the VFX team does as a first step]. So literally in the script there were about three or four lines that said: “They get into the car, drive through Los Angeles as all the buildings are crumbling around them, and then arrive at the Santa Monica Airport.”

30 NINJAS: So they just left everything up to your imagination.

VE: Kind of. The writers, Harald [Kloser] and Roland [Emmerich], had an idea of what they wanted, and they knew it would be a bigger sequence, but they didn’t write it out. First, we got a beat sheet from them. It’s basically the earthquake starting as they’re running out of the house, all the way until they reach the Santa Monica Airport. So the beat sheet would say: OK, they’re running out of the house, they hit the Porsche of the lover of Amanda Peet, and they [drive], and shit sprays the windshield, and the freeway breaks apart, and so on. So we take that, and we first build the entire landscape there as a previsualization, so it’s a fairly simple build: The houses are just boxes, with some textures mapped on it.

30 NINJAS: When you say you build it, you mean you build it on the computer screen.

VE: In the computer, yes. Because you really model it as if you modeled it for real, only you do it in the computer. So you literally build a set in the computer that’s an entire residential neighborhood, and then a business district, and so on. And we animated it, all the way through: We animated what the limo was doing, from hitting the car, to this and this and this, the freeway breaking, and so on, and then we would treat it really like a live-action shoot. And that’s something that, as far as I know, has never been done that way: where we would then think about a master shot, and we would basically mount a few cameras (again, inside the computer) that are following the limo, and we would render the entire thing through, from the beginning to the end. And we’d give that to editorial, and editorial would start immediately cutting it together, and then would come back to us. We would look at it together with Roland and say, “OK, we need to add additional insert cameras here, this would be a good moment to add [this or that],” and we would do that — add those cameras — and we would render it out, and this way really build the scene in the computer and develop the script, so to say, through previsualization in the computer. And I think it worked really, really well, because it’s a tight sequence that I think feels very fluid.

8,000 Square Feet of Terror

30 NINJAS: Now, that sequence also combined your computer-generated visual effects with “practical” special effects, didn’t it? I understand that Roland had a great big deck built — with palm trees and streets and cars and houses — that actually shook the bejesus out of the actors and everything around them.

VE: Yes, in regards to the earthquake, to start with the special effects rigs on the set — the practical effects — we had what we called the Shaky Floor, where the actors were actually able to be in a real earthquake environment, and that was 8,000 square feet. Of course, it’s frankly not that many shots; it’s probably a couple dozen where that happened — when they’re inside the house and running out of the house just when the earthquake starts — and then we immediately go wider and are in the limousine [moving through a computer-generated environment].

Of course, there’s lots of meetings involved, and from a very early point on we were in those meetings in Vancouver, where the live action was shot, and all the practical effects were done during the shoot, and were discussing with our practical effects supervisor and his team which part he would do, and which part would be ours. And the dynamic is for some reason, to me, in every movie the same: There’s a lot of good intentions to do way more practically than in the end are done. And a lot of times it’s the director who says, “OK, but if we do it like this, what are my limitations?” And of course, if you do things physically, and if they’re really big, there are a lot of limitations, so many times you just get to the point fast where you go, “OK, we’ll do this part here physically, but the rest just has to be a big bluescreen behind it and we’re gonna augment that [digitally].”

30 NINJAS: So did that happen here, where the proportion of practical special effects was reduced from the original list?

VE: Of course, they’re doing a great job, but we found out, for example, that when it comes to flipping cars in that earthquake sequence, where that parking structure starts to collapse and then our protagonists are driving in their limousine past that structure, and they have cars bouncing onto the street left and right — the original idea was to do a lot with practical cars being dropped while the limo drives. And I think [in the final movie] it’s one cut that still has actual special effects cars in it — one real car rolling by and one car smashing behind the limo. That’s real, and everything else in that piece of that sequence we did in CG with completely photo-real cars smashing onto the asphalt with, you know, the windows shattering into a million pieces —

MW: Actually, I think it ties into what we talked about with miniatures. Because, of course, full live action has exactly the same limitations. One thing we discussed very early on with the special effects supervisor is that if that parking garage crumbles, or the freeway falls down, the cars that are coming out of this parking garage have to come out in a very, very specific trajectory. So first, you have to make that happen, but even if you can make that happen, which is very, very difficult — because he said, “We can easily drop cars from a crane” — but then they fall straight down, which is not what we want. Or we can flip cars left and right through the picture, but that’s also not really what we want, because they really need to come at a very certain angle down. So even if you could do that for real, which you could — it would be very expensive — then you’re still very limited by that. What if the director comes afterwards and says, “You know what, it really doesn’t look so good, how the parking garage crumbles; I’d like to do it a slightly different way”?

30 NINJAS: And how many takes are you going to get when you’re destroying cars each time you toss them?

MW: One, that’s it. Yes, exactly. So if you calculate all that, and you think that even a second unit team costs you about 100 grand a day, so if you shoot, let’s say for only three days, $300,000, for that same amount, you can get a visual effects crew of 120 people for one week. So you can imagine that those people could do quite a bit more than you can do in three days by just dropping a couple of cars.

Whatever You Do, Don’t Pull a Shatner

30 NINJAS: Can you talk to me a bit more about how, in this earthquake sequence, it was decided which portions were going to actually be built and how the shaky floor idea was implemented?

VE: We knew that the only things that would make sense with this shaky floor is to have the parts where the actors run out of the house and then later when they’re actually sitting inside the limousine. This was a different rig inside the studio where there was just a car on top of one of these rigs, and so the actors were inside the car in front of bluescreen. I mean, it was almost like this good old Hitchcock rear-projection set-up, only we’re now using a bluescreen for that, and they were shaking like crazy — and everything else was done by us. But Roland’s decision to actually do it with those physical effects when they’re running out of the house was just based on the fact that he didn’t want them to act it; he wanted them to feel it. Otherwise, we would have called it a Shatner, like on the bridge of the Starship Enterprise. So you don’t want to do a Shatner on that one — it’s gotta be 100 percent real. You want to smell the fear.

Read Part 2 of our exclusive interview with 2012’s VFX geniuses to learn how they used thousands of computer processors to flood the Himalaya Mountains (and send that poor monk scurrying for his snorkel) Read It Now!