Method Studios: VFX for Abraham Lincoln: Vampire Hunter
COW Library : TV & Movie Appreciation : Randy Goux : Method Studios: VFX for Abraham Lincoln: Vampire Hunter
Abraham Lincoln: Vampire Hunter "Train Escape" clip
Randy Goux shared his experiences on Abraham Lincoln: Vampire Hunter with Creative COW.
Method Studios was approached by Fox's visual effects team to work on Abraham Lincoln: Vampire Hunter. They had a pretty ambitious part of the movie they were looking to solve. It was very tasty stuff - a train sequence with burning bridges, full CG fire, all in stereo. As CG artists, this is the kind of meaty work you really want to go after.
Three major houses worked on the VFX for Abraham Lincoln. WETA had a pretty substantial sequence, dealing with a stampede and Spin in Toronto dealt with the Civil War sequences. We worked on this 8-minute train sequence, during which Lincoln and his partner Will are battling vampires on top of a moving train boxcar. In the second half of that sequence, the train goes over a trestle bridge that has been sabotaged and set afire. We also created five or six large CG environment shots, including a digital crowd of 5,000 people listening to Lincoln's Gettysburg Address.
Before we were awarded the job, we flew down to New Orleans and had a dinner with [director] Timur [Bekmambetov]. The dinner was amazing. It was a half social/half technical dinner. He talked to us about how he makes movies and how he likes to constantly explore things creatively. It can be a bit nerve-wracking when the director wants to leave all options on the table, but that's what made it exciting to us. He wanted to know a lot about what we were doing, to make sure that he wasn't going to ask us to do something that wasn't realistic and whether he could push us somewhere that we could actually achieve.
That's exactly what happened. We were pushed very far but he wasn't asking for anything impossible. We did lots of look-dev all last summer, honing the effects before editorial got the sequence together. By then, he had an idea of what we were capable of and pushed us even more.
We started serious look-development just as they finished shooting, so Timur could really focus on what we were doing. As visual effects supervisor for Method, I faced going down a creative path while making sure that we don't promise a technique that'll break the bank. There were numerous factors that made the battle sequence on top of the boxcar challenging. The sequence takes place at night, so there's a nighttime grade on top of it. Smoke is another major component. Smoke is constantly barreling through the scene, and the humans and vampires are constantly going in and out of it. We knew we couldn't shoot with smoke; when you do, you usually end up replacing it with CG smoke. The train is also going through a forest, and moonlight casts shadows of all the branches and foliage of the forest onto the smoke, vampires and people. Plus there are the burning embers from the locomotive. All these factors combine to create a really tense, really dramatic feel with a handheld camera and hand-to-hand combat.
There were numerous factors that made the battle sequence on top of the boxcar challenging. The sequence takes place at night, so there's a nighttime grade on top of it; smoke is constantly barreling through the scene, and the humans and vampires are constantly going in and out of it; the train is also going through a forest, and moonlight casts shadows of all the branches and foliage of the forest onto the smoke, vampires and people; and there are burning embers from the locomotive.
The look-dev for all those pieces of the puzzle was ongoing for a pretty intense three or four months. We had to figure out exactly how the vampires came in and out of the smoke, how intense the dappled shadows were on the boxcars. And we had to fully integrate the characters, digitally fluctuating and modulating the shadows on their faces. We had to develop a flexible system for creating these volumetric smoke effects.
Look-dev was important, because Timur really wanted to exhaust all our options, go through all the different feels and looks before he had a decent edit. That way, when he went into editorial, he'd know what was perfect for the moment.
We shot the sequence on top of the train on a stage in New Orleans with two boxcar tops and a green screen with fairly neutral lighting and some fans blowing around. The actors on top of the car are doing their choreographed fight sequences. Those were the plates.
Method Studios' finished scene and plates of the boxcar fight sequence.
Another challenge was that Timur's directorial/visual style integrates a lot of speed ramps. Because speed ramps were in constant flux editorially, we had to create all of our effects at 250fps, at least, in order to be able to keep up with editorial changes.
We'd get a new editorial download almost on a daily basis with different time ramping. We had to make sure all our effects could be slowed down to almost zero time.
With regard to the fact that the movie is a stereoscopic 3D release, a decision was made to do almost half of our shots in full stereo and have Stereo D convert the movie, which was shot in 2D. Although the conversion process is a whole other pipeline that you'd think is not in our court, it did impact us. Characters embedded in smoke and other VFX we created are very volumetric. So we'd hand over to Stereo D all our comp scripts and elements, everything they could possibly use to get a more successful conversion, rather than delivering a finished shot and having them start from scratch. We had a pretty specific relationship with Stereo D. Their stereographer Graham Clark spent a lot of time with Timur and us, talking about what was possible, what we could do, and how to think about this differently.
We didn't sign off a lot of shots until late in the game, not because they weren't good enough but Timur would have an idea. Once we had solved the technical issues, we turned to creative iteration. In the last three or four months, we were constantly trying something different. Timur would say, Maybe we're hiding this part too much in smoke, let's back off. The push at the end of the show was about making it look dirtier and more real. We gave our cameras a slightly different shake especially when the train was on the bridge.
For the volumetric smoke, they developed a compositing technique that made the smoke more efficient. Since they were doing speed ramps, a true volumetric smoke solution is very expensive, especially for the number of shots they were doing. Led by their Compositing Supervisor Abel Milanes, they developed a system in Nuke where they could actually take a rendered sequence of 8K smoke and map it onto a sphere in Nuke.
To do this, we had to keep a tremendous amount of shots online and when you're dealing with CG fire and smoke, that's a hell of a lot of space. We were up to 80+ terabytes, and we're not talking about what you can get at Best Buy. To create visual effects, we need high-speed, disc array storage, and that's rather expensive.
We did develop some of our own techniques specific to the kind of effects we were building. For example, for the volumetric smoke, we developed a compositing technique that made the smoke more efficient. Since we were doing speed ramps, a true volumetric smoke solution is very expensive, especially for the number of shots we were doing. Led by our Compositing Supervisor Abel Milanes, we developed a system in Nuke where we could actually take a rendered sequence of 8K smoke and map it onto a sphere in Nuke. Once it's in 2D you're not doing all the computation of the smoke for every shot. There's an economy there. Developing that technique was major in getting the sequence done. Of course we did have to do volumetric smoke to accent all over the place, but the majority of the shots used this technique we developed in Nuke. To give an idea of how much time it saved, if we got a call from editorial that Shot XYZ had a new speed ramp, we could turn it around in two hours when it was a 2D solution. If it were a full 3D solution, it could potentially have taken two or three days. When you're dealing with 100 shots, that's a major time-savings, and we pulled it off.
Next, the sequence moves to a 12-storey trestle bridge that has been set ablaze. We couldn't do those 2D tricks with fire. If you map a 2D fire from an element shoot onto a card and try to do that in stereo, it looks like flat. We had to do the fire in stereo. So we rolled up our sleeves and got the fire system going. This was actually a major task. We started over a year ago, taking our existing solutions for fire in Houdini and getting prepared to do a real large-scale fire. We went through every possible scenario of the scale of fire we need to do. We're not talking about torches but 100-foot flames attached to 20-foot beams falling and crashing on one another.
Our technical R&D team, with our Digital Effects Supervisor Sean Lewkiw in the lead, looked at how we were going to create the fire. We had to be prepared for really intense, photoreal, close-up, no-cheating-in-comp fire. This amazing group of effects TDs took everything to the level we needed. The thing you worry about with fire is detail and performance. Fire can look very different depending on the exposure of the sky behind it and how hot it is, among other factors. Part of getting the quality we needed was the Houdini renders with fire. But we needed control of all the detail in the passes we used for the compositor. Our shader for fire was built around real world factors: the more fuel in it, the hotter the fire, the more violent it would feel, and the larger it would roil at the bottom. We could export fuel and heat passes.
If you look at those passes individually, they looked like purple paisley, but the compositor could use them to taste. Compositors would only dial in these heat or fuel passes depending on what we were looking for. For example, if the shot showed the actors, we needed to see details in their faces and costumes, and thus the details in the fire would be blown out. If we're focusing on the fire, we see more detail in the fire, and the actors were more silhouetted. As we always do, we had to think like filmmakers. Like a cinematographer, we were exposing for the situation in this moment and then adjusting for that.
We also did a handful of environmental shots, which we handle a lot at Method. We had to re-create the crowd for Lincoln's Gettysburg Address. The production filmed Abraham Lincoln and the people on the stage, and right out in front of the stage are a couple rows of extras in costume. Behind them is an empty field. The camera is on a crane, completing a graceful move, swooping around and revealing the audience of 5,000 people listening to one of the most famous speeches in history. We had to create that audience.
Method created the audience of 5,000 people - quite a task when the subjects are still and silent.
"You can get away with some cheats when the crowd is very dynamic."
It was a pretty ambitious, 1200-frame shot, covering most of the speech. We had to take our crowd system, which we've been developing for many years, and push it to its maximum level for detail. We needed to be sure that our crowd held up at that size and the speed of the camera. You can get away with some cheats when the crowd is very dynamic, but in this scene, everyone is standing silently and the camera comes right over Lincoln's shoulder and stares at the crowd. We had a team of ten people who up-resed all our texture maps and geometry and really opened up the hood of our crowd system to turbo-charge it. Our render times were pretty extensive on this as well.
I've been on a few shows that have gone this far; I call them marathon shows. I feel that once I cross a threshold of seven months of intense work and still have four or five months to go, I know from experience that I'm doing something that is worth it. But I was dealing with a lot of extremely talented artists who had never been on a marathon show. To them, there's no end in sight. I saw this wasn't your average show, but one of the biggest effects show we'd ever done in terms of the intensity of the fire. My producer Jinnie Pak and I rallied the troops by making sure the crew knew what was going on with production at all times and helping them predict what could happen down the road.
When I look in my office, I'm surrounded by concept art, and that reminds me of how we partnered up with Timur, this talented director, in a really engaged, creative way and accomplished the show's climactic sequence. We all felt really connected to the production side, to the director, more than other shows we've worked on. That could be part of being on a show for so long. But we also made a pretty major accomplishment in our effects capability, and we rose to the occasion. The director was very pleased. The studio was pleased, and that always makes you happy.
Title art: Abraham Lincoln (Benjamin Walker) and his closest friend Will Johnson (Anthony Mackie) make their final stand atop a speeding train. Photo courtesy Twentieth Century Fox.
All images ™ and ©2011 Twentieth Century Fox Film Corporation. All rights reserved.