‘X-Men: Days of Future Past.” All images © Twentieth Century Fox. Images courtesy of The Third Floor, Inc.
As you examine the visual sophistication of a film like X-Men: Days of Future Past, it’s not hard to see why major studios continue to expand their practical and strategic use of previs. No longer just an elaborate animated storyboard, previs has become an essential part of every aspect of film production, from the earliest look development to the final editing postvis.
As one of the industry’s leading previs companies, The Third Floor has worked extensively on many of this year’s biggest films. I recently had a chance to talk to previs supervisor Austin Bonang about his team’s wide-ranging work on the latest X-Men film, which included Sentinel design, onset SimulCam previs asset integration as well as prevising key Moscow and Washington D.C. sequences.
Dan Sarto: Describe the scope of The Third Floor’s work on the film? What was your role?
Austin Bonang: The Third Floor started very early on the film. Right before our Christmas break in 2012, Bryan [Singer, the director] pitched us his vision for the movie. We came on the project as soon as we finished the holiday break, storyboarding and blocking out ideas for a couple months before joining the production in Montreal, where The Third Floor also has offices. The production set up shop at Mel’s Studio, which housed the film’s sound stages. We worked with Richard Stammers [overall VFX supervisor from MPC] and Blondel Aidoo [VFX producer] initially at our offices and then on site in Montreal and back at Fox.
One of the first things we worked on was the Sentinels, both the future and 1973 versions. We built different types of models, exploring a range of designs while staying loyal to the original Sentinel concepts. Richard and his team were doing versions as well. The production designer was involved as was the director, who would review the work and give notes.
The future Sentinels were something new. They’re different than anything in the comics so it was necessary to figure out what they looked like, how they moved and how they behaved. The production designer John Myhre compiled all of the designs for a presentation to the director and Fox. From there, final decisions were made. So designing the Sentinels was a big collaborative effort on many different levels, with Framestore ultimately handling the design of the finished 1973 Sentinel and MPC taking on the future version.
DS: After the initial Sentinel work, what came next?
AB: We moved on to the Washington D.C. sequence. We worked on that for a bit, then set it aside based on some discussions and changes being made to the story. We moved on to the Rogue – Anna Paquin sequence that ultimately is not in the final movie. Brian Smrz, the second unit director, came to The Third Floor to oversee some of the sequence direction, coming into the office every day, giving us revision notes and supervising our work. We shot a couple actors in mo-cap suits at The Third Floor’s home base. We also started working with a virtual camera, introducing it and planning how it would be used for the rest of the shoot. We finished most of this sequence, kind of as a trial run before joining the filmmakers in Montreal.
DS: Tell me about the Moscow sequence.
AB: We started on it early. It wasn’t the first sequence they planned to shoot, but it was probably the most complex sequence in the movie just because there were so many mutants. Also, new Sentinels were being introduced and the tone and pacing of a darker X-Men film than anyone had seen before was being established. There wasn’t much dialogue as the story is told visually, mostly as a flat-out action scene. Bryan said several times this sequence had to be as effective as the “Night Crawler” sequence in X2. It really sets up the whole movie. All types of mutants and powers had to be introduced as well as the Sentinels who had their own unique abilities.
We did many versions of this sequence, probably 30 different passes, animating the previs and cutting it together. Once John Ottman and his editorial team came on, they started cutting it together with sound. Their expertise really helped us. Eventually, some things were moved around story-wise. We were a bit nervous showing the result to Bryan, but we knew we’d nailed it when his face lit up, he paused and said, “Now that is a significant improvement.” We’d finally arrived at a version he could shoot with.
DS: What source materials were you working from? Were there any concept drawings, storyboards, models or digital assets to guide your creative efforts?
AB: Bryan likes his previs to be as polished as possible since he’s using it to determine the directions he wants to take the film. We produced a lot of refined and detailed previs though sometimes the ball would be rolling so fast it was hard to keep up. We did have script pages. We were constantly getting updated script pages. Sometimes though, the description of action would be a bit vague. That’s when we’d go over things with Brian Smrz, who shot the majority of the second unit action stunt sequences.
For Moscow, we had the actual shooting location scanning data, which we used to recreate the set in virtual 3D space within Maya. So we knew exactly what we had to work with. Understanding and working with the limitations of the actual set was crucial to prevising sequences that could actually be shot. That was important because we had to make sure our 12 foot tall Sentinels weren’t bumping their heads on the ceiling. The script pages took us on a basic path from A to B. But we had to fill in all those little fight scenes. Hypothetically, Iceman is going to fight over here. But how are other events going to affect where he’s fighting further into the sequence? We didn’t want arbitrary fights going on in random places just because it was cool. Every fight had to serve a purpose within the story. The fights all had to flow into each other.
We got some assets from the art department, including initial SketchUp models for environments that we would then convert in Maya. We received one-sheets of the character art. But the majority of the assets were created from the plans and concept art.
For the Sentinels, initially we built a rough version based on the concept art. But later, we replaced that with the high-res version built by MPC. The previs needed to feature some detailed shots of the creatures’ mechanisms which our models didn’t have.
DS: So some of that sequence work would be considered techvis?
AB: Yes. We were always very mindful of the actual sets. Once the sets were built, we always went down and made sure the camera crane they were using would fit and all the camera moves within the previs were doable. For example, in the Quicksilver bedroom sequence, the scenes were pretty tight because the set was very small. We actually rehearsed that scene on set and realized that the crane didn’t allow for a particular shot because there was a wooden beam in the way. We had to have the set designer remove the beam. Our head of virtual production, Casey Schatz, built a virtual camera rig for us modeled after the one used onset, so we could make sure not to overextend the camera and crane within the previs.
For the Quicksilver sequence, they also didn’t want to build a full CG character. They wanted to shoot it with the real actor as much as possible for believability as well as cost. So for example, we had to figure out how to get him running horizontally on the wall. We ended up using two different treadmills, one green, one blue, and when the character jumped, they just flipped the camera. It worked really well. Through techvis, it was possible to figure out how to shoot the sequence without having to resort to all digital actors.
We did some reshoots on the Washington D.C. sequence. Originally the sequence was shot on an outdoor set, but the reshoot had to be moved indoors on a smaller soundstage. So we had to measure how much room was available on the new set to ensure the camera moves could be reshot with enough space. We recreated the new set virtually inside Maya just to make sure everything fit within that space.
DS: So the goal there was to see whether or not and in what way the actual set or the shot itself had to be modified?
AB: Exactly. We wanted to make sure all of the previs camera moves could be shot on the actual set. We didn’t want any of the camera moves to feel too “CG.”
DS: You eventually came back to the Washington sequence?
AB: Yes. We did. It became a big part of the monastery sequence because of the juxtaposition of the two, which were both happening at the same time. During the film’s big climax, there’s a lot going on in both locations because it’s a time travel sequence. We were working on them almost at the same time. We edited them together so we could make sure they flowed together. The Washington sequence changed a lot. We came back to it many times.
In the shot where Magneto lifts up this giant stadium and flies with it through the air, for example, we had done something completely different initially using the Washington Monument. That stadium idea came much later.
DS: You mentioned virtual cameras. How often were they used in the previs as well as during the actual shoot?
AB: Once shooting started, Casey was onset not only running the virtual camera but also a SimulCam setup. Before onset shooting began, a lot of the Moscow and future monastery battle was framed up on the virtual camera. Brian [Smrz] came in and shot it himself, because ultimately he was going to shoot a lot of that onset.
We’d animate the action Brian wanted and Casey would put it into MotionBuilder. Then Brian [Smrz] could come in and shoot the action with the virtual camera exactly how he wanted the action to take place. It gave him a lot of control.
DS: So you put together previs sequences and assets at that point that wound up onset in the SimulCam?
AB: Right. And that becomes very important when you have CG characters like the future Sentinels that are 12 feet tall and 1973 Sentinels that are even taller. You need to make sure the actors’ eye-lines are where they need to be while fighting against these creatures. With the SimulCam, the director, the DP and everyone who has a monitor offset can see exactly what’s happening and where that CG elements are supposed to be within the shot composition. It allows the filmmakers to give more accurate direction.
DS: Overall, what were the biggest challenges your team faced on this project?
AB: Does the weather count? [Laughs] Probably the biggest challenge was just the scope of everything. There were so many characters to consider and so many mutant powers. How can you give them all screen time and make them all interesting without milking everything for too long? Also, there were a lot of moving parts and people working on them. But everyone came together with great camaraderie and the goal of making just a really good film.
This film had a lot of effects. It would have been easy to just go wild and crazy with any idea, like having 12 digital characters running around on screen at the same time. But we were always looking to make sure these sequences could be filmed in a realistic way.
Dan Sarto is Publisher and Editor-in-Chief of Animation World Network.