Once Zack Snyder decided to ground Superman in a more relatable reality for Man of Steel by shooting most of it hand-held at 24 fps, it made perfect sense to go for a super fast stylization for the flying and fighting sequences in Smallville and Metropolis. This was the opposite of his signature slow-mo fight language for 300, Watchmen and Sucker Punch.
The urgency of the Superman flying takes hold during the Smallville Kryptonian battle. In addition to superhero speed they would throw punches that would minimally break the sound barrier so there would be shock waves. At the same time, they didn't want to lose the force of the punches completely in motion blur (except as a gag with Zod's partner, Faora). They added effects to it by putting mock cones on the forearms, shockwave effects when they hit blows as visual cues.
"I did a lot of R&D with Guillaume Rocheron and MPC for the all Kryptonian Smallville battle," explains Desjardin. "That R&D carried over into the more amped up Superman fight with Zod in Metropolis, which was handled by Double Negative."
But the Smallville battle was different from Metropolis because it's a practical environment that's been stitched together amid character animation. Metropolis is a more conventional virtual battlefield.
"Smallville was a real location [in Plano, Illinois] that they wanted to keep the integrity of," DesJardin continues. "That's a problem in a fight. You need to find a way to render that location for camera set ups. We still want the animation to drive the camera all the time because that's where you're looking. There are major parts, and often quite short parts, of the Kryptonian fight, where you need to animate that punch or foot stomp. None of that you're going to do until later, so we not only had to separate out the people but capture that location at that spot so we can post animate the camera in the environment later. We did on set performance capture with the suits and triangulating the witness cameras all the way around.
Desjardin continued to use the ShantyCam, which is a six still-camera set up on a rack, to photograph the character you're going to change to CG to capture all the lighting on set and you can redo their performance frame by frame so you have different expressions and movements at super high-res. And then you can post project that onto geometry and get them from real to CG mode.
"We added one more camera to our arsenal, which was the EnviroCam. It's a still camera body with a 50mm lens on a motorized head on a tripod. That camera will take a sphere of high-res images that are stitchable. We had all these positions that we knew our cameraman, John Clothier, was going to use to follow the fight. After we yelled, 'Cut,' everyone would hide and we'd just shoot the environment in that lighting. You could project that onto simple geometry and with that one position you could interpolate another position. You could even dolly or zoom like crazy because the resolution's so high. That allowed us the freedom to do the animation later between two positions and two character moves and then also animate the camera in post so it would fit that new character animation."
Kryptonians are not constrained by earth physics and have abilities way beyond human limitations. It was established that each time a character would move at speed or fly, their digital version, with CG set replica, would be used so the camera could follow the characters' action instead of trying to fit their animation within a pre-established camera move on an empty plate or limit their action to the constraints of wire rigs.
Superman, Zod and Faora required the creation of digital doubles while Namek, an 8-foot-tall Kryptonian warrior, was created as full digital character using on-set performance capture. Zod’s armor was also fully CG, placed over the actor's body.
The sequence also required complex simulations. As Superman battles with his enemies, buildings are destroyed, vehicles are wrecked and roads are ripped apart. The mayhem necessitated the use of MPC's proprietary destruction toolkit, Kali, enabling the FX team to handle the demolition of different elements, including brick, glass, wood and concrete. The team also used Flowline to generate volumetric dust, smoke and fire around the destruction.
For Metropolis, Dneg (under the supervision of Ged Wright) rendered 32 square miles, which they situated along the eastern seaboard. The city is nearly all CG and was developed in previs form by Pixel Liberation Front. It was a mash-up of Chicago, New York and LA. "Dneg, which had already done a virtual Gotham for Batman, developed a procedural city basket for Metropolis. Their team did lidar from the ground up along with texture photos. We also got a helicopter with a lidar mounted on a steadicam to go down some of the city canyons. We wanted to get roof detail but ended up getting detail all the way down to the streets. They had a complete survey of Chicago, which was then used to tear apart. Then they could throw it back together again procedurally. Then there was another procedural process to layer in the structures inside a building they knew was going to break. Office building interiors became easier to do in CG because the camera moves were impossible. We broke our rules at the end just to keep up with the fight, shooting on greenscreen sets and the rest of it was all replaced later.
Meanwhile, Weta Digital (under the supervision of Dan Lemmon) handled the organic-looking Krypton interiors designed by production designer Alex McDowell as an outgrowth of biomech physics in addition to creatures, robots and space craft.
But one of the coolest sequences was the Kryptonian history lesson that Jor-El (Russell Crowe) provides his son in the Fortress of Solitude craft buried deep in the Arctic. McDowell showed the production team an art installation comprised of tiny cubes that were suspended in a grid and you could look at a representation of yourself in that array of cubes shot by a video camera at low-res.
They decided to utilize the effect in a series of displays in a more tactile way. "We looked at a push-pin effect by Michael Fink in the first X-Men movie," DesJardin recalls. "It took a long time and a lot of R&D. But for the history lesson, the trick was to take deep images and make them look like forced perspective elements in this space using particles of varying resolution sizes. Sometimes we wanted chunky details; other times we wanted finer detail when they were smaller.
"We gave Weta a lot of license to play with the tactile display. But Zack had a very specific idea for the history lesson that was going to be more stylized than the other displays. We looked back at the beginning of Hellboy and Francis Ford Coppola's Dracula. We had to bridge the exposition with imagery that wasn't redundant."
First they animated the real 3D objects and worked out the blocking and pacing and then that would get turned into the particle version by compressing the Z depth and filling in the shapes. This became the volume of the display. Lighting was an additional challenging because they wanted it to look like a museum diorama so they hid all the lights around the edges.
The result of Man of Steel is modern sci-fi with an emotional hook: the first alien contact story. "There's only one high-speed shot in the whole movie, which was shot on film in Panavision, the dreamy flashback of Clark wearing the cape as a kid on the Kent farm," Desjardin concludes.
He can't wait to continue the Superman saga in the sequel that's already in the works. But that's another story.
Bill Desowitz is former senior editor of AWN and VFXWorld and the owner of Immersed in Movies (www.billdesowitz.com). He's also a columnist for Thompson on Hollywood at Indiewire and contributing editor of Animation Scoop at Indiewire. Desowitz is additionally the author of James Bond Unmasked (www.jamesbondunmasked.com), which chronicles the 50-year evolution of 007 on screen, featuring interviews with all six actors.