Within the superhero genre, Marvel Comics characters from the films Iron Man
Iron Man 2
, and Thor share a fictional cinematic universe, with characters from one film occasionally appearing in, or referencing, another. The latest entry in this Marvel universe is
Directed by Joe Johnston and distributed by Paramount Pictures, Marvel Studios’ Captain America
: The First Avenger is set in the 1940s during World War II. Chris Evans stars as Steve Rogers, a 98-pound weakling rejected by the U.S. Army for regular duty but recruited for a secret mission. Once physically enhanced and transformed into
, he fights Hitler’s henchman, The Red Skull (Hugo Weaving), who leads the evil HYDRA organization. It’s early in the Marvel universe:
parachutes behind enemy lines by jumping out of a plane owned by
Tony Stark’s father.
Although Captain America
’s story is all-American circa 1940, the production was a 21st century global effort. “
, brought to you by the world,” laughs Chris Townsend, visual effects supervisor. “We had 13 studios working in Germany, Australia, the UK, and the US, and we shot it in England. I’m English, but
is the first film I’ve worked on in my home country. There is some irony in all this.”
Actor Chris Evans plays Steve Rogers, who in the first third of the film is so physically weak that the army rejects him for regular service, but later recruits him for a secret mission. He then becomes Captain America, the first Avenger. The wimpy version of Evans is a visual effect, added to the shot at left. At right is the original film plate before the effect.
Double Negative in London was what visual effects producer Mark Soper calls the “cornerstone vendor,” and that studio created most of the hard-body assets—the tanks, planes, and trains—and many environments. Framestore, a second Soho studio, began the work on Red Skull, who is missing his nose, and then when the number of shots overtook the time the studio had available, Lola picked up the scent. Led by VFX supervisor Edson Williams, Lola, which is known for “youthenizing” actors and otherwise buffing up movie stars’ digital makeup, also shaved and warped the images of actor Chris Evans to create the thin, frail character that Steve Rogers was before becoming a superhero.
Townsend and Soper consider the work by the Lola crew on the character they dubbed “Skinny Steve” to be the most cutting-edge effects in the film largely because of the number of scenes in which the altered main character appears. “He’s on screen in the first third of the movie,” Townsend says. “That’s a lot of screen time. Our biggest compliment would be if, when people talk about the visual effects, they don’t mention Skinny Steve.”
The other studios working on the film include The Senate in London, where artists transformed Manchester, England, into 1940s Brooklyn, New York, and created Skinny Steve’s digital bare feet for a car-chase sequence’s 180 shots. The German studio Trixtor created blue bolts of energy in CG. “Complicated work,” Soper says. “They turned it around in a remarkably short time.
One of the most visible visual effects was the alteration to actor Hugo Weaving (Red Skull), who is missing most of his nose thanks to Framestore and Lola Visual Effects.
Fuel in Sydney, Australia, built environments, digital doubles, and motorbikes. Ride FX in Berlin helped with some tricky effects at the end of the film. Method Studios, based in London, Los Angeles, New York, and Vancouver, enhanced an important sequence in which Captain America
jumps from a plane behind enemy lines. Matte World Digital in Novato, California, built and painted digital backdrops and set extensions, and Whiskey Tree, down the road in San Rafael, painted New York in the 1940s as well as handled a fight-sequence montage on a huge tank. Evil Eye Pictures in San Francisco handled composites. Luma Pictures in Santa Monica, California, floated digital boats in a scene, and Look Effects, another Los Angeles-based company, joined the team effort, as well.
“Look Effects came on toward the end to create a digital floating car,” Townsend says. “We had shot a car on a gimbal, but we needed to have it float more organically, so they jumped in, found quick solutions, and did good work.
“It’s amazing how many people work on creating the visuals, and there are so few credits at the end,” Townsend continues. “These people have put the pictures on the screen for a lot of the film. It’s a shame. But, that being said, it takes a whole city of people to make a film.”
Manchester vs. Brooklyn
It might seem odd that the production unit filmed a sequence in Manchester and Liverpool, England, that sent Captain America chasing a car down an alleyway and through the streets of Brooklyn, New York. But, Brooklyn 2011 doesn’t look much like 1940s Brooklyn, either. What’s amazing is that in the film, the chase takes place in a complex pattern through Brooklyn over 180 shots, but the cars on location drove on only two or three streets. The Senate created the illusion.
“They’d drive the cars up and down the street and occasionally turn a corner,” says Richard Higham, visual effects supervisor. They’d film that, then put the cameras in a different position, have the cars drive past the camera again. Then, they’d reset the cameras again, maybe looking back at the cars. They’d keep shooting and resetting and shooting and resetting until they had enough to assemble a chase sequence.” Editors then assembled all the pieces into a sequence, trying to separate repeated buildings by two or three cuts, and using the cuts to make it seem as if the cars traveled farther.
Once The Senate had the assembled, edited sequence, the crew drew a map of the journey that showed the left turns, right turns, repeated intersections, and so forth. “When we had that, we could determine how many views we had to re-create,” Higham says. “All the modern buildings had to go, and of course, so did the dead ends. We added buildings that gave the sequence continuity. We also added extensions when the cars are supposed to be driving down long streets, and added other cars in the background to busy up the plate.”
To insert the buildings in the correct positions in the plates, the crew worked from tracking data. “The point cloud from the 3D track told us where the buildings began and ended, and if the dots were on the same plane, we knew they must be on a wall,” Higham says. “So, we created geometry that lined up from that, extended it into the Z dimension, and projected images onto the simple geometry.” For the projections, the crew photographed buildings in London and found images on the Internet.
When the camera moved too much, the simple projection—that is, the photographs and paintings—didn’t work, so modelers would add more geometry. And, modelers built 3D fire escapes, the old style with ladders, that the compositors added to many of the buildings.
“On the whole, the process is straightforward,” Higham says. “You’ve got your scene, your tracking point cloud, the geometry you derive from the point cloud, the paintings, the perspective, and then you populate the scene with cars and people. But it was time consuming. We had to rip out modern buildings and then bring [Captain America], who is running across the foreground, back in front. And there were so many angles and views; we had to make sure all the perspectives matched. In addition, Chris [Townsend] wanted everything to feel like we filmed it in camera, so we matched lens distortion, lens flares, and motion blur, and kept refining to make it look real. It took artists with good eyes.”
Captain America is barefoot when he chases after the cars, so during filming, Chris Evans wore flesh-colored boots that The Senate artists painted out and replaced with CG feet. “We cyber-scanned [Evans’] feet, modeled and rigged digital feet with an IK bone structure, and added muscles so they’d flex in certain ways,” Higham says. “Then we articulated the joints right down to the toes and applied photographs and textures to the surface. We matched what his boots did in the footage, but we had to take some license because they were stiffer than his feet would be.”
To create their shots, The Senate used Autodesk’s Maya for modeling, animating, rigging, and for projecting textures onto geometry; The Foundry’s Nuke for compositing and tweaking, and adding 3D elements; Science D Vision’s 3D-Equalizer and The Pixel Farm’s PFTrack for tracking; and Mental Images’ Mental Ray for rendering.
Soper and Townsend began work on the film in November 2009. Shooting commenced in June 2010 and continued through December. Postproduction began in December and continued into June 2010.
There was no previs. “Joe [Johnston] made a statement at the beginning that there would be none on Captain America
,” Townsend says. “He’s a real shoot-from-the-hip director. He visualizes the film in his head, storyboards it, and then creates his shot list for the day. We’d walk onto the set and he’d say, ‘Forget the script. Forget the storyboard. I think we’ll do it differently.’ So, I learned to fall into reactionary mode for six months, and I loved it. It might have been easier with previs, but we wouldn’t have gotten the same natural feel.”
On set, Townsend would watch Chris Evans to be sure that his performance wouldn’t create Skinny Steve problems later. “One time, [Evans] was massaging his lip while watching a movie,” Townsend says. “I went to Joe [Johnston] and said, ‘Chris can’t do the hand thing.’ Joe said to Chris, laughing, ‘Hey can you not do the hand thing. It’s too hard for visual effects.’ When you see the film, you’ll see that it’s there, though. But, we had backup shots in case that didn’t work.”
Before giving the Skinny Steve work to Lola, Soper and Townsend considered using digital doubles, a CG head replacement on a small body double, and forced perspective and other in-camera tricks. “We spent many months in R&D before we decided on a technique,” Townsend says. “The general approach we used was to take the performance in the 2D world and stretch, thin, and re-size it. But, we also had a body double who would watch Chris’s performance on set and repeat it. That gave us reference, at minimum, for Edson Williams and the crew at Lola. They could also blend bits and pieces from the body double—shoulders, legs, back, and so forth—into Chris’s image, as well. It was a hodgepodge of techniques.”
Thirteen VFX studios worked on this film, creating invisible effects for the most part, including set extensions, armies of soldiers, and backgrounds, such as in this shot.
The main problem in using the 2D approach was in maintaining a physical volume, especially for Skinny Steve’s face. “When the head faces the camera and then turns right or left, the actual volume doesn’t change, but in the 2D world, if you don’t nail it, the face slides off the head a half a frame too early,” Townsend says.
The biggest challenge, though, was in maintaining Evans’ performance throughout the Skinny Steve sequences. “In one frame he could look like Chris Evans, and the next like Lady Diana,” Townsend says. “Or, if we dropped the eye half a pixel, he’d go from melancholy to sad. It was crazy. Is his neck too thick, chin too long, nose too pointy, eyes too low? Also, Chris Evans is a real chameleon. He looks different in various lighting conditions. We spent many late nights staring at a computer screen, flipping back and forth from Chris Evans to Skinny Steve, and many interactive sessions with Edson and his crew.”
The goal for Skinny Steve, of course, was to create an invisible effect, and that philosophy extended to the rest of the film as well.
“When Chris and I started working on this movie, we talked a lot about how it wasn’t a big visual effects movie,” Soper says. “There aren’t things that jump out at you. Green Lantern is a visual effects movie. In this movie, we’re creating the world of the 1940s, World War II
Europe, and Brooklyn. We created a lot of objects in a CG world—the hard-body planes, train, cars—but they have a photoreal look.”
Jump, Captain America!
To film a sequence during which Captain America jumps out of Iron Man Tony Stark’s father’s airplane behind enemy lines, the production unit put actor Chris Evans inside a set piece built to look like a plane, placed it on a gimbal, and filmed it with a camera on a Technocrane. Method Studios tracked the camera, composited the backgrounds outside the window, changed the plates from day to night, added foreground mist whipping past the open doorway, and placed reflections on the body of the plane for exterior shots and inside the window frames in the interiors.
“The plane is polished aluminum with rivets,” says Sean Faden, visual effects supervisor. “If we had done a strict day for night, it would have looked fake. So, we adjusted the values of the [real] plane shot during the daytime to what we thought it would look like at night, built a CG plane, matchmoved the shots, and then rendered a moonlit version of the plane as a CG element.” To have good reference for the interiors, Faden drove into the mountains near Santa Monica, California, on a full-moon night and took photos from inside his car looking out.
During the second half of the sequence, the plane comes under attack, so the crew added CG tracers and flak explosions, and created the resulting interactive light show on the clouds, the window frames, and the body of the plane. “We have the reflections of the explosions on the plane,” Faden says, “and, the interactive light passes, the orange glows in the clouds. It helps everything sit together.”
A digital double jumps out of the plane wearing a CG parachute created with hand-animated blendshapes and animated with Maya’s nCloth. “We have one shot with the digital double, background clouds, window comps, tracers…the whole enchilada,” Faden says. “Masa Narita did a lot of the work on it; it was a good shot.”
In addition to Autodesk’s Maya, Method artists used Autodesk’s Flame and The Foundry’s Nuke for compositing, completing their shots in two and a half months.
Adds Townsend, “We had a whole slew of invisible effects—full CG environments, interiors, mountains, underwater shots. We find better and more efficient ways to do that and to decide when we need to create full-CG environments and when to use digital matte paintings. But, a film like this isn’t there to push the boundaries.”
Thus, we have one last irony for the globally-created Captain America
. Joe Johnston was a visual effects art director at Industrial Light & Magic for Star Wars: Episodes V and VI, and won an Oscar for best visual effects in Raiders of the Lost Ark. Since then, he has directed several science-fiction movies, including Jumanji, Jurassic Park III, and, more recently, The Wolfman. And now,
, a superhero film that you might expect would have the most visible visual effects of all, but doesn’t.
Many of the backgrounds in the film are nearly all-CG, including the factory (above left) and the snowy mountainscape and tracks (above right).
“Joe [Johnston] and I both love to create beautiful images and put them on the screen,” Townsend says. “But we want those images to further the story. That’s why I go to movies, to be enchanted by the story.”
Barbara Robertson is an award-winning writer and a contributing editor for Computer Graphics World. She can be reached at BarbaraRR@comcast.net.