Search form

'Children of Men': Invisible VFX for a Future in Decay

Alain Bielik talks with the visual effects wizards given the task of transporting audiences into a believable, bleak near future in Children of Men.

Invisible effects help sell the believability of the bleak future in Children of Men. All images, unless otherwise noted, © Universal Pictures. Courtesy of Double Negative.

Invisible effects help sell the believability of the bleak future in Children of Men. All images, unless otherwise noted, © Universal Pictures. Courtesy of Double Negative.

Set in the near future, Children of Men (which opened Christmas Day from Universal Pictures) is a gritty thriller in which humanity faces the perspective of its own extinction. Due to a mysterious global phenomenon, women have lost the ability to give birth. In this childless world, even the most civilized societies are collapsing. The only hope lays in the hands of a small group of British rebels that tries to bring the only pregnant woman in the world to safety. Adapted from P.D. James' best-selling novel, the movie was directed by Alfonso Cuaron with a unique naturalistic documentary shooting style. The entire movie was filmed hand-held with wide lenses (mostly 18mm or 21mm) in order to give the footage the feeling of an actual newsreel. The director also favored extremely long takes - there are no more than 460 shots in the final cut.

For lead visual effects vendor Double Negative, London, this approach implied the creation of more than 160 "invisible" effects shots of the highest complexity. The core team included vfx supervisor Frazer Churchill, vfx producer Rupert Porter, CG supervisor David Vickery and 2D supervisor Andy Lockley.

Like many future set films, the world of Children of Men is dark and bleak.

Like many future set films, the world of Children of Men is dark and bleak.

"There were two main categories of effects," Churchill explains. "First, we had to enhance environments to make them look like the action was taking place in 2027. Then, we had to combine several takes to create impossibly long shots. One of them is a nine-minute hand-held tracking shot made up of six different takes. Obviously, the shooting style created a true challenge for us. The handheld camera meant that there were no tripods, no dollies and none of the usual points of reference. It also meant that the height, tilt and roll of the camera were always varying, making them impossible to measure and to repeat precisely. Our matchmove and tracking team really out-did themselves on this project!"

Visualizing a World in Decay

Although the movie is set in the year 2027, Cuaron didn't want the action to take place in a futuristic world. This is the world that we know, only with a slightly more sophisticated technology. The London of tomorrow was mainly realized by the integration of multiple animated billboards on buildings and vehicles. These elements were blended as much as possible into the background, as not to draw unnecessary attention. Some billboard images were purposely created in poor quality and projected onto screens in a bad state of repair -- this is a civilization in decay.

The lengthy opening shot features many of these subtle enhancements, but the viewer focuses immediately on lead character Theo (Clive Owen). In one continuous shot, the camera follows Theo out of a coffee shop when an explosion blows the place apart right behind him. The impressive shot was created by Double Negative from two different takes shot over two consecutive days.

"For all these shots that we had to blend together for the movie, we started with approved storyboards," Churchill continues. "We then went on set a couple of weeks before the actual shoot, and filmed several tests in video, rehearsing the whole action. Then, back at Double Negative, we tried to determine the ideal transition point between every two consecutive takes. We did several versions and showed them to the director. Most of the time, he would tell us: 'This is too easy! Try something else...' So, we kept on refining the transitions in video until we met Alfonso's approval. Then, we would go on set and rehearse the take with the actors and the cameraman operator, based on the approved video test. Most of these takes were logistical nightmares as there were so many cues to hit: actors, camera, extras, pyrotechnics, stunts, etc. If something went wrong, it was so complicated to reset it all...

"During rehearsals, our vfx editor Andy Hague would take the live feed from the video assist into editing software Final Cut Pro, and create test transitions directly on set. Using this reference, the camera move and choreography were then adjusted to provide the best transition points possible. Since the takes were all hand-held, we filmed the set with video cameras to see precisely where both the actor and the cameraman stopped at the end of each take. We then referred to these images to place them in the same position at the beginning of the next take. As soon as we had two plates, we would test the transition again and determine if we had what we needed to create a seamless blend before moving on to the next shot."

The transitions were mainly completed in Maya using a 2-1/2D reprojection technique. First, the end of A-roll and the beginning of B-roll were tracked in 3D. A third camera was then created and used to blend the two camera moves. The 3D data generated by the three cameras was then exported back into Shake via proprietary software. This allowed the compositor to take the 3D data from the three cameras, and complete the physical merging of the plates within the compositing software, enabling him to retain maximum image quality from the original plates. The team often added foreground elements over the transitions to help with continuity issues and positional inconsistencies.

Many of the subtle additions to scenes blend the current urban landscape with simple futuristic flashes.

Many of the subtle additions to scenes blend the current urban landscape with simple futuristic flashes.

Merging and Blending Multiple Takes

For the coffee shop explosion scene, there were two "hero" takes to blend together. "On the first day, we shot the shop interior," Churchill notes. "We would later add screen inserts for the televisions, and digital billboards outside. At the end of the shot, the camera would exit the shop door out onto the street and stop there. On the second day, the interior set was emptied and rigged with explosives and debris. The camera then duplicated the exit from the coffee shop and continued the move down the street. Our task was then to create the illusion of a continuous camera move. For this, both plates were tracked, stabilized, and stitched together merging different sections from both plates into the doorway. So, as we exit the door with Theo, we are seeing the pavement and people from the day two shoot, composited over the passing bus from day one. As the bus passes, the rear of it 'wipes' on the rest of the day two plate, revealing the buildings and traffic."

Shot separately on greenscreen with a wire rig, a stunt couple was composited into the café doorway to be "blown" across the street. Extra debris was also added in the shot to enhance the original explosion, and cracks applied to foreground windows. Finally, a new handheld camera move was generated, so the transition didn't have a smooth stabilized feel to it.

Most of the transitions in the movie were completed on the environment while the camera was briefly panning away from the main actors. For the coffee shop scene though, Theo seems to remain in frame throughout the shot, but at the transition point, the camera cleverly pans off of him to show his reflection on a window. "You get the feeling that he was in camera throughout the shot, when actually, for a couple of seconds, you only saw his reflection... You just don't notice it."

Even more complex was a shot in which the camera follows Theo in the middle of an urban battlefield and inside a building. At one point, in a remarkable cinematic sleight of hand, the camera actually becomes the character's point of view. The shot was captured in five separate takes over two locations. The exterior section of the shot was split into only two takes and the interior split into three. On top of the already demanding transition effects, Double Negative also had to deal with a huge amount of set and action enhancements. Some buildings were painted out, others were extended with extra stories, new structures were added in; hundreds of bullet hits were also composited in, using practical elements.

During the interior section of the shot, the environment remains visible behind windows. Since the plate had been shot with greenscreens, the team needed to replace them with a digital panorama of the exterior. First, digital stills of the exterior set were stitched together using proprietary software. The resulting panorama was mapped onto a 3D cylinder, which was then imported into a 3D scene containing the matchmove data from the tracked greenscreen plate. From the 3D package, the cylinder and the matchmoved camera were then exported into Shake via another in-house application, and the 'Cyclorama' was composited into the greenscreen windows.

Multiple takes were combined and tweaked to accomplish the final long shot of the anarchists' attack on the car.

Multiple takes were combined and tweaked to accomplish the final long shot of the anarchists' attack on the car.

Crafting a Nine-Minute Long Shot

The most complex shot of all probably was a nine-minute long scene in which the characters have an extensive dialog while driving a car, and are then ambushed by a group of anarchists, resulting in one of the character being shot dead on its seat. The shot was filmed in six sections and at four different locations over one week and required five seamless digital transitions. Moreover, the camera records the action with a continuous movement that would actually be impossible to create in reality. In many instances, the camera ends up shooting the actors from a seat where we had just seen another actor the second before...

The plates were shot from a "doggy cam" shooting through the cut-off roof. The director, the cinematographer and the camera operator were actually seated on top of the car, thanks to a special rig, while the vfx crew and other technicians were hiding out of camera range around the traveling car. Altogether, 13 actors and crew members were on board for plate photography!

Given the length of the scene, the team opted to use as much of the original plates as possible, re-timing, warping and painting to reposition actors and parts of the vehicle where they didn't quite line up from section to section. Photographic textures of the entire interior of the car were taken to create a 3D model that could be used to align the 3D tracking data for each section of the shot. The roof was replaced throughout the entire shot, while the dashboard, windscreen and parts of the front doors had to be created in CG in several instances to allow for a smoother transition between plates. Defocusing the Maya elements was achieved using depth passes from software packages, including a proprietary plug-in that uses real world camera and lens measurements to calculate correct focus levels. Focus distances were then animated by hand to match in-shot focus pulls. "During filming, each location was photographed using an 8mm lens over a range of 12 stops to produce an HDRI environment -- inside and outside the car -- that would allow us to light the CG elements," Churchill says. "Using proprietary tool Stig, we created tiled panoramas of each environment that we then used to join the surroundings from one location to the next."

The live-action ambush was greatly enhanced by CG Molotov cocktail, a shattering digital windshield, a bullet hit and blood spurt and even a CG biker and motorcycle to augment a stunt performed during plate photography.

Framestore was given the task of bringing new life into the world of the film. Courtesy of Framestore CFC.

Framestore was given the task of bringing new life into the world of the film. Courtesy of Framestore CFC.

Tackling the Holy Grail of Visual Effects

While Double Negative was crafting invisible effects shots, Framestore CFC, London, was busy trying to create the most realistic CG human being ever put on screen. During the climactic scene, the pregnant woman finally delivers her baby into Theo's hands. After trying to use an animatronic newborn, Cuaron decided to call for the expertise of the company that had created the highly successful digital Hippogriff for his Harry Potter and the Prisoner of Azkaban. Creating a realistic human being has always been the holy grail of CG animation, but here, the challenge was made even more difficult by the fact that the shot was a three-and-a-half-minute long take, filmed with a hand-held camera, and lit by a hand-carried hurricane lamp.

"It was definitely one of the most demanding projects I ever worked on," vfx supervisor Tim Webber admits. "This baby was the key point of the whole plot and the emotional climax of the entire movie. We wanted to make the audience believe that the producers had actually convinced a pregnant actress to deliver her baby on screen. So, we couldn't afford to miss..." Working alongside Webber were CG supervisor Andy Kind and animation supervisor Michael Eames.

The plate was shot with Owen holding a limbless dummy to help him focus his eye line and body language. The plan was to simply cover the dummy with the CG baby. However, during post-production, Cuaron decided to make the newborn a premature. As a consequence, the dummy was now much too large and couldn't be covered by a scaled down digital baby. After tracking the plate in Matchmover and boujou, the team used Commotion to painstakingly reconstruct the missing parts of the background, as well as Owen's hands and costume, without any clean plate to work from...

As usual on any digital human being, the key to the success -- or failure -- of the CG baby would be the skin, how it moved as well as how it looked. "The baby was modeled in Maya and animated from reference footage that we found" Webber notes. "We even asked our colleagues if anyone had filmed a childbirth and would be willing to show us the footage! We wanted to have as much reference on a newborn look and body language as we could. For example, we noticed that newborns have a very floppy and wrinkly skin. It meant that we had to do a lot of additional work that we wouldn't normally do on a wrinkle system. The skin had to move in a very specific way.

"Also, we knew the sub-surface scattering would be critical. Luckily, RenderMan had just released a series of new rendering tools, including sub-surface scattering techniques that were precisely what we needed. We obviously employed global illumination techniques, but we actually lit the baby to look good. The CG lighting was not the exact reproduction of the real lighting set-up. In all honesty, I don't think we did anything groundbreaking in terms of the technology. In my opinion, it is the first totally believable CG human being, but I think this success has more to do with the enormous amount of work and care that went into the texture, the lighting, the rendering and the compositing [Shake], than to some fancy new algorithm. I'd like to say so, though!"

Alain Bielik is the founder and editor of renowned effects magazine S.F.X, published in France since 1991. He also contributes to various French publications and occasionally to Cinefex. Last year, he organized a major special effects exhibition at the Musée International de la Miniature in Lyon, France.