Dream Teams

The VFX company behind Inception reveals the movie's biggest secrets

“You have to be on your A-game.”

Originally Published: 

In most movies, the job of the visual effects department is to go unseen.

You want all those computer-generated explosions to look so realistic, no one even realizes they didn’t happen. But sometimes, a movie’s special effects are so incredible and so groundbreaking the entire industry takes notice.

The Matrix is one example — so visually arresting, it became ripe for parody almost as soon as it was made. Christopher Nolan's Inception is another. Through a canny blend of the real and the almost real, its dreamlike landscapes and dizzying stunts have muscled their way into cinema's hall of fame.

Whether or not you’re convinced by the film as a whole — or can take Joseph Gordon-Levitt seriously as he says “Paradox” as a zinger after pushing a man down a flight of stairs — your jaw will struggle not to drop at Inception's technical effects. Its eerie scenes of exploding debris, Paris folding in on itself, and characters drifting in zero gravity around hotels all linger in the public consciousness.

These VFX sequences are the work of visual effects company DNEG (formerly Double Negative) — a British outfit led by Paul Franklin, Pete Bebb, and Andy Lockley — that got its big break working with Nolan.

Inception was a big deal for me,” Franklin tells Inverse. Of course, it wasn’t easy. “You have to be on your A-game,” Bebb adds of working with the Dark Knight director.

In interviews, Franklin, Bebb, and Lockley unpack some of Inception's most memorable VFX moments and reveal the secret methods that turned Christopher Nolan’s wildest dreams into a cinematic reality.

New kids on the block

In 1998, Paul Franklin became one of the four founders of DNEG. Pete Bebb was the company's first runner and Andy Lockley joined from another facility. The trio has known each other for over 20 years. Coming from the same part of England — the North Midlands and northwest of the country — they’ve always had a similar sense of humor. Bebb was best man at Lockley's wedding a few years ago. He describes the three as a “merry band of brothers.”

When DNEG got their big break working with Christopher Nolan on Batman Begins in 2005, Bebb says that the director was skeptical about VFX. He wanted to use the technology only when the effect couldn't be achieved without it. DNEG, then only a small company, managed to impress Nolan, and he used them on several other films, including The Dark Knight.

By the time he got to Inception, his most ambitious film at the time, Nolan fully trusted DNEG — and Franklin in particular.

A big one

“I spent about 30 minutes trying to work out what was going on.”

DNEG / Warner Brothers

Inception was daunting. Franklin remembers getting the call from Nolan's office while he was in the middle of Mickey's Toontown at Disneyland in Anaheim, California. As this was in the aftermath of The Dark Knight, which had made just over $1 billion worldwide, Franklin expected the film to be a modest one. It was not. He went to Warner Brothers and read a physical copy of the script in a locked room, as is standard for a Nolan project.

“I remember I spent about 30 minutes just reading and rereading the first 10 minutes, trying to work out what was going on,” Franklin says.

The film would require around 500 VFX shots. “It was very, very clear what needed to be done, and you could feel the awesome scale of this moment,” Franklin says.

This was when their history together was important. “The good thing is that we'd been there before,” says Bebb.

The explosions

Every exploding object had to be painstakingly built.

DNEG / Warner Brothers
DNEG / Warner Brothers

Nolan wanted the dreams in the story to be presented with “the absolute conviction of photo-realism,” says Franklin. The first scene that had to confront that challenge was the scene in which Elliot Page's character, Ariadne, is talking to Leonardo DiCaprio's Dominick Cobb outside a Parisian cafe, and he tells her that she is dreaming. There are then a series of explosions that fire material violently into the street. The explosions start rapidly but then immediately slow down, almost suspending the debris in mid-air.

“That sequence, if it's not done correctly, could be very boring,” says Bebb. “Because everyone likes slow-mo...but if it's just slow-mo, then it slowly becomes very, very boring.”

Nolan and his team did shoot some real explosions near the cafe, but the vast majority is computer-generated for the obvious reason that detonating actual glass near human beings isn't a great idea. Around that time, Lockley says, the tendency was for computer-generated explosions to be too choreographed. Nolan wanted his explosions to bear “the disappointment of reality.” There needed to be objects flying in directions you wouldn't expect.

“You could feel the awesome scale of the moment.”

Lockley says that every object that exploded — fruit, cobbles, books — needed to be painstakingly built and textured in 3D. The team needed to simulate them exploding at 1,000 frames per second, even for the parts in which they would appear at 24 frames per second.

“You can't do a simulation at 24 frames per second and then slow it down,” says Bebb. Lockley adds that at first the task seemed quite straightforward, but “it immediately became apparent that it needed to be a lot more complex than just shooting a load of stuff out.”

As the film was generally in either 4K or 6K (4,000 or 6,000 pixels across), the team had a difficult job. With that amount of detail on screen, the CG elements needed to be all the more authentic. Once they had convincingly made the numerous 3D elements — a job that was mainly Bebb's — they needed to multiply these objects, as there were three stages to each explosion. They spent a huge amount of time refining the simulations to not only perfect the timing but also work on the strange interactions between the objects so that, on repeated viewings, an audience could see more and more complexity.

Folding Paris

A mix of old-school images and “tomfoolery” created this iconic visual of Paris folding in half.

DNEG / Warner Brothers
DNEG / Warner Brothers

Another challenge was simulating Ariadne's vision of a folding Paris — an entire chunk of the city that folds on its hinge and closes like the lid of a box. Taking inspiration from a scene in Batman Begins, Franklin explained his vision to Nolan. Lead CG artist Alison Wortman built a wireframe model and the team went to Paris to take around 250,000 photos, scanning the buildings with a LiDAR scanner.

“Bizarrely,” says Bebb, “the weirdest challenge was: What do the top of buildings look like? We had to rely on old-school images and a bit of tomfoolery, shall we say.”

If you fold a city, it will fall into complete shadow because the sun will have no way in. So, in order for the actors to be visible, the team effectively created four suns, the light coming into the scene from various places at various points. “The artists have to have a phenomenally good understanding of light and how it works,” says Franklin. Ultimately, in the film, little windows are left open to “a blue void beyond the Paris landscape.” The shot, finished in November 2009, made it into the first trailer.

Working side by side

For post-production, a period of several months, Bebb, Franklin, and Lockley worked together at DNEG's offices in Fitzrovia, London.

“It's the closest-knit team I've ever worked on,” says Franklin. “Me, Andy, and Peter all sat together in the same room.”

Every day at 4:30 p.m. they and about 15 colleagues would report to Nolan in Los Angeles. Lockley says that with any other team, the intense nature of the work might have been “almost unbearable,” but the three took comfort in the fact that they trusted each other.

Hall of mirrors

A 15-foot, two-and-a-half-ton mirror was built to create the infinite loop of reflections.

DNEG / Warner Brothers

Another memorable scene is one in which, still in Paris, Ariadne creates an infinite loop of reflections by pulling two huge mirrors in line to face each other. To do this, Nolan requested that a real mirror be built.

“It was a huge thing,” says Bebb. “I think it weighed about two and a half tons, and I think it was about 15 feet across. When it moved, it slightly wobbled.” (After the VFX team erased its frame in post-production, Elliot Page could still be seen stepping over it in the film.)

Bebb explains that in order to paint out the crew who were visible in the reflection, they needed to build the entire environment that was seen only as a reflection. This was a challenge. To digitally remove something like a wire is easy for a VFX artist because only the minority percentage of the frame needs to be replicated. For the mirror scene, they needed to computer-generate the majority of the frame using still photography. They had to track the reflection in the mirror as it moves with Ariadne, maintaining some of the real wobble.

Lockley is fond of these idiosyncrasies. “All these little imperfections and little peculiarities are the things that really help sell CG as being real,” he says. “You don't want the surfaces of objects to be computer-clear; you want all these dinks and dents and ripples.” (Colleague Peter Chiang would say something was “a little biscuity” if the texture looked too dry, and would talk about adding a “whoosher” if a shot really needed to sing. This could be a lens flare or a camera shake, for example.)

Zero gravity

The “world’s longest ethernet cable.”

DNEG / Warner Brothers

Near the end of the film is a remarkable scene in which Gordon-Levitt's character, Arthur, wraps several of the protagonists up in what Franklin calls “the world's longest ethernet cable” and floats them down a hotel corridor into an elevator. All of the action was meticulously planned before VFX got their hands on it and, like every scene, there is both a huge amount of “in-camera” (i.e. real) footage and visual trickery.

“The clever thing about that whole sequence is not really the visual effects; it's how it's shot,” says Lockley.

“The compositing was extremely complex,” says Franklin. “This is where I'm 100 percent reliant on Andy.”

No green screens were used. The whole set was mounted on a gimbal, which was rotated 90 degrees. Nolan was on set, pushing Gordon-Levitt around on a kind of seesaw. The bodies — stunt performers standing in for the more famous actors — are hanging on a rig of body-shaped fiberglass molds and supports. When the camera comes close to the human bundle and shows the faces of Cillian Murphy (Robert Fischer) and Ken Watanabe (Mr. Saito), these have been digitally inserted.

The initial shot as the camera tracks through the hotel room took about five weeks, Lockley says, as the team had to remove wires, paint out elements like Nolan himself, and digitally repair some of the set because chunks of it had been removed in order to admit the cables attached to various things in the room. Gordon-Levitt was one of the most complicated elements because he was wearing pinstripes. When painting out his wires, Lockley says, they had to create CG fabric and line it up with his clothing. As Bebb says, you never want to be taken out of the world of the film for even a second.

“Yes. This is good.”

The team sensed the film was going to be something spectacular when they saw a 10-minute trailer. Various DNEG employees went back to watch it again and again. It had been a challenging job — people worked till 9 or 10 p.m. and on weekends — and this made the blood, sweat, and tears worth it.

“I think it really inspired people, and people really went that extra mile to make things work,” says Lockley.

At the cast and crew screening in Leicester Square the day before the premiere, the film was even better than people had hoped. At the premiere itself, the hairs on the back of Lockley's neck stood up. When it came out, Franklin watched it in cinemas and loved listening to the room as the effects lit up the screen and the crowd oohed and aahed in the darkness.

DREAM TEAMS is a series from Inverse that takes a look back at the greatest team efforts of the 21st century and what they mean for our ability to collaborate in the future.

This article was originally published on

Related Tags