The highly stylized, computer-generated look of Speed Racer was built in a way that frequently blurred the lines between different production and postproduction departments. Photo courtesy of Warner Bros. Pictures.
As Speed Racer hit theaters, there was much debate about the reasons for its disappointing first-weekend box office take (only in Hollywood could $18.5 million in three days be considered disappointing). On the question of the look of the film — the so-called “photo-anime” design and color scheme — however, everyone seems to agree: This is one tripped-out movie. Kim Libreri, Digital Domain’s co-visual-effects supervisor for almost 500 of the more than 2,000 visual-effects shots in the movie(BUF Paris, Industrial Light & Magic, and Sony Pictures Imageworks were the other major vendors on the project), in fact, refers to Speed Racer as “the most blinged-out movie eye candy yet.”
How that blinged-out look was achieved technically is important certainly, but a better question might well be: What is the significance of this form of moviemaking, and how will its techniques, tools, and philosophies impact the industry moving forward?
Members of the core team that collaborated on behalf of Speed Racer‘s elusive directors, Andy and Larry Wachowski, have definite thoughts on this subject. They firmly consider themselves to be collaborators on something truly unique — so much so that they suggest the movie’s significance may well lie in the blurring of lines between departments. Co-editor Zach Staenberg says the film will play a role in the “rise of post-production” for the making of highly stylized digital movies.
The team says Speed Racer has added a unique, new chapter to a colorful and growing genre that includes 300, Sin City, A Scanner Darkly, the Wachowskis’ Matrix films, and Sky Captain and the World of Tomorrow. The similarities between all those pictures, they add, lies not so much in terms of mimicking each other’s techniques, because the approaches varied widely. Rather, the filmmakers’ general willingness to experiment with, and rely heavily on, a wide range of postproduction techniques as a primary filmmaking tool for creating highly stylized imagery lumps such projects together. In that context, Speed Racer would appear to be a state-of-the-art poster child.
“This movie feels very much like live action, but it is clearly a hybrid movie,” says co-visual-effects supervisor Dan Glass. “The way we put the layers together reeks of animation techniques, but it is definitely photographic-looking. I think we’re getting into an exciting phase of movie-making. Digital is contributing to this by bringing things together that were separate processes before. From the outset, the art department and visual-effects lines were blurred. Visual effects and shooting were interrelated, even hard to distinguish sometimes. And the editing was very complicated in terms of the goals of the shots and the multiple number of elements that were not only edited together, but often composed by the editors. One shot sort of bleeds into another — elements and actors and animals that lead you from one scene into another. Multiple layers of people in different poses and moments that are interluding in time visually in front of you — all this makes this movie very rich and exciting.”
The design for Speed Racer‘s unique visual palette emanates from the same place its editors found inspiration: classic, cel-animated, anime-style cartoons. Photo courtesy of Warner Bros. Pictures.
John Gaeta, Glass’ partner as co-visual-effects supervisor and a key player in the anime-inspired design and palette of the movie, goes even further. He coined a phrase that describes what editors Staenberg and Roger Barton did: “editography.” This term essentially suggests that pieces of the traditional role of the cinematographer were taken over on this project by editors. Further, he suggests much of the inspiration of this approach emanates out of the same source as the inspiration for the images themselves in Speed Racer — classic, cel-animated, anime-style cartoons.
“If you step back and view the more masterful [classic anime works], you see an overtly expressive quality to the presentation of emotional moments, including a kind of pseudo-cinematography, or customized lensing, if you will,” Gaeta says. “If you forget for a moment that it was cel-animated, the lenses suddenly become whatever you can imagine — mixing focal lengths from foreground to background, experimenting heavily with focus, and especially with motion blur to such a degree that they push it to the borderline of hyper-real. So our goal with photo-anime was to build a universe that created a photo analogy to that aspect of anime, and that meant developing some rules: pure colors, very poppy, the use of color to define space and depth, nuanced camera motion, pushing more expressive production design, super clean, texture-less, no heavy grain or noise, and so on.”
To accomplish all this, certain decisions were made — none of which were in and of themselves brand new, but when all lumped together, certainly resulted in unique imagery and a unique workflow. Among those decisions: shooting HD in order to capture the opposite of a film image; selecting Sony’s F23 digital-cinema camera as their primary acquisition tool; shooting the entire movie greenscreen; building complicated virtual backgrounds out of high-resolution, digital-still imagery in the form of “bubbles,” as the production called them; and heavily layering virtually every single shot in the film and routinely playing with focal length and perspective, as dictated by the directors.
That meant those involved in the greenscreen shoot were there essentially to capture elements that would be stitched together in various forms later on. This is where Gaeta’s concept of editography comes from.
“The cinematography, in general, manifests itself more after the live-action shoot,” he says. “One can take the whole idea of lensing and customize it — combining focal lengths across layers, creating impossible depth-of-field situations. For example, creating infinite depth of field for telephoto [POV], or having extreme narrow depth of field on a fisheye. Lensing in that sense could essentially become something you design yourself.”
Consequently, the job of the cinematographer was different on this project than on more standard fare, and cinematographer David Tattersall freely admits this.
“There are changes here for the cinematographer — technical, procedural, and collaborative,” Tattersall says. “On Speed Racer, we had fuzzy job descriptions. It was more of a team sport. And, yes, it was unusual for me to not be shooting all the principal photography. [Backgrounds] were shot by a still unit, some principal material was farmed out to second unit, and all race scenes involved comped material. But, going in, I was aware of this approach, so we all got into this thing of sharing, and that largely happened in post. Usually, the DP concentrates squarely on the images. It’s his job to follow through and control the photographic images, from preproduction through to the finished print. But that was not the case with this show. A graphic strength or weakness of an image might happen to be shaped in the editing room — re-framing cars and backgrounds, making midground and foreground choices.”
Speed Racer‘s visual artists used gamma warping during the DI process to change color contrast in certain spots without affecting other parts of the frame. Photo courtesy of Warner Bros. Pictures.
Therefore, while Tattersall was shooting live action on greenscreen stages, a small still-photo team fanned out across the globe (northern California,Europe, and northern Africa) to capture high-resolution still images with a series of Canon EOS-5D 12.8-megapixel cameras to build a library of panoramic, 360-degree, spherical, virtual backgrounds — images taken from dozens of perspectives at each location. Instead of merely stitching some of those images together, however, they were transformed into the equivalent of ultra-high-resolution QuickTime Virtual Reality-style “bubbles,” inside of which elements from the live-action shoot, or CG material, or both, could be inserted and seen from multiple, changing perspectives.
These bubbles were, essentially, sliding transitional elements between shots to express changes in time and space and emotion during the movie, making the entire picture, in the words of Digital Domain’s co-visual-effects supervisor Mohen Leo, sort of “a flowing collage, constantly transitioning.”
“We used to call these virtual backgrounds — projecting them onto relatively simple geometric forms to represent buildings or structures,” Glass says. “With them, you could recreate camera moves to get a sense of parallax, of perspective changing, but with high-resolution and photographic textures. But Speed Racer’s origins come from cartoons. We loved that the original cartoon series was made out of simple, flat layers — and that had a charm unto itself, and a visual familiarity. That’s different than virtual reality, but we wanted to make a photographic version of it for our photo-anime world. Thus, in a sort of retro fashion, the bubble backgrounds are not reprojected, but presented flat to camera with clearly warped perspectives at times.”
Pieces of live action from Tattersall’s greenscreen material were also captured from multiple perspectives and layered into virtually every shot, and the editors then added those bits and pieces to construct shots in post for even relatively simple situations.
“I like to call it ‘intraframe editing,’” Staenberg says. “On this movie, the creation of these kinds of shots would finish up in the editing room. Each character was shot in a separate pass, and we could slide them around for composition and choose where the focal plane would be, and we could work on the size of the image since we had such a high-end camera. We could play around with blowups as much as we wanted to, and shrink things down as well. Final composition and focus and image size in the frame were often determined in the editing room. As a simple example, we might have a shot that would look like an over-the-shoulder two-shot conversation. But each character was shot as a separate element, a separate pass without the other character. So, when you put them together, it is more flexible. You can edit both characters independently, you can speed them up, and work with them within the frame of the final shot.”
The epic track-racing sequences in Speed Racer were almost entirely computer-generated affairs as opposed to the rest of the movie and the rally races, which incorporated live-action photographic background elements. Photo courtesy of Warner Bros. Pictures.
Staenberg’s editing partner, Roger Barton, worked with Tattersall on the second Star Wars trilogy, and therefore, he was well familiar with all-digital workflows. (Speed Racer’s
editorial team used seven Avid Adrenalines hooked into 4TB of mirrored
Unity storage on location in Germany during the greenscreen shoot, and
added three more Adrenaline systems when the editorial process returned
to the United States.) But as he divided up the work with Staenberg,
who has now cut five movies for the Wachowski brothers, Barton says he
realized the cutting style on Speed Racer was radically different from anything he has ever worked on.
“The idea behind it was to take all your
selected takes and feel as though the camera was flying through the
action, panning and zooming to find the next piece of action, so that
whatever you wanted to reveal next was either stacked behind the
foreground, or just out of frame,” Barton says. “And we’d have to
design all these whips, zooms, and wipes — always keeping in mind where
we wanted to go next.”
Because final shots were built during the
editorial phase, however, it required the production to design a
workflow that would allow filmmakers to visualize those shots on stage.
The first part of that choice involved selecting the then-brand-new
Sony F23 digital camera system as their primary acquisition tool,
picking it from their then-available set of camera options largely
because its single 2/3in. sensor allowed them to capture greater depth
of field than they would have been able to capture with a film camera
or other available digital cameras at the time.
They next chose to record an uncompressed
4:4:4 signal from the F23 to both Sony HDCAM SR tape and a Codex
hard-drive recording system simultaneously, with the Codex being used
as the platform for efficient file management to send data off to the
various entities that needed it almost as soon as it was shot,
including an on-set compositing system developed by Digital Domain
engineer Brian Smith, dubbed “Sparky.”
“Sparky basically ran on a standard
Intel-based videogame PC you might use for home gaming, running a
Nvidia  graphics card and a Blackmagic Design Decklink HD Extreme
video card, and we developed code to take advantage of the incredible
processing power of that modern GPU,” says Digital Domain’s Kim
Libreri. “That gave us an on-set compositing system to take the live
feed out of the F23, and combine it with [a library of virtual
The use of Codex hard-drive technology also
allowed the production to instantly export DPX format data for the
visual-effects team to use in building temp shots, and MXF format data
for editorial to ingest efficiently and use for cutting without a lot
of time wasted on format-conversion work.
“The Codex system, and its transparent ability
to facilitate the data flow, while allowing us to operate in a more
familiar way, was a linchpin to accomplishing the throughput of the
photography,” Gaeta says. “That system, in combination with the F23,
allowed the directors to make final decisions on stage, and that was
Later, in keeping with the theme of building
the movie’s final look in postproduction, the digital intermediate —
handled in collaboration with colorist Maxine Gervais at Pacific Title
— was where the movie’s exaggerated spectrum of colors was executed.
“There was a lot of color in the light, in the
set design, in the costumes, of course, but in the DI, we did gamma
warping — using digital filters to manipulate the contrast in selective
areas of the color space, which let us change contrast in certain
places without affecting other parts of the frame,” Tattersall says.
“That’s how we baked the exaggerated color into the negative. There is
an exaggeration to the color saturation that we added — beyond what is
really possible in the photochemical world.”
Many backgrounds in the film were designed and built during the editorial process, using high-resolution digital still-photo image elements captured in places ranging from Europe to Northern Africa. Photo courtesy of Warner Bros. Pictures.
Among the end results of these processes used on Speed Racer
is the fact that the approach has left a lasting impression on the
people who made the movie. They point to the crossover approach as
having infiltrated virtually every aspect of how Speed Racer
got made. Digital Domain officials, for instance, point out that much
of what would have once been considered classical art design
essentially merged with visual effects. The film’s lead concept
designer for production, Jeff Julian, for example, also served as
internal art director and spent much of the production integrated into
the visual-effects team at Digital Domain, helping to design virtual
environments, the race tracks seen in the movie, and so forth.
Barton, meanwhile, considers such moviemaking as so overwhelmingly collaborative that he says that Speed Racer “is neither directed by Larry or Andy, but rather, by one collective spirit.”
And fellow editor Staenberg says, philosophically, the whole concept “can’t help but give you pause.”
“It seems likely to me that we have entered
into an ‘age of postproduction,” where post takes on a bigger role in
production of the movie,” Staenberg says. “I’m not exactly predicting
it, but it is possible. This kind of movie shows you can create things
you couldn’t create before, in terms of camera moves, shots, and making
locations without going to them. That changes the dynamics of
production and shifts the balance toward postproduction for certain
kinds of movies.”