One Secret of WALL-E's Success

Iain Stasukevich has a delightful article on WALL-E in this month’s American Cinematographer (called “Expert Eyes Enhance WALL-E”). Unfortunately it’s not included in the magazine’s online material, so I’ll try to give it a little summary here, with all credit to Stasukevich of course. Those whose interest is piqued are strongly encouraged to seek out the original article on newsstands before the August issue takes its place (and look in June's issue for a discussion of Prince Caspian).

WALL-E was directed by Pixar veteran Andrew Stanton. He’s best known for directing Finding Nemo but he has also contributed to nearly every Pixar film, including writing several of their screenplays and codirecting A Bug’s Life. As with all Pixar films, the cinematography credit on WALL-E was shared between two technical directors, Danielle Feinberg over lighting and Jeremy Lasky over camera. Very early in the process Stanton knew he wanted to upgrade the film’s visual look over that of previous productions, to make it “more filmic.” To do so required revamping both departments and rethinking much of the process of how Pixar’s images are created.

The first step was reevaluating the focal parameters of Pixar’s proprietary animation software Marionette. For this a good old-fashioned Panasonic 35mm camera was brought on the premises for a screen test (they shot little cardboard WALL-E cut-outs) to compare visually with Marionette’s CG-equivalent lenses, focal lengths, and camera movements. The result was surprising for the programmers but confirmed Stanton’s suspicion: Pixar’s images did not match traditional film. (“That put a bee in their bonnet,” he said.) Among other problems, the depth of field (amount of material in focus) was consistently too great, at least for the anamorphic look Stanton wanted to replicate. Camera movement was also off, and other minor issues distinguished the CG images from the photographic.



So the programmers revamped all of their codes, getting the Marionette “lenses” to look and act more like actual lenses, including on focus pulls (the above shot is static, obviously, but notice the fuzzy background and even, unless I'm mistaken, WALL-E's left hand going a bit soft as it gets too close to the "camera"). For me one of the most interesting changes they made was in how the lens pans and tilts. Formerly Pixar’s “lenses” had been configured nodally, that is they rotated (tilted, panned) upon the point where they were located, causing a limited shift in perspective. Real cameras aren’t aligned this way, of course, but rotate behind the lens where the camera head is attached to the tripod, jib, or dolly. Adjusting the CG codes to move like a real camera added another layer of filmic authenticity, even if the audience doesn’t know why. “Most people will never notice it,” Lasky said, “but subconsciously, it makes you feel like there’s a camera in the CG space.” Eventually two styles of movement were adopted: a relatively loose “handheld” style for the Earth and a more locked down, track-and-dolly-based approach for onboard the spaceship.

So much for the camera. To make Pixar’s lighting procedures look more like live action Stanton made the logical choice to bring in a live-action cinematographer, Roger Deakins (above, shooting In the Valley of Elah), whose lengthy production credits include, most recently, No Country for Old Men. Deakins analyzed Pixar’s workflow and held a lighting master class for Lasky and Feinberg. One result was greater collaboration throughout the process between the two departments (previously all camera work was blocked out before the images were “lit,” a holdover from the days of 2D storyboards). On the screen, the result was a greater contrast ratio, the range between lightest lights and darkest darks; a lot of area was allowed to wash out into black, something generally eschewed in animation, or at least in Pixar’s films thus far. Particularly on the spaceship, pools of light were allowed to exist in their practical location with the characters moving through them as necessary. Every shot, in other words, wasn’t lit around the characters; the characters were allowed to exist in a lit space.

The result is a film that looks more filmic than Pixar’s previous cartoons. It’s not photorealism per say, although the CGI textures continue to improve, but the feeling that an actual camera was there in this cartoon environment to film events as would be done in any live-action shoot. And that is what Stanton envisioned all along.

(By the way, just googling Stanton turned up some other great material, like this round-table discussion on the Pixar blog and this interview on a film-related website.)