A world of hurt (page 2)
Note: Plot Spoilers!
|
Photo: The film commences with a ninety-second pullback within the narrator's skull. Digital Domain
handled the brain passage sequence, the design of which evolved from authentic renderings by a medical illustrator. Reflecting the director's wish that the environment be evocative of a night scuba dive, the brian passage featured seaweed-like particulate matter that appeared to interact with the virtual camera. Arcing electrical forms known as action potentials were also revealed during the shot, their appearance mimicking that of objects recorded via electron microscopy. The camera eventually exits the skull through a hair follicle, emerging on the narrator's face, above his nose, before tracking down to reveal a handgun wedged in his mouth. Skin, hair and sweat beads were additional CG elements created, animated and rendered at DD, as was the pistol, matched to an oversize model used for the only-live action component of the shot - a final rack focus to Edward Norton.
|
Fight Club begins with a passage through the human brain, over which the main
titles are displayed. Though the director had always intended to open the film in this
fashion, budgetary concerns had kept the title sequence from being awarded until January
1999 - leaving Digital Domain with a fairly short turnaround time for producing the
ninety-second shot. Fincher and Haug had instigated extensive previsualization early on,
enlisting the assistance of medical illustrator Kathryn Jones to develop the brain trip
with ridefilm-like action, and utilizing PLF for the later part of the shot, which emerges
on the surface of the narrator's face and tracks back to reveal a gun. "Kathryn put together
a whole book of illustrations for us," Haug noted. "She also made up a map that showed
all the various 'rooms' of the brain and the 'doors' that open onto them. From a scientific
point of view, she clued is in to the fact that the brain was basically colorless, which fit
David's monochromatic intent. But he made it tough on her design-wise. While he wanted to
keep the brain passage looking like electron micorscope photography, that look had to be
coupled with the feel of a night dive - wet, scary, and with a low depth of field."
After extensive consultation with Dr. Mark Ellisman, profewssor of neuroscience and
bioengineering at the U.C. San Diego School of Medicine, Digital Domain wound up departing
from the initial shot design, mainly to boost visual interest. "We decided to change
some areas of the brain being shown," noted DD visual effects supervisor Kevin Mack, "but
the overall cinematic concept remained the same. The camera would always be pulling back;
and we treated the shot as though there were an actual physical camera moving through this
tiny space."
The shot begins in the synaptic cleft, where neurotransmitters are released to blanket
a huge spine-like formation of dendrites, eventually revealed to be part of a much larger
dendritic forest, with turbulence causing dendrites and axons to sway in their liquid
surroundings. "To grow a digitally created tree that figured prominently in What Dreams
May Come, Mack had employed L-systems - used in botany to classify plants by their
branching structure - rather than model the complex organic form. The even more dendritic
shapes on display behind Fight Club's titles were realized through essentially the
same approach. Mack began by growing neurons, then handled the work off to animator David
Prescott, who took L-systems much further to grow other aspects of the brain, then
animated the brain matter with Houdini.
Photo 1: Digital Domain computer graphics supervisor Matthew Butler and animator David
Prescott at work on the brain passage shot. The pullback incorporated extreme twists and
turns reminiscent of ride-film camera moves.
Photo 2: A wire-frame rendering from the brain passage. DD visual effects supervisor Kevin Mack elected to utilize L-systems - an approach that allowed branching elements to be 'grown' rather than modeled - to create the brain's highly complex dendritic forms. To reduce overall render times, individual models were instanced, allowing them to be placed in a variety of locations within the frame. Dendrites were animated to sway appropriately via a low-frequency noise function. |
While L-systems added a large measure of complexity to the brain matter, another
approach - instancing - would increase the density of imagery even further. "For an object
that's going to be repeated often during a render," Mack explained, "it was possible to
instruct the computer to load a single model, then 'instance' it, rendering the form in
many different places. This was preferable to loading the same geometry over and over for each use, and the technique allowed us to have fairly complex dendrites and axons. While we had created a small variety for each form, this number was made to seem musch larger, simply by orienting the instanced geometries differently. By nesting instanced dendrites among axons and spines in this fashion, we avoided a cloned look. Instancing provided vast amounts of geometry without our having to pay for terabytes of RAM."
The nooks, crannies and pores on the surface of dendrite and cell bodies necessitated the construction of rather complex profiles, which were created through the use of displacement shaders written by Dan Lemoons and Darin Grant. " These shaders were used for animation as well, which was something new," Mack commented."Global deformation couldn't be applied to animate the various pieces of geometry - that was the drawback to instancing. Instead to create the swaying action within the dendritic forest, we added a low frequency to the shader, causing it to displace the geometry in world-space, producing an animated effect. More and more, I believe we are going to be getting away from the geometry/surface paradigm and moving toward shader paradigms for solutions." In addition to the work of Prescott and Lemmon, some brain matter was hand-animated by Chris Yang, with CG supervisior Matthew Butler providing oversight.
To help convey the notion the of an actual camera filming within the brain, a shallow depth of field - achieved via ray tracing - coupled with a roving focus, was utilized. "We choreographed events to pass camera in as interesting a way as possible," said Mack, "which meant animating this shallow-focus focal plane. If something of particular interest passed, the camera follow-focused on it for a time before racking away to some new object, directing the viewer's eye to the next area of interest."
The camera pulls out of the brain through the skull, passing fat cells and entering a skin pore. Then it emerges onto the narrator's face by exiting a hair follicle above the nose. "Huge beads of sweat are seen on his face, and our passage bends eyebrow hairs while we descend along the nose," Mack continued. "Since the character has been badly beaten, there is caked blood visible on his face, in addition to blackheads and nose hairs. All of these details were developed by Judth Crow, who integrated the eyebrow hair and follicle - realized by Zachary Tucker - into the forehead as well. Crow's sweat beads refracted the multitextured skin beneath, via Darin Grant's software adaptations, and could run together and pool while bending the eyebrow hairs."
|
Descending past the upper lip, the camera retreats from the narrator's face and enters metallic-looking groove - pulling back along the length of what turns out to be the sight of a handgun. The CG pullback ends at the weapon's rear sight, with DD creating a transition that blended its digital scene with production's live-action rack-focus, which begins on an oversize gunsight and winds up on Edward Norton's profusely sweating face. PLF's previz for the camera track-back along the gunsight proved so reminiscent of the Death Star trench from Star Wars - one of the director's favorite films - that a joke version of the animatic was produced, with a pair of TIE fingers popping up to chase the camera.
Holding the gun wedged into the narrator's mouth is Tyler Durden, calmly awaiting the destruction of several downtown highrises that his urban terrorist forces have rigged for detonation. From their vantage point on the upper floor of another building nearby, the narrator imagines the pyrotechnic setup far below, requiring another effort from BUF. "This was another case of snatching the audience's eyeballs from their heads," stated Fincher, "dropping them over thirty stories while traveling at the speed of thought. The camera needed that kind of incredible fluidity to keep pace with the narrator's voice-over. The shot came to me while we were on the highrise building set, when I decided to shoot a plate looking into the set from outside the window. I told Pierre Buffin we'd need a reflection for the window glass, and after we did a tilt-down move off the set, he would have to match that and carry it down the side of the building, right through the sidewalk and into an underground parking structure. From there the camera flew through a bullet-hole in the windshield of a van to reveal containers of nitro inside. Next, out we go through the van wall and across a six-lane street - passing through the van wall and across a six-lane street - passing through hideous corporate landscaping - and into another garage with its fifty-gallon drums and bomb timer." BUF used nearly one hundred photographs - shot in downtown Los Angeles and Century City by Michael Douglas Middleton - to serve as a basis for its digital 3D street and parking lot models.
With this moment of crisis established, the film flashes back to the narrator's earlier life - before he is introduced to Durden. But even in these scenes, Durden makes a kind of appearance via subliminal single-frame cuts of Brad Pitt. "The idea - which doesn't pay off until much later - is that our hero is creating Tyler Durden in his own mind," Fincher noted, "so at this point he exists only on the periphery of the narrator's consciousness. The Tyler frames are all images of Brad Pitt making eye-rolling expressions that seems to say: 'What is it with you - come on! Let me spring from your forehead fully developed.' But, no one will see any of this in the theater - we did this for the DVD crowd."
Gray Matter FX was responsible for nesting these subliminal images into background plates. The company also provided effects work for several other shots, including an amusingly self-conscious moment when Durden - a sometime projectionist with a penchant for splicing single frames of pornography into family films - casually points to the top right corner of the film frame. A changeover mark - the 'cigarette burn' flash that alerts the projectionist to switch reels during a showing - appears, as if Durden has conjured it out of midair. "That was my way of pointing up the break that is coming later in the story," said Fincher. "When a projectionist handles the changeover of reels correctly, the audience is never aware of the fact. That becomes a thematic element in the movie. At the beginning, the audience is presented with a fairly subjective reality - we jump around in time some, but otherwise remain pretty consistent, so the changeovers seem to be going over as planned. But that reality is completely subverted later in the film - suddenly it's as though the projectionist missed the changeover, the viewers have to start looking at the movie in a whole new way."
Having fallen prey to the rampant consumerism that drives members of his generation, the narrator arranges to populate his apartment with trendy furniture. "We see him ordering this stuff on the phone," Kevin Haug stated. "The camera pans the room, which fills with his purchases - couch and chairs, plus funny little lamps and vases with bubbles in them - becoming a virtual catalog showroom."
A large number of motion control passes were executed onset to aid in the assembly of the shot, which consisted of fifty two layers. "We pulled difference mattes to isolate each piece of furniture," recalled Nick Brooks, who worked in conjunction with PLF to composite the shot. "When an object goes into a room, it creates a lighting changes - not just adding shadows, but altering the room's overall ambience by occluding and reflecting light, which created a second level of sell to the whole thing. The appearance of each piece of furniture was preceded by a fading-in of the ambient lighting effect caused by that item's presence. This kept the transitions from looking abrupt and unnatural." As the furniture pops in, catalog captions - generated at PLF- turn up, hanging in the air. (Continued)
If you have new information on Edward Norton (and you can provide a
verifiable and reputable source), please email me-
Susan
Note: Articles and images have been posted without permission for noncommercial and nonprofit use
with no intention of copyright infringement. The purpose of this reprinting is to disseminate correct information about the
actors, films, and studios. I have included author names and links to sources whenever possible.
