Step By Step: The Day After Tomorrow
In 20th Century Fox's The Day After Tomorrow, director Roland Emmerich presents an apocalyptic vision quite different from his famed pyrotechnics for Independence Day. This time the threat comes from environmental disaster, and one shot of a massive storm over New York City illustrates what visual effects supervisor Karen Goulekas calls “the big freeze.”
“We start on the needle of the Empire State Building, looking down at all the rooftops below,” she explains. “As the camera tracks down the side of the Empire State Building, frost begins covering it, and the freeze causes the windows to begin blowing out. We come down lower and start to pan and tilt a little so we're looking past the building to the rest of the city. Those buildings are also becoming covered in frost.” On the streets far below, people run for cover before the freeze reaches the ground. As if that weren't ominous enough, the shot ends with a view of a “super-cell” of turbulent clouds in the distance, which is the rim of a giant hurricane that has engulfed the city.
While several effects companies contributed to The Day After Tomorrow, including ILM and Digital Domain, the San Francisco-based house The Orphanage was chosen to illustrate the big freeze. Leading that crew was visual effects supervisor Remo Balcells, who characterizes this particular shot as “250 frames of pure invention.”
The process began in 2002 with previsualizations done in Maya by Emmerich's inhouse team and a spin-off group called Crack Creative. Next came the task of amassing information on the Manhattan cityscape, which included purchasing data about the basic geometry and placement of buildings from Urban Data Solutions. “Having the correct heights of buildings in the right position allowed us to mock up the camera moves,” says Goulekas.
Once the camera moves were decided, she enlisted Lidar Services to map the area that the camera would see. “We looked at the streets we'd be using repeatedly, which was about 135 buildings,” she recalls. The production also hired three teams of photographers who shot reams of still photographs that could later be used as texture maps on the 3D model buildings that The Orphanage would create.
Working in Maya on Windows PCs, The Orphanage used the databases to build groups of 3D buildings for the foreground and middle ground. Distant buildings were handled as 2D elements. Particle effects were used to create snow flurries as well as the broken glass that fell from the bursting windows. The Orphanage attached flat polygons to the latter particles and they spun in the air like shards of glass. Balcells' team also created “little speckles of actors” as CG elements, animating those models with motion-capture data.
Fortunately, the CG humans were seen from afar. The Orphanage's up-close challenge was creating the frost that rapidly consumes the Empire State Building as the camera pans down. “It's an animated texture,” says Balcells. “When frost is happening on surfaces, it generates a pattern that looks like a flower. Roland called it an ice flower. It's generated as the frost moves forward and little tentacles grow and expand at the same time.” (Goulekas reveals that the development of this look owed a debt to a fake ice kit for kids and the application of a hair dryer.)
The Orphanage's frost effect comprised several 3D-CG layers. “A big challenge was to find a balance between the gray building and the white frost,” says Balcells, “A good transparency showed the detail of the frost and also the building underneath. Anything too opaque looked fake.” The frost included a reveal layer of finger-like motion that was developed procedurally in 3ds Max. Despite The Orphanage's broad use of Maya, Max was also used as a gateway to get CG files into SplutterFish's Max-compliant renderer Brazil. That renderer was chosen to get the most photorealistic look possible, and it was used for everything except the motion-blurred snow flurries, which were rendered with SiTex Graphics' AIR renderer.
Taking advantage of the global illumination and radiosity features of Brazil could have been a compute-intensive proposition, but Balcell's team applied an efficient approach. “We rendered around 12 key frames as close to reality as we could, with global illumination and radiosity. Those frames were handed out to the matte painters and texture artists to add even more details. We then did 12 expensive renders where we baked in the radiosity, the global illumination, and the artists' enhancements. Then those 12 frames went back into the 3D world. We re-projected them through the camera view onto the geometry. It matched that geometry because that's where these frames came from to start with.
“The reason we needed 12 frames was because the camera moves so far that at some point we'd start to see tearing of the texture. When that started to happen, we replaced it with the next key frame that we'd rendered. We rendered about two-dozen frames using each of those key frames. By doing this, we got an enhanced photoreal render that wasn't expensive.”
“The frost elements were rendered separately,” adds Balcells. “Karen and Roland wanted the freedom to fine-tune the transparency in the comp, instead of having to re-render the whole sequence.” The compositing, done in Adobe After Effects, combined the models and animated effects with one final element — a matte painting of the super-cell. While that was done as a matte painting in Adobe Photoshop, Goulekas notes, “there's a little bit of motion added to it, to imply that the clouds in the super-cell are spinning around.” As if to add false hope to the impending doom of the scene, the massive wall of the approaching super-cell is lit by golden light.
In the end, notes Balcells, it was a purely computer-animated scene. “There wasn't a single pixel in there that came from anything physically real.”
Credit Roll
Middle Ad
CPN Newsletter
FREE eNewsletter
Subscribe to our FREE eNewsletter
Subscribe to our FREE eNewsletter
Middle1 Ad
print edition
