The Nightmare Before Christmas was converted after the fact to 3-D, something more complicated than it sounds. © Touchstone Pictures. All rights reserved.

The second pass is fairly complicated, as convergence points are adjusted and multirig is used to mix different stereo volumes for different elements within a single shot. "You might want to use a longer lens, which has the tendency to make the characters look like cardboard cutouts. So we will actually use a different stereo setting for the character to enable the character volume to feel nice and round without having the backgrounds go so deep with that lens that it won't hurt your eyes," he says.

The last step is McNally's, and it's mostly a blending pass in which he focuses on how shots are jumping across the edit and mitigated to avoid creating eyestrain for the audience. McNally says this kind of 3-D process will have a profound impact on filmmaking, though it will take experimentation and time for it to evolve beyond the traditions of 2-D. "If I kind of project out 10 or 15 years and assume 3-D has become successful and people have been able to develop the idea, I think we're going to see a much more passive camera style and composition," he says. "Meaning the frame and camera style will be more passive and the dynamics of a shot will come from the 3-D staging within the space."

At Sony Pictures Imageworks, Senior VFX Supervisor Buzz Hays says 3-D has been a learning process of constantly refining tools and techniques.

Each of the 3-D projects the studio has worked on has been different, and those differences have had a huge impact on how the work is done. For example, given the length of development process, many films were not conceived as 3-D films, and many that were converted to 3-D were done so either very late in the 2-D production process or after the 2-D version was completed.

Hays says that while it's possible to start integrating 3-D at any point, the best results and the most efficient processes come from being able to integrate the 2-D and 3-D versions as early as possible.

"The longer we wait to integrate 3-D into the process, the more it tends to be a separate process, or at least a separate department," he says. "Whereas if it's integrated into the show from the beginning, then the entire show is both 2D and 3D and there is no differentiation between the two."

3-D used to be a separate department at Imageworks on Beowulf (left) and Open Season (right). © 2007 Paramount Pictures and Shangri-La Ent., LLC. All rights reserved, © Sony Pictures Ent.

The 3-D was a separate department on such studio projects as Beowulf, Monster House and Open Season. That's changed and the 3-D process will be fully integrated into the current production, Cloudy with a Chance of Meatballs (opening Sept. 18), all the way back to the camera layout phase, Hays offers.

"It saves a lot of resources internally to do it that way," Hays says. "At least as far as we can predict, it's more cost effective to do it that way because the same artists are dealing with both the 2-D and 3-D."

The decision to do a film in 3-D still can come late in the process, especially when a studio is uncertain whether the cost is worth it. Hays says the standard number thrown around for the cost of 3-D is between 10% and 15% of the below-the-line cost, but that shrinks the more 2-D is integrated into the process to the range of 6% to 10 or 12%, he says. "But it's hard to say because every situation presents itself differently."

Additional pipeline issues come from there being few commercial tools for working in stereo. Animation studios and visual effects houses have had to come up with their own solutions and tools for automating the generation of a second image without having to render it from scratch.

One such solution, Hays says, is photogrammetry, or reprojection, where the depth and color info from one eye can be used to generate at least part of the other without having to go back and re-composite it.

Animation offers a certain type of control that live-action does not. Hays says CG animation allows each element to have its own virtual camera, allowing the depth of each object to be fine tuned. Adding a 3-D camera alters more than just the pipeline -- it creates new aesthetic challenges.

"There's a language to telling stories in 2-D that is distinctly different from 3-D, which hasn't really been fully explored yet -- in fact, it's been barely explored," he adds.

Hays says camera placement becomes the most important change 3-D brings to movie making. In 2-D, filmmakers have long relied on focus, camera position and framing to direct the audiences. 3-D is different in that it requires changes in framing, pacing and editing. The depth of field look that gives 2-D images the illusion of depth are unnecessary in 3-D, and the comfort of the audience becomes a factor as cuts that force eyes and brains to adjust too radically can cause headaches.

But that doesn't mean 3-D has to be slowed down, Hays says. "There are certain things to avoid so you don't make an uncomfortable experience, but that doesn't mean you can't cut quickly or use fast motion or anything like that," he says. "But I do think people need to experiment a bit."

Taking a slightly different approach is Blue Sky Studios, making its 3-D bow with Ice Age: Dawn of the Dinosaurs (opening July 1).

"What we've done is we've made a parallel pipeline for stereo versus the mono pipeline, which is our traditional pipeline," says Jayme Wilkinson, technical supervisor for stereoscopic development at Blue Sky. "At certain locations in the mono pipeline, we take data out to adjust additional camera data for the stereoscopic view of the second camera."

Though the process separates 3-D from a lot of the 2-D work in the Blue Sky system, the 3-D planning takes place early on. "After the previs layout, we'll take the 3-D data scene and bring that into our world of stereo and do some depth composition, using the tools and tech we've developed internally," Wilkinson says. "We'll set up how much depth do we want to get out of this, how much volume."

That 3-D info is put back in the standard pipeline as 2-D camera data and rides along with each shot through the process. After the shots are animated, the settings will pulled up and checked to ensure they still work for each shot.