An Imagina session dedicated to the subject of "bluescreen filmmaking" focused on Sky Captain and the World of Tomorrow. ™ & © 2004 Paramount Pictures. All rights reserved.

However, vfx supervisor Darin Hollings uncovered the tip of the iceberg in his presentation of the making of Sky Captain during the HD and vfx conference (hosted by Guerville, the co-author of this article). A completely digital workflow along with extensive preparation and previs allowed Hollings to provide his crew members every morning with a call sheet detailing each shot complete with camera rig and lighting placement. This enabled the entire crew to work extremely autonomously and effectively, achieving an incredible average rate of filming of one shot every 12 minutes. The grid system installed on the set and matched by a grid in the 3D space of a virtual environment enabled the team to synchronize perfectly real and virtual elements on the set with on-the-fly compositing.

New Technologies
Once again this year, the technological and research presentations included many image-based approaches. In his Vision for Animators conference, Steve Seitz, associate professor at the University of Washington, discussed the current limitations of standard capture tools that are often invasive, labor intensive and can only reproduce reality. He showed how, by extracting movement, texture or geometric data from video footage, it was possible to use live elements to recreate ultra-realistic animated sequences simply and effectively. For example, by analyzing the optical flow of water particles in a very short clip of a waterfall, Seitz’s model could make the waterfall loop endlessly and, simply by drawing a line representing a new course for the waterfall, it could recreate a new waterfall just as realistic as the original. From then on, one can "paint waterfalls" any which way, without even being limited to water particles; fire, smoke or other animated elements can also be used.

Seitz also presented the mapping of flowing video elements on a 3D particle stream. He used a video-clip of water pouring from a garden hose mapped on a Maya generated particle flow to animate the water at will, without the need of any kind of texturing, lighting or rendering. Seitz covered two other projects, one based on animated rotoscoping and the other on video-based facial reconstruction and animation.

At the other end of the research spectrum, Ron Fedkiw, assistant professor at Sanford University, presented ultra-realistic heavy-duty simulations of natural phenomena. With his approach, 3D objects are no longer hollow envelopes but instead behave like solid objects that can be squashed, deformed, melted or broken, just as real world objects would be. In addition to impressive water and smoke simulations that can take into account complex parameters of vorticity, thus recreating rich and subtle turbulences, Fedkiw also presented dynamic simulations of rigid and soft bodies combined together; for example, solid objects floating or sinking in a tank of digital water (and even scooping water from it), a water spout that deformed cloth, as well as cloth burning, ice cubes melting, etc. Every single movement and deformation was completely dynamic and animated by physics, not by manual control. A very impressive demonstration.