American Cinematographer - November 2011 - 62

◗

3-D on a Budget

Left: Cinematographer Geoff Boyle, FBKS, who supervised the 3-D workshop, shows off a rig that combines two Arri Alexa cameras with a P+S Technik Freestyle Rig. Right: Thomas Harbers, a Munich-based post specialist, provided 3-D post for the workshop.

decisions in 3-D prep is the choice of convergence: which angle to adopt between the two cameras’ lines of sight. During the workshop we will try four different approaches to shooting 3-D: Parallel, Normal Converged, Derobe Converged and French Converged. When the two cameras’ lines of sight are parallel, the 3-D convergence is arranged in post by introducing a Horizontal Image Transformation. A HIT is simply shifting one of the images slightly to the left or right. HITs are measured in pixels, and will typically not exceed 40 pixels for a 2K image. On a set with parallel cameras, filmmakers can preview the HIT by using external boxes, like the StereoBrain Processor we used, to create a horizontal offset that can be fed into a stereo-capable monitor. (The StereoBrain is a 3-D HD videoprocessing unit developed for use in stereoscopic 3-D production, post and broadcast.) Boyle notes that “if you’re doing visual-effects work, it’s better to shoot parallel” because of the possible keystone effect in converged images, an artifact that is especially evident with buildings and other rectilinear shapes. Because
62 November 2011

parallel shooting involves a HIT in post, the composite 3-D image will have black bands on the edges. Harbers explains, “With parallel shooting you have to do the HIT in post, so you get black borders on the sides, and you often have to zoom into the image. This means recalculating the whole image, and the image becomes less sharp.” Boyle adds, “If you finish in HD, you need to shoot at 2K or 4K, because you will need slop on the edges.” When the two cameras are angled for Normal Converged shooting, their lines of sight will intersect at what will become the screen plane. Boyle says converged shooting requires more time on the set, if only because of the inevitable discussions about where to converge. One advantage of converging on the set is that the director and cinematographer can preview the 3-D with glasses, without the need for an intermediary box. Boyle adds, “The reason I stress IA in the workshop is that, along with pitch or tilt, it is the only thing that cannot be changed in post. Convergence is less important, because it can be changed later. I’m teaching low-budget 3-D, which has to be fast. Should I
American Cinematographer

discuss convergence on set, [where costs can run to] $25,000 per hour, or in post, at $500 per hour? The answer is quite clear to me!” “Theoretically, you have less work to do with converged footage in post,” Harbers says. “When you shoot converged I can take the image as it is, without scaling or interpolation.” However, he cautions that if you shoot footage converged, it’s difficult to make a major change of the screen position in post. “It won’t look right if you change the screen plane a lot. The relative offsets in the foregrounds and backgrounds won’t match anymore.” A third approach is the method defined by French 3-D pioneer Alain Derobe. In Derobe Converged shooting, the cameras are angled so that the lines of sight to the farthest object in the frame are offset by the interocular distance of 2.5". This fixed-convergence angle is typically small, between 0.5 and 1 degrees. The IA is then varied during the shot to follow the point of interest (for example, an important character in the film). The Derobe Converged method was used on Wim Wenders’ documentary Pina (AC Sept. ’10). This

American Cinematographer - November 2011

Table of Contents for the Digital Edition of American Cinematographer - November 2011