Following the success of James Cameron’s Avatar, a 3-D science-fiction epic that became the highest-grossing movie of all time, many producers chose to convert previously filmed features to 3-D prior to release. Many have criticized the results. How do you convert a movie to 3-D after it has already been shot?
Draw a map of each shot and let a computer do the rest. 3-D movies are normally filmed using two slightly offset cameras. Both images are projected onto the viewing screen, with those cheap plastic glasses feeding one image into your left eye and the other into your right. Without the benefit of the second camera during filming, producers have to generate the two offset images based on the single flat picture that they have. The first step is to separate the shot into somewhere between two and eight layers of depth. Take, for example, an image of a man standing in front of a brick wall, with a blue sky behind the wall. The graphic artist might separate the shot into three layers: the man, the wall, and the sky. Then, he would take each layer and draw contour lines around any object that appeared there. He’d start by marking depth lines on the man using a computer, turning the image into a sort of topographical map. He’d repeat the process for any objects in the other layers. (If there were a bird in the sky, he’d draw lines there, too.)
Once this is done, the computer takes over. Software creates a new, offset image of the man by moving the various regions of the contour map to the left or right and smoothing everything out. The closer bits—the tip of the man’s nose, for example—would be moved the farthest, while the more distant parts—the back of his shoulder—would be displaced a bit less. Then the process would be repeated for the other two layers of the image: the wall and the sky. (The former would move only slightly; the latter hardly at all.) Broadly speaking, this process must be completed for every object in every shot of the whole movie—an undertaking that might take months, even with a team of 30 or more artists. (The workload depends in part on the amount of chaotic motion in a scene. If an object is relatively still, the artist can map out a few representative frames and let the computer interpolate the rest. Otherwise, he might have to work frame-by-frame.)
So how does a converted 3-D film compare to one that was shot with a genuine 3-D camera? It’s not as good, but you probably wouldn’t notice anything amiss unless you’d seen a lot of 3-D flicks. During the conversion process, the artists and the software have to fill in a lot of blanks. Consider the example above, where the image of the man is shifted against a background of a brick wall. That displacement will leave a blank space in the image—the portion of the wall behind him that wasn’t in the original picture. At that point the artist has to cut a piece of the background from elsewhere in the frame and paste it into the man-shaped hole. If this cut-and-paste job isn’t done perfectly, even an untrained viewer will get a sense that something isn’t right. Another problem arises from the fact that the shot has been converted into three, four, or eight layers of depth, sort of like the way digital music is a series of snapshots rather than a continuous wave of sound. A stereoscopic camera shoots an infinite number of layers—so it produces an image that more closely resembles what the human eye sees.
Still, some directors of 3-D movies choose to convert after the fact, rather than shooting with stereo cameras. One reason is cost. Top stereographers charge millions dollars for feature films. A full conversion from 2-D to 3-D usually costs somewhat less, although it can run into seven figures. (Some conversion companies are now saving money by outsourcing the work to Asia.) Another reason to convert is unfamiliarity. Shooting a film in 3-D requires some careful decision-making so as to maximize the depth effect while minimizing potential eyestrain. Directors may feel constrained by these limitations. In any case, not every 3-D director agrees that conversion works just as well. James Cameron, for one, has criticized Tim Burton for using this approach in his upcoming feature, Alice in Wonderland: “It doesn’t make any sense to shoot in 2-D and convert to 3-D,” he said.
Explainer thanks Tony Ascroft of Tony Ascroft Productions *, Eric Robinson of Mr. X Inc., Daniel Symmes of the 3-D Film Preservation Fund, and Ray Zone of Ray 3D Zone.Correction, Feb. 1, 2010: The original misidentified one of the sources for this piece.