Still in its infancy, the labor-intensive process of converting 2-D images to 3-D images in post is becoming an increasingly popular alternative to shooting native 3-D. Recent releases that underwent the post conversion include The Last Airbender (AC July ’10) and Alice in Wonderland (AC April ’10). Most of The Green Hornet was shot 2-D and converted in post, and the movie’s stereographic supervisors, Grant Anderson of the Sony Pictures 3-D Technology Center and Rob Engle of Sony Pictures Imageworks, recently talked to AC about their work on the project.
American Cinematographer: Describe the stereoscopic-conversion process for The Green Hornet.
Grant Anderson: When you convert any 2-D movie into stereo 3-D, you have to think of every shot as a visual-effects shot. We scanned all 2,426 shots at 2K, and then the actors and objects in each shot were rotoscoped and split into separate layers. The layers were offset to give individual depth to each actor or object, and the gaps between the plates were painted in. During this process, you’re synthesizing the other eye. Some vendors keep the original scan plate as the right eye and only create a left, while others create both a left and right eye. Creating one eye means less data management, but creating two new eyes can result in less paintwork. These plates then come back for a depth review, where the overall depth and roundness of the various actors and objects are approved. Then the vendors go into a cleanup phase where the final paintwork is completed and any edge artifacts from stray roto are massaged out. The idea is to nail it without too many reworks.
John Schwartzman told us, ‘If you were to make a list of all the things you shouldn’t do when shooting in 2-D for conversion to 3-D, you’d find we did them all on this movie.’ He cited anamorphic lens flares, anamorphic depth-of-field and atmospheric effects as some examples. How did you overcome these challenges?
Anderson: All of these things make the conversion process more difficult. Dimensionalizing lens flares is especially challenging because of their transparent nature — the background comes with it. Great care has to be taken, or you have to remove them and comp CG flares back in. Atmosphere and smoke have the same issues: you need to put them at the right depth without pulling the background forward. When you’re looking at a shot in 3-D, your eye jumps directly to the thing that’s closest to you in depth. When you have a shallow depth-of-field in 3-D, the foreground object your eye jumps to is a blurry mess, and the scene becomes a strain to look at. If you were actually shooting in stereo, you would deepen your depth-of-field to make those objects a little sharper. With conversion, we don’t have that advantage. The objective, then, is to minimize the degree to which those out-of-focus objects are coming out of the screen so they’re not sitting in your lap.
Many cinematographers have expressed concern about the stereoscopic-conversion process, particularly when they’re not involved in it. What would you say to them?
Rob Engle: The most important thing we can tell any cinematographer is that his or her work is always the starting point for our work. We’re not changing the original composition or lighting.
Anderson: We try to stay true to the depth of the scene as it was shot. This goes for overall depth and also for the roundness or internal depth of the actors and objects in the scene. When you stretch it too much, it becomes very obvious that the scene is no longer realistic. It becomes a caricature of itself.
How much input did John Schwartzman and Michel Gondry have in Green Hornet’s conversion?
Anderson: Michel, Seth [Rogen] and others had a lot of input, John not so much at the post stage because of scheduling. Michel had a clear vision for the 3-D tricks he wanted to perform. He wanted to exaggerate the 3-D, giving slightly more volume to the actors and the scenes. He was always talking about how he loved his [Fisher Price] View-Master as a kid.
Engle: Even though John wasn’t on hand during the post conversion, he and I were able to collaborate on the live-action portion of the shoot. It’s becoming less common for productions to decide to convert to 3-D after they’ve wrapped. I’ve had many conversations with directors, cinematographers and visual-effects teams during prep, and I’ve even gone on technical scouts to discuss the best ways to shoot with 3-D conversion in mind. I encourage any cinematographer who’s informed about a production’s decision to go 3-D to reach out to the stereographer.