[ continued from page 2 ]
Matting methods utilizing UV lighting and the red emulsion bandwith were pioneered by Erland and Roger Dorney for the film Firefox (see AC Sept. 1982) and later adapted by Digital Domain for use on Apollo 13 (AC June '95) and The Fifth Element (AC May '97). With this technique, the frontlit beauty, key and fill passes for models are filmed against black; a backlight pass is then shot, with a screen painted with UV-sensitive paint positioned behind the subject. When hit with a Kino Flo, the screen fluoresces a very narrow bandwidth frequency of red, making the model appear as a black silhouette against red. The backlit image only registers on the red layer of the film, creating a perfect matte.
However, the solution itself became problematic on Alien Resurrection. Says Fichter, "I got a call to come over to see some com-positing tests. I saw some problems, but I couldn't understand what was causing them. I asked if they were using a different matting process, but because neither my French or their English was very good, we had difficulty communicating. It dawned on me that they were using a color-difference system, which employs different colors to create mattes. The technique we were using a UV-irradiated redscreen with a black core, which became the matte for the model was a problem for Duboi, because the black core had residual UV red contamination. In the U.S., where we use the more exact contrast-matting system, we eliminate that contamination by adjusting the contrast, but this was pure disaster with Duboi's matting system.
"Since we couldn't use a black core, we decided to put some color in that inner matte. The ideal color was the reverse of the UV redscreen, so we tried all kinds of combinations where we lit the foreground model with Composite Components' blue globes. We had to run a 2E filter for the ultraviolet light or we'd get crosstalk, and we had to separate all our passes, which became incredibly difficult."
Fichter and Erland soon devised a photographic method to function with the color-difference system in a single matte pass. The process employed antiquated filters to create the exact reverse of the UV redscreen color background for the core matte. "We'd go to the Kodak filter table and evaluate what color of light would be passed and absorbed by certain filters," Fichter remembers. "John and I figured that by lighting everything with blue light and using a certain magenta filter, we could have red on the screen and blue on the model, and nothing else, so we could get everything in one matte pass. I thought that test worked magnificently, but when we brought it to Duboi, we again had problems: the matte edges were not sharp.
"We ended up using a two-pass system. I would shoot the screen and the foreground blacked out, and then go back and reverse all of the lighting, backwind, and double-expose the blue core. But if the camera was not totally steady, it wouldn't work. We needed all of this intense light to create the God-light look, then we had to change everything for the blue-light pass, and then we had to change everything again for the UV pass. We had to move things constantly without bumping the model or the lights. We never knew what was going to happen. Once we started shooting, we just couldn't stop. The exposures were very long and we had a very high number of passes, so our shoots would sometimes last 15 or 20 hours! We would try to shoot certain passes in order, so that before we stopped shooting, we could dump some of the film in the lab and evaluate whether or not we were okay. Frankly, there were times when we found that somebody had slipped or nudged something; we then had to make an evaluation about which passes had to be reshot.
"Duboi had to rotoscope those shots that got bumped slightly," Fichter states. "Ultimately, though, the process really worked, and John was ecstatic because we had tested out all of these really theoretical things nobody had ever really done before."
Fichter believes that solving such color-difference matting problems may eventually facilitate more artistic and effective composites for future films. "If the software can be modified, the color-difference system could conceivably be better than the contrast matting technique," he offers. "Employing the theory John and I were experimenting with, effects cinematographers could use magenta filters to pass red and blue, hold back green, and ultimately shoot mattes in a single pass, including spill mattes. Given the efficiency of corporate filmmaking, whether we ever realize this process is another issue, but the color-difference system could conceivably be a very good way to go."