At this time, the ultimate image quality, least apparent grain, highest resolution and accutance, and best tone range and color saturation are produced in first-generation, or camera-original negative. Copies or generations of film derived from an original exhibit the familiar degradation of each of these qualities. In special effect matte photography, the most difficult and rewarding results are obtained when all components in a shot are composited “in camera.” This method, also called the “held take” technique, places all the difficulties of color balance, exposure, film shrinkage and matte size/fit in the hands of the miniature stage photographers. Many shots in 2001: A Space Odyssey were produced in this manner. However, in a film such as Star Wars, where the shots pose difficulties in miniature photography which have not been attempted in such quantity, the held take method would preclude a realistic production schedule. Our approach was to transfer the difficulties of composite matte photography from the miniature photographers to the optical cameramen. This achieved an effective balance between capturing the images on film, and finishing them into composites; it also gave the optical cameramen the challenge of maintaining maximum second generation quality in the optical composite negative.
It was decided to use black and white three-color separations for all primary images in the composites. The emulsion of 5235 color separation stock has a wide contrast latitude and grain definition which is superior to 5243 color interpositive: it is the best choice for quality.
Any emulsion on standard celluloid shrinks and expands during developing. With a regular four-perforation image size, the shrinkage difficulties are largely controllable. In the double frame format (eight-perforation VistaVision), the shrinkage is a definite problem. Kodak Estar base was used to overcome this problem, as it is about three times more stable dimensionally than celluloid base. By using Estar we were able to have better stability over eight perforations than acetate would provide for a conventional 35mm frame.
High resolution over large image areas in one-to-one reproduction is a classic optics problem. A lens capable of resolving 160 lines/mm over a VistaVision format is roughly four times as difficult to construct as a lens for conventional 35mm. We researched all stock lenses and found that the 100mm f/2.8 RepNikor (lens #1) met our requirements. But this lens was only suitable for transferring an image from one projector into a printer camera. On an aerial image optical printer, or a printer with two projector heads, the transfer of an image from the film plane most distant from the optical camera necessitates both a second conventional copy lens (lens #2), and a “field lens.” Positioned behind the first projector film plane, the field lens is required to bend the diverging light rays coming from the furthest film plane, through lens #2, into the Nikor (lens #1). Introducing a field lens cuts the performance of any lens dramatically, and to solve this problem, we designed a special lens which required no field lens, while reproducing an exceptional 180 lines/mm resolution, with no distortion. This meant that a blue screen shot could be placed in the first projector, and its matte in the second projector, so that as far as the camera was concerned, the two images were essentially the same. Result: improved matte fit.
Having now attended to the problems of film quality, lens resolution and distortion, and film shrinkage, using the best solutions available to us, we were confronted with constructing optical printer projectors and cameras which were able to work in the largely obsolete double-frame VistaVision format. The essential mechanical problem in the VistaVision format is film weave across the short side of the frame. In conventional 35mm, the weave is only half as noticeable. Fortunately, we were able to draw on the experience of Jack Cauldwell and Paul Lerpae, who struggled with producing VistaVision composites at Paramount for such classics as The Ten Commandments. It was decided to use a third register pin, nine or more perforations from the standard two register pins to eliminate the weave. Even with the third pin, the printer movements required continual adjustment to keep this problem within bounds.
Laying this foundation allowed us to tackle the larger task of the optical department: to produce 360 multiple element composites, with 560 separate blue screen elements, and about 900 other elements (planets, stars, lasers, flak, etc.). We were required to develop techniques for matting fast-moving miniatures with streaked edges and matte them against planet backgrounds without generating matte lines, blue fringes, or double exposures—and do this in 32 weeks of production.
Basically, our task was to create a continuous tone matting system that could be controlled very accurately. Bill Reinhold, one of the most experienced Industry printer operators, provided invaluable support in approaching this problem. We began an intensive testing program to set standards for sensitometric control of separation and matte exposure. We sought advice from the labs and were greatly assisted by Wes Shanks at MGM and Ted Wilson at Deluxe General. We had to maintain a close relationship with the labs because the result is a function of the relationship between exposure and processing of the mattes.
We found that we needed eight or nine separate black and white film elements to produce a blue screen composite. Some of our shots involve as many as eight separate blue screen elements. That makes sixty-four separate B&W elements crossed over each other and the background. As the first ship streaked across the frame, the second ship it is in front of must be seen through the streak of the first, and so on down the line of criss-crossing elements.
Often we had to integrate animated Laser and Glows between the ships. The next level of difficulty came when, with all this going on, the ships being matted suddenly exploded. We used combinations of exploding miniatures, animation, and superimposed explosions. In any case, it is very difficult to matte the delicate smoke and tiny particles produced by explosions without defects. (Another complication was the 150 or so shots that had all this going on and also included a live action foreground previously photographed in England. A large number of these shots had reflections of the blue screen in the foreground, which required precise hand-drawn mattes on a frame-by-frame basis.) By far the largest matte difficulty was maintaining the integrity of the miniatures when they were small in frame: the smaller the image, the less quantity of grain defines the image, and the more likely it is that the image will disintegrate in composite.
Over ten thousand B&W film elements had to be prepared for printing (“lined up”), logged, cleaned, and expedited through the jungle of complications, mistakes, and directorial changes. Out of the problems of this project, a line-up system was developed by Paul Roth to rapidly mass-produce composites. All exposures and problems were logged as the work progressed. Some elements were used in a number of different shots. Some were flopped for changing key directions, some were flopped upside down and others needed their directions reversed. Often combinations of these, including added alterations to the original stage photography, occurred in the same composite. All of this had to be accurately recorded and accessed quickly to facilitate use of the elements.
At last, after months of running the printers 24 hours a day, the 365 shots lost their numbers, designations and individuality, and merged with the rest of Star Wars as a completed film.