All that matters is if the audience will believe it on the screen. The fact is that people who know nothing about how these things are done can still tell us whether the effect is good or bad. You don't have to know what a 10K is to know whether you like or dislike a movie, and that's something we tend to forget. We say, "What do they know?" But they know. They've used their eyes all their lives and they know when something doesn't look exactly right. There isn't a little sign that says, "matte line" or "jiggle" or "dupe". It just says, "Oops!" It distracts them. It takes their minds off the story. The Empire Strikes Back has a terrific story, an interesting story, a strong storyand that's where your mind should be. Anything technical that jars you a little bit is going to distract you. So your effects have to flow and fit in neatly. I hope they do in this picture. I think they do. I've seen it once and I was quite pleased.
For our front-projection set-up we had a Bell & Howell camera box, which we also used for the bi-pack capability on top with a 100mm Cinemascope (Bausch & Lomb) lens.
The projector, which was a VistaVision projector, was constructed at ILM; the lamphouse was designed and constructed partly at ILM; some of the optics were designed outside. So the lamphouse, which was sort of made from scratch, was set off 90 degrees from the camera. We had a beam-splitter piece of glass through which we projected the image onto front-projection material, which we put behind the glass. As a result, because we had this material behind the glass, we had to tilt the painting slightly so that you wouldn't see a direct reflection of the light source back into the camera. That kind of solved the problem.
We didn't want to put the front-projection material on the front of the glass initially because we weren't sure how much of the image we would use and we thought thatwould create more problems for us. But we tilted the glass (we had front-projection material in back of it) and then generally shot the painting and the plate in two passes, or, in the case of separations, obviously in four passesbut we shot the paintings separately from the front-projection. We did this not because we had to (you could turn on the lights and shoot the painting and the plate all in one), but because it gave us a little more control, in that we could vary the amount of light on the painting in relation to the plate. We could do exposure wedges on the painting, as well as the plate, and bring the two of them together, rather than having to be locked in to one set of exposures for both. That offered us a bit of advantage.
The camera had a motion-control device that allowed it to tilt and pan. Obviously the camera lens was not a zoom, so we were not able to move in or out, but we found that we could eliminate a number of moves from the paintings. I had originally instituted moves, but found that they really weren't adding anything to the shot, so I moved probably far less in this picture than I did for The Black Hole. I think that's also one of the reasons that the shots are quite shortone second, two second, four seconds, maybe six seconds; the longest one was eight seconds. That's a pretty short amount of time and introducing a move could sometimes be a little disorienting and not particularly prove anything. There is nothing worse than an unmotivated camera move. I think that probably out of all our shots we had maybe three or four moves. It wasn't because the equipment wasn't up to it, but simply for the reasons I've stated.
The type of front-projection we were using is different from conventional front-projection in which you have live action going on in front of a projected plate, but it creates its own set of unique problems in that the screen is much closer to the camera than 3M (screen manufacturer) recommends. You have to be much more critical with your line-up in order to avoid matte-line problems. Anytime you try something that is markedly different you may solve one problem, but create three more. It's like going ahead one foot and slipping back three. But this method had been used fairly successfully on The Spy Who Loved Me and that really got Richard Edlund thinking that front-projection would be a very viable method for The Empire Strikes Back.
[ continued on page 4 ]