"We tried something new on this sequence that we hadn't done before. It is based on what Coppola did in One From the Heart. Ordinarily we don't deviate much from the storyboards without going through George first, but we came up with the idea of trying the whole sequence with little toy models even before the storyboards were done and shooting it just like a kid game. Joe Johnston and I shot this thing in about a week. It consists of a hundred cuts based on what we discussed with George and we did it with a little solid-state Hitachi camera. The set was about four by eight feet. It consisted of some two-foot high trees left over from E.T. and some cardboard tubes that had been painted brown, set on a piece of plywood with a carpet over it. We used some little rocket bikes about a foot long which one of the guys made out of some spare parts and put a couple of action figures on them. We hung them from little rods and, with me hand-holding the camera and looking at a video monitor, we found some really neat angles. It isn't static like a storyboard and you can move on it and see how the perspective changes throughout the shot. We did the whole sequence that way very quickly and then transferred it to film, which gave George an actual motion medium with which to cut the sequence."
"That worked out so well that we actually set up a video department and did a reel on the Rancor pit, of which we did two entirely different versions before we hit the final one. Then we did much of the rest of the show that way. The same cameramen were shooting the animatics who would actually be making the shot, so we were learning what the problems were and finding the angles as we were doing it."
Edlund pointed out that there are pitfalls to the method. "It didn't work as well for the space ships because it was so much harder to get the matte work right in the final version, making it difficult for George to get a good idea of the movements on the video. You have to be very careful. If you start cutting for the animatic and then you ask for the shots to be done like the black and white image of toys and props, you may find the frame count is all wrong when you go to a Panavision color comp of the shot. Color makes it twice as hard to recognize things in a short time, so you have to be flexible and usually have to lengthen everything by about 20% to 50%.
"For the bike chase, we first decided we could build a model forest and get a snorkel camera and shoot it that way. We did a rough mock-up of how big the set would be on our stage and figured how much stage time it was going to take. It was going to be a humongous set to get even a four-second cut, which is all we needed to get through any one scene. The set would have been about 80 feet long and the backing about a hundred feet wide just to get to the farthest view. And then once we got into—it, it could have been one of those things where we'd still be shooting on it months later and still be tweaking up the lighting and trying to make it right.
"So, I began thinking maybe there's another way. We took a plane up to the location, a real redwood forest, and I could see that we could get every shot. We could get all of the side views and three-quarter views from a car and then blue-screen in the bikes in front.
"The question was, how would we get the front view without seeing the road or a crane or something where we couldn't get a helicopter in to do it. If we set up a cable, the thing would just be going in a straight line. we thought of all sorts of ways—at one point we even considered getting Nelson Tyler's rocket suit, but that proved to be very expensive and dangerous. So what we came up with was Garrett Brown and his Steadicam.
"We did what Garrett called 'the human cannonball,' shooting with the Steadicam at about a frame a second, with him walking at normal speed with a gyro similar to those they used in Das Boot. The gyroscope is on the camera and the pathway is camouflaged over but marked out with positioners at intervals that he could line his eye up to. Over a run of about 400 feet it would represent about a four-second cut. The moves were very slow. If he would bend slightly to the right, it would take 200 or 300 feet, so he had to think differently. We had elevation marks too, so the camera would not go up or down too much, and he had a television camera on it with the telephoto in all the way, giving a much wider view than the camera lens, so he could see if the camera was drifting off, which it easily could do shooting at such a slow speed. The fact we were shooting one frame a second gave him time to correct each frame as he was walking, whereas if he had done four per second he might have gotten an incredible weave going on where he was trying to correct but couldn't because it was too late. This way he had time to zero in maybe three times on each frame and just keep a little target in the video.
"Then we shot on a couple of trips to the redwoods up north and each time laid out paths and had Garrett walk through. Later, we shot a bunch of blue screen foreground with the actors at our stage with a lot of flashing lights on them to simulate going through the woods. I had one of the best camera operators here, Mike McAllister, shoot a lot of bike stuff while driving through Lucas Valley Road on a motorcycle with a camera mounted onto the front and side to see what the lighting changes would actually be on him. It was undercranked to represent 120 miles an hour. It wasn't nearly as interesting as it should have been, so we dramatized it and made it a lot more violent looking. We shot for a week or two and covered the sequence from a number of angles. George was there and Marquand was directing the actors. They had a lot of footage to work with and they assembled a sequence that way."
The illusion of speed created difficulties in maintaining an effect of scale, according to Muren. "Some of the redwoods were six feet thick, but when you're going past at 120 miles an hour it doesn't look like a six-foot redwood. The whole scale seems reduced, as if you're looking at a miniature, because you never move that quickly and if you don't have any people or objects to relate to it loses the scale. This stuff almost has a graphic art look because of the speed and the streaking. Sometimes you get a different sense of speed depending on how high above the ground you are. If a scene needed to be a little more frightening, for example, we'd use a lower camera field."
Miniature bikes and riders were utilized for scenes that couldn't be done with live action, Muren explained. "We had them racing away from the camera tremendous distances and coming toward the camera from a long way off. A go-motion man manipulated the figures for some of this, moving the figures or heads."
Rotoscoping—the making of hand-drawn traveling mattes a frame at a time—figured importantly in the bike sequence. "We're relying a lot more on the roto department than on any of the other shows," Muren revealed. "It was a decision made at the start because it cuts down a lot of shooting time. If it takes three hours to get a shot it might take a half hour to get the glare off things, the glints off the lights and so on. If you eliminate that you can cut down a lot of time and shoot faster on the stage. We're limited by the number of cameras, the amount of space and the number of experienced crew members we have. The rotoscope artists can fix these things in a matter of minutes."
Some of the composite photography is unusually difficult because it occurs against light backgrounds which emphasize matte bleeds. "There are some really tough shots in the skiff sequence at the opening of the show involving blue screen work against desert skies," Muren said. "That sunlight caused headaches and took extra time for getting all the bugs worked out. Fortunately, a lot of the daylight stuff is in the forest, which has highlights and shadows and objects that are pretty neat camouflage for a lot of problems—plus it looks interesting. They go through that forest so fast it's like a light show, a kaleidoscope of images moving by."
The walking machine sequence uses a lot of go-motion effects," Muren said of the third major sequence for which he was responsible. "It uses mostly the two-legged walkers, or 'chicken walkers,' rather than the four-legged ones we featured in Empire." Go-motion technique puts a blur of motion into the animation, eliminating the stroboscoping effect that has plagued model animators since the earliest attempts some 85 years ago. "We used it in Dragonslayer, then in E.T. we had to go beyond that to matte the kids on their bikes in front of a bright sky," Muren explained.
"It took a lot of time to set the motors up and set the blue pylons to get rid of the rods holding the motors that were attached to the figures' heads, so I spent a lot of time looking for another way of doing it. Mike Foreman, who's a model maker, was one of the first to suggest that we didn't need to do some of this stuff with motors. He figured that since we shoot at a continuous speed, about one frame per second, on a lot of this stuff, that if it was something where we didn't need to stop and do any supplementary stop-motion animation, a person working with wires could do it. It worked, and so a lot of the go-motion shots are done without stop-motion and use fewer motors than we used before.
"On some of the walker footage, guns are firing and the heads are turning to specific counts, so we did all that with our traditional go-motion one frame at a time, with the major axis programmed and moved by motors, and then Phil Tippett or Tom St. Amand would do supplemental animation between each frame of the detailed stuff. It speeded things up so that we did most of the walker scenes—about 50 shots—in about two months.
"I didn't use much motion control on this. George likes to have a lot of footage when he cuts the films, and on a show like this where there's a lot of money at stake the last thing the producer or director wants to hear is, 'Wait a minute—we want to play back the tape to see if the move is right.' So I chose to put the energy into being free on the shots rather than being locked off and dictating how the shots would be. Then, if the director wants to pan the camera, we can pan the camera, but we'll figure it out later, as a post-production thing in the quiet environment back at ILM. There's lots of ways to do it without having pre-recorded in the field. We have a few shots with camera pans that were made here by projecting a plate up through the camera, then using the field recorder to check a grid point off the plate to pan the camera from. Once it was entered in the field recorder we played it back with the puppets. If the person who plotted it picked the right points to grid off of, then we'd have a pretty good match. Plus, we did a lot where we didn't try to lock in things too carefully. We had them walking against the direction of the pan instead of with it, so if something wasn't quite right it wouldn't show up. We got the show done faster and nobody cares if a walker is going off to right or left. We had a tremendous number of shots on this show— about 500 or so—and some of those Ken's been doing have about 50 elements in them. With such an enormous quantity of work, we were trying all the way down the line to come up with ways to avoid problems and still have it look fresh.
"We took a lot of chances on this show. When we did the walkers in Empire, I made a blue screen test, and matting those things walking slowly against the bald skies showed up a lot of matte lines. It was too touchy, so we instead did all of it in the camera with glass paintings and backgrounds and blue screens as well. In Jedi, because it was happening in the redwoods, I figured that the image is so jumbled with highlights and shadows and colors that matte lines, even if they were there, wouldn't show up. So we just blue screened them in, and that's how we got a shot a day even though it took a lot of time to set up those shots. By the time the artists have rendered a painting and we're tweaking the lighting and getting a set built and then we have to animate it, a lot of time has gone by."
Muren feels that practical considerations sometimes inhibit creative methodology. He explained that "to go berserk and just do something for the sake of being different can lead into a blind alley—and there is a responsibility to schedules and costs. So we're battling two factions. One is that we know we can do something a certain way—that's a given. But is it really the best way to do it? Maybe we can try some other way, but do we really know it is going to work?"
