Saturday, February 16, 2013

Steadicam Work in Return of the Jedi...

I read this earlier today and thought I'd post it here, courtesy of American Cinematographer Magazine, talking about the making of the steadicam background plates used in the speeder bike chase in Return of the Jedi, as told through the word of Garrett Brown, the camera operator, but first, some concept art by Ralph MzQuarrie:



"Just 100 more yards. My left hand is slippery with sweat. Legs tired. Brain hurts most of all. I wonder if really intense concentration produces any lingering harmful effects...look at Bobby Fisher! (At a camera speed of only 3/4 frame per second, a thousand feet of precise walking gets you just 16 feet of VistaVision in the can, and every take you forget a few more important phone numbers, or a family birthday.) Is it worth it? At this rate I will be illiterate in a week and a vegetable in two.

75 yards to go. The physical part should be easy. I've made hundreds of walking shots as long as this. The body appears to be merely strolling through the woods, carrying what looks like a portable X-ray machine. The mind, however, feels like Franz Klammer's on the downhill... One must rapidly cycle the attention between five discrete matters: navigating along the path, framing a distant target exactly on the cross hairs of the monitor, keeping the camera level, maintaining the correct camera height, and not straying side-to-side. The latter two tasks are only made possible by flying the camera directly alongside the thread and above the line!... (I'll pass on "above-the-line-jokes... more later about thread and line.)


My companions, Dennis Muren (recent Oscar winner for visual effects on "E.T.") and ace assistant-cameraman Michael Owens, are attempting to help me stay on course, each with his own deeply meaningful monologue, i.e. "Higher!, lower! perfect, no!...too much, good" etc. (I must remember that Dennis' voice refers to height, and Michael's to matters lateral.) Perhaps frail mankind was not meant by the Creator to imitate Prussian blue-steel motion-control machines. Her punishment for my impertinence seems to me quite adequate— the repeated electric shocks in the old forebrain as I cycle around my mental checkpoints, only to find each going ruinously astray: 1. "Step over the log, 2. Target! you're losing it!, 3. Level!! (How long has it been off, is it correctable?) 4. Whoa, the camera is a foot below the string, 5. Stay on the line!" etc. etc. All while lugging a VistaVision camera and two gyro stabilizers on the new Model III Steadicam. I think my EKG at the moment must look like a seismograph reading in Java.

A good take is of course proportionately exhilarating. As I am carried back to the start mark in a pillowcase, I reflect on the fact that this is quite a phenomenon a first. Lucasfilm, through its Industrial Light and Magic facility, has gambled over a week of shooting and a lot of engineering—dough that the Steadicam can deliver acceptable VistaVision background plates for the straight-ahead and straight-back views, over which will be matted one of the most exciting action sequences in The Return of the Jedi. In combination with Mike McAllister's side-looking camera car shots, and short dolly moves for "passby's," the scene will be constructed to run about two and one half minutes, with perhaps one hundred cuts. Mark Hamill, as Luke, and Carrie Fisher, as Leia, are pursued on their flying Speeder Bikes through the forests of the planet Endor by a horde of similarly mounted Imperial Storm Troopers. At 120 miles per hour! Wow. The effect on screen will be sensational. Speeder Bikes are a cross between motorcycle and rocket. The chase swoops at incredible speed through narrow openings between trees, with the battling participants and the lens clearing obstacles by inches.

In the beginning, during the planning stages, there was a certain amount of head-scratching on the part of the producers as to how the Speeder Bike chase—should best be executed. Dennis Muren, in charge of visual effects for the film, originally considered acquiring these shots by means of miniature stop-motion photography. A small model was built in order to determine the minimum acceptable scale, however Dennis concluded that the final "forest" model would have to be over 100 feet long, with a sky backing 100 feet wide at the end.

Then, on a trip to the redwood forest location for the live action part of the sequence, Muren was struck by the fact that the only really difficult plates to acquire by live shooting would be of course the straight-ahead shots and straight-behind shots. ILM produced a budget for getting perhaps one shot a day by laying a "hell-of-a-lot" of Elemack track, straight and curved, plus disguising the rails with ferns which would then be swept aside as the dolly progressed. These runs were contemplated to be about 300 feet long each, and Dennis figured that about five different setups would suffice, given the options to flip or reverse the film. Still, one can walk through a redwood forest. Muren even attempted a hand-held stop motion test with an undercranked Bolex, which, he reports, was terrible to behold. In the meantime, it was considered that a camera might be mounted to a motorcycle which could race at high speed over a road or path disguised with undergrowth, but this idea was ultimately dropped.


One consideration which might have ruled out the Steadicam from the beginning, was the weight of the VistaVision cameras. However, the possibility of lightening one of the small "butterfly" versions sufficiently was discussed, and finally Dennis suggested to George Lucas that since it is "easy" to walk through the woods, the possibility of using extremely undercranked Steadicam footage should be investigated. George said, "maybe, why not" etc. and I was called soon after and consulted as to whether I thought it was possible to produce results steady enough for matting. My initial, private reaction was the usual fear-of-failure etc., but honor and greed dictated a cautiously optimistic response to Warren Franklin, sequence producer, and the suggestion arouse that perhaps we could test the concept inexpensively by shooting spherical 35mm with my IIc Arri running at 3/4 frame per second.

Dennis Muren pointed out that he wasn't necessarily after the usual perfectly smooth tracking shots, as seen in the space sequences for Empire. He had been half looking for a new, less rigid, more realistic style; and he indicated that a small amount of roughness would perhaps be desirable in a chase near the ground, which might then feel as though it was shot from a pursuing aircraft. This was encouraging. I had once before, on the ill-fated Heretic, produced a 38-frame hold of sufficient stillness for an effect matted over the open mouth of James Earle Jones even after a violent 100 yard running shot. I began to think that it could be done. One of the early Steadicam prototypes actually incorporated a Kenyon gyro stabilizer, which is an amazing device, but which was ultimately dropped from my plans due to noise and power problems and because it tended to resist rapid panning moves. In this case, however, it could be used to artificially increase the inertia of the system in at least two axis, and would perhaps make the difference between success and failure. I suggested that we acquire at least one Kenyon gyro and attach it to the spar of the Steadicam.

April 12, 1982 found us chugging along the corridors of ILM, gyro humming, with my camera being run by a little outside motor at roughly 3/4 fps, and with 100 feet aboard of a test stock that could be processed and viewed immediately right in the building. The results were instantly encouraging, and by the end of the following day, we were testing large-scale shots within a local redwood forest, having worked out nearly all of the curious requirements for producing acceptable plates with the Steadicam.

All that is necessary for perfectly smooth results on this or any other Steadicam assignment, is the elimination of all of the six kinds of motion that plague say, the hand-held camera. We must avoid any angular deviations in pan, tilt and roll; and any unwarranted moves in the spatial planes of up-and-down, side-to-side and back-and-forth (here meaning variations in walking speed). Nothing to it!—Of course, this would be the all time mother of a Steadicam shot and virtually no errors would be tolerable. I am happy to say that the three of us—G.B., Dennis and Michael Owens, managed to blast through the problem and come up with schemes for each of these worries which eventually produced the usable footage in the final sequence.

To deal with the angular deviations, we mounted a side-finder video camera with a long lens, figuring that if I could acquire a distant target, such as a sunlit leaf or a piece of hand-kerchief tied to a branch, and hold a telephoto image of it on the cross hairs for the entire walk, then pan and tilt accuracy would be good enough for the VistaVision negative, especially considering that the big camera was carrying a wide-angle lens which would forgive small errors visible on the telephoto video. the problem of "roll," which is equally troublesome no matter what the focal length of the lens, was to be dealt with by continuously checking a very precise bubble level mounted next to the monitor.

To handle the spatial motions, we came up with what we modestly agree was the brilliant expedient of stretching a taut thread through the woods beside the chosen course, and making a straight dotted line of chalk directly along the ground, both of which would be invisible to the camera at speed, and which would positively guide us to the correct camera height and the straight and true path. Bear in mind that even a slow deviation in any of these directions occurring over the course of many yards, would produce a violent bump in the shot when seen at projection speed. I suppose we planned to have someone count cadence to keep my speed constant, but this turned out to be no problem and in the heat of the moment, we gave the cadence-counter the elbow.

The weight of the VistaVision camera was still a worry. I had learned that the prototype of the newest model of the Steadicam, the so-called Universal III, would be available in time for the shoot. This would be its first use on a feature, which was timely, since the III is lighter than previous models and much more flexible as to monitor positions and the distribution of its components. In addition, it provides for the display of level and adjustable frame-lines directly on the monitor screen. I hoped that the electronic level indicator would help with our critical roll axis, since it is more sensitive than bubble levels and easier to see. However Cinema Products was unable to finish this gadget in time for the shoot, and I was forced to rely on the old bubble throughout. Gene Whiteman, ILM's talented master of the machine shop, was given our careful estimate of the maximum weight allowable for the butterfly camera, (with about eight pounds of gyro and video gear already subtracted from the figure). We had found during the test that the gyro made enough of a difference to be worth carrying along. In fact, we ordered a second unit from Kenyon, so that when they were mounted at a 90 degree angle from each other, we would have double the effect in the critical pan axis and single in tilt and roll. If we ever do this again, we'll probably get a third unit and festoon the thing with gyros. (With any more than three, I might even be able to drop out of the harness and sneak away unnoticed. By the way, they take about five minutes to come up to speed, and at least ten minutes to stop, so don't try to drive away from the location right after the wrap. Your car won't be able to turn left or right for the first couple of miles!)

During the next two months, I kept in touch with Michael Owens regarding the weight problem, and his progress with a new wrinkle which had come up in the meantime. Dennis had decided that we must also have a remote-controlled "roll cage" on the camera, so that he could follow along and dial in smooth banking moves as I made the turns, to be used for certain POVs required in the chase. This however meant more weight and the need for the camera to stay exactly balanced as it pivoted around the lens axis. Whiteman and company did a fantastic job, since, as of the first shooting day, the camera/gyro/electronics/roll-cage/video-finder/motors/film/lens etc. totaled no more than our maximum limit of 26 pounds!

On June 14th, I flew to San Francisco, stopping at ILM to check the last minute arrangements, and on to Humboldt Redwoods State Park, to commence shooting in the "Avenue of the Giants." The crew had come prepared with lightweight aluminum ramps which could be placed over the odd ravine and disguised with tons of potted plants, rubber leaf-mats, ferns, etc., which they also brought, along with tools for clearing a navigable route, reels of green thread, green wands to suspend same, and squirt cans of chalk to indicate the True Path. Dennis had previously selected the courses he wanted to use, and the crew had them roughed-in by the morning of our first shot.

For the next three days, we made about five or six "official" shots per day. Michael Owens and I winged a number of other shots, just walking through the trees, burning all of 45 frames of raw stock per minute, but I must say that most of that stuff was merely exciting rubbish. With perhaps one exception, only the prepared shots ended up in the sequence. Even with Michael holding back and boresighting a line for the camera, and shouting instructions to me to raise it, lower it, move right, etc.,; these attempts lacked the last few percent of precision necessary to be usable. We made several refinements in the technique for the regular shots. It turns out that you can't just stroll along squirting the chalk line for the path without the risk of it wandering from side to side, even if it looks true to the eye. We actually had to stretch a string along the ground, to get a straight path, and then go over it with the chalk. In two visits to the big trees, we used up many pounds of chalk and a mile of thread and string. I admit that as statistics go in the movie business, these are less impressive than say, the quantity of cement in the chariot race set on Ben Hur, but they're all we've got. Obviously we won't impress anyone with the amount of raw stock expended.

I gradually acquired the knack of shooting curves as well as straight-ahead shots. It's hard to make thread stretch in a curve, so I had to get used to less help from that quarter. In addition, as soon as I started a turn (ever so slowly), I had to give up my video target, and let the lens drift across the terrain until a suitable new target swam into telephoto view along the next straight leg of the course. We even tried a scheme in which I would ease the target onto the trailing frame edge of the video camera, and then zoom slowly out, keeping the target thereon, and then zooming quickly in when a new target arrived at center screen, and so carrying on, I must report however, for purists still following the technicalities, that as I got fully wide, the angular resolution diminished to the point that my shooting was no longer as precise. We made a number of these long curved shots, and the best ones were done by just letting the landscape drift at a constant rate, until the straight bits came along.

The roll-cage got a workout during these curving shots, and we made at least one take with banked turns for each pass which was kept level. It proved extremely difficult for Dennis to judge the correct slow rate of banking with his remote control, and when you consider that his moves are to be seen at 30 times the shooting speed, I think he did amazingly well. Certainly his years of laying out motion-control timings helped in this regard.

On the first location I was in some pain from a minor bit of surgery on my left foot, but dealing with all of these complications was so absorbing, that I literally felt nothing until the end of each take. It was the only time in my career during which the walk back to the start was worse than the event itself. Owens is terrific, schlepping the rig back for me each time to cut down on the wear and tear. In addition, he had to practically dismantle the camera to reload the special daylight-hundred-foot spools, plus keep track of the footage expended by means of a stopwatch, since we had no frame or footage counter onboard not to mention calculating exposure at 3/4 frame. The forest was so dark much of the time, that we were exposing at T.4 notwithstanding an effective ASA of 2560!

I enjoyed watching these two "indoor" guys deal with the realities of "outdoor" shooting. Dennis and Michael are accustomed to the timeless air-conditioned confines of the stages of ILM. Shooting in the real world provided them with a number of surprises. We could not locate the thermostat for regulating the temperature of the forest, nor the dimmers for adjusting the amount of light, which itself seemed to constantly shift direction from morning to night. Dennis was also annoyed by hordes of the local "Mini-flyers of Endor" strange biting small craft which stubbornly refused to attack him at any speed other than 24 frames per second. (Since I was undercranking, I was of course immune!)

The final shoot commenced on September 20, and was designed to give us a chance to get any remaining shots which were needed for the cut version, and to re-do any that didn't work the first time. One of these latter involved a rapid move under a fallen log which was useless because the original course was too short. As I recall, we remade this one, but scrapped another experiment that actually involved a live person, our amiable and talented second assistant Randy Johnson, who was in costume to double Luke Skywalker, brandishing a sword at 1/30th speed as we roared down at him at a simulated 90mph. (He eerily resembles Mark Hamill, so the scheme should have worked, except that once again it is difficult to judge moves appropriate to this degree of time compression. I am told that in dailies it looked like we were running down a hyperactive puppet grandmother.)

This second adventure in the forest gave us an opportunity to experiment with the fearsome "sideways" version, in which the Steadicam is allowed to float horizontally, and the camera is re-erected by turning the roll-cage 90 degrees. I made this ill-advised suggestion in response to a request from Dennis Muren for some shots that would appear to fly directly over the ragged undergrowth. It worked beautifully but proved to be difficult to tune, since the camera was aimed off to one side, and somewhat above vertical, and therefore any balance adjustment to the straight-ahead-looking Steadicam caused the camera to assume a different angle. The new model III proved its worth by permitting a much larger number of adjustable components to actually adjust!

There you have it! A tour de force, if not a tour detour. From the standpoint of cost, it was clearly effective. Dennis informs me that the miniature model scheme would have gone at least five times as much, and even the restricted dolly version would have cost 50 percent over our final budget, despite the large sums spent upon lightening the VistaVision camera. It has certainly been a very gratifying assignment for me. I don't think I have permanently harmed my brain, and having just passed my 40th birthday at the time, I was happy to find that I could still get the old legs up over 100 mph. I look forward to the opportunity to shoot even faster plates, perhaps breaking the sound barrier. As soon as I can remember my address, I will collect some clean socks and I'll be ready."