This story begins over fifty years ago. I was recently out of the army and I'd been making a living writing for photo magazines and designing gadgets to be used in making TV commercials. I had an idea for building a computer-controlled zoom lens that would be used in shooting movies and commercials. I had no engineering background, but I convinced a small group to finance the idea. Two years later, I was nominated for an Academy Award, and the next year my lens was used to shoot the opening scene of The Godfather.
Along the way, I got married, someone gave us a honeymoon in Paris, I got to shake hands with Prince Phillip, appeared on the evening news in London, and finally submitted it to The Academy of Motion Picture Arts and Sciences for an award from the Sci/Tech group.
But I think that the most interesting part of the story is how many puzzles I had to solve that were clearly impossible. As I said, I don't have a formal background in engineering and in many ways, I think that worked in my favor. Someone with the right training would have understood that these puzzles couldn't be solved. But lacking that training, I wasn't aware that things were impossible, so I went ahead and did them.
But first, let's take a look at what Francis Ford Coppola and Gordon Willis did with my computer-controlled zoom lens. This is the opening scene from the Godfather, just voted the greatest American movie by a group of industry professionals.
In this opening scene, an undertaker petitions Don Corleone in his quest for justice. At the beginning, the undertaker's face fills most of the frame. As the scene progresses, we zoom back very slowly and you can see the subject getting smaller and smaller. At the end of the zoom, we can see the full scene, with the undertaker framed by Don Corleone, seen from behind. This scene, which takes almost three minutes, would have been impossible with other zoom controls.
This is how an article in the March 1999 issue of American Cinematographer magazine described the photography that cinematographer Gordon Willis used in making The Godfather:
To further simulate the film's 1940s milieu, he and Coppola abandoned all modern filmmaking tools, such as zoom lenses, opting instead to frame their tale in a "tableau style" of painterly, classically composed frames (the one exception to this rule was the famous opening pullback shot).
From Time Magazine, March 14,2012
That Opening ShotUsing a “high technology” computerized zoom lens, Coppola started with a tight shot of the undertaker’s face, and then pulled back slowly for 2 min. 20 sec., before holding the shot for another 30 sec. while the undertaker whispers in the don’s ear.
This is what the Angenieux 25 - 250mm zoom lens looks like in its unmotorized form. It's quite large and quite heavy, weighing over ten pounds. The two rods are attached to the camera and run underneath the lens to support its weight. This is the bare lens with no zoom control attached.
Here are some of the design and philosophy issues that were involved in the design and building of the computer-controlled zoom lens used to shoot the opening scene of The Godfather.
Where no man has gone before -
It would have been nice if someone else had already built a computer-controlled zoom lens. Then I would have had some guidance, and could have just worked on building a better model. But no one had. I would be the first. In fact, my zoom control was probably the first piece of computer-controlled equipment ever built for motion picture use. So, Like Captain Kirk, I ventured forth.
There were some controls for zoom lenses available at the time, but they were very simple. A typical one was fitted to a small handheld box with a little lever that did the actual controlling. Push the lever a little bit and the lens zoomed forward slowly. Push the lever a little bit further and the lens zoomed faster. Push the lever in the opposite direction and the lens zoomed in the opposite direction. That was it.
So it's no surprise that anyone who could produce exactly the zoom that the director or the cinematographer wanted was highly sought after. Zoom lens operators might be booked several pictures in advance. They were considered as artists for this ability. But even with all of this skill, the best that could be hoped for was just an approximation of the desired zoom. The use of the zoom lens in motion pictures was limited because of these problems.
It's one thing to do a zoom, but it's something else entirely to be able to repeat exactly the same zoom again and again, take after take, until the director is satisfied with the result. The characteristics of an ideal zoom would be to hit the start and end points accurately, and to make sure the zoom would take exactly the right number of seconds, every time.
With the old zoom controls, you would fiddle with the little lever until the starting point for the zoom looked pretty good, then mark that on a piece of tape around the lens. Then repeat the process for the framing at the end of the zoom, and make another mark on the tape. Then fiddle countless times until you got the right time for the zoom between the two points and make a mark on the zoom control showing how far to push the little lever. It was a long and complicated process to set up any but the simplest zoom.
Room at the top -
This was always a sore point with both the director and the cinematographer. They wanted to be able to set up the zooms themselves. The framing at the beginning and the end of the zoom is critical, as is the time it takes to get between the two points. You would rehearse the actors multiple times and then measure the time with a stopwatch. This would tell you how long the zoom should take if it was going to match the speech.
Wouldn't it be nice if the director or cinematographer could just look through the camera, set the start and end points for the zoom, and then, somehow, set the time that the zoom should take? Wouldn't it be nice if the zoom could be repeated any number of times, producing exactly the same zoom at the touch of a button?
For this, you needed a computer, and this is where I was headed.
Infinite speed range -
It would be nice if the zoom control could do a very fast zoom -- say half a second from end to end. And it would be nice if the same control could give a zoom that went on for minutes, as in the opening scene of The Godfather.
This is a problem I didn't even think about when I started. The existing zoom controls had a very limited speed range. You might have a range, for a full zoom, from several seconds to about thirty seconds. Zoom lenses required a lot of torque to turn them, and this required small, powerful motors. But motors of this type had a limited range of speeds.
I would be using a similar motor in the zoom control I was building, and I would have to find a way around this limitation.
Soft start and stop -
This is a little something extra. Instead of an abrupt start and end to the zoom, it would be nice to have a soft start and stop. If done properly, the start would be imperceptible, and the end of the zoom would be equally smooth. There should be an adjustment for this softness, from none to maximum.
In 1969, computers that could do the sort of real-time processing that would be needed to control a zoom lens were huge, drew thousands of watts, and needed very complex assembly language programming to do the simplest task. In addition, there was no way to hook one to a zoom lens, or to easily input the setup for doing a zoom.
To give some idea of how primitive the state of computing was, your smartphone has more processing power, memory, and storage then all of the computers in the United States had in 1969. Obviously, I would be headed in another direction.
In the end, I designed a computer to control the zoom lens. It weighed less than two ounces, measured less than 3 x 4 inches, and drew only a few thousandths of a watt of power. Like something out of science fiction.
Motor control -
The motor control would have to be solid state, using minimum power for the longest battery life, and be capable of driving the zoom lens over an infinite speed range.
Portability - Easy to carry and easy to set up -
The control unit that would program the computer would be handheld and should weigh about a pound. The power unit with the batteries and the motor control circuits should weigh less than ten pounds. On the movie set, the system should be easy to set up and easy to break down at the end of shooting. In addition, there would be cables hooking the system together. There should be no way to put things together incorrectly.
The lens with its attached motor, the computer unit, the power unit, and all cables should fit into a padded, easy-to-carry container that would be as rugged as any piece of movie equipment.
Here, in the twenty-first century, computer interfaces are very advanced, with touchscreens, mice, and computers that respond to human speech. When I built the zoom control, there were no computer interfaces, other than a primitive display terminal or a Teletype terminal. The interface for the zoom control would have to be easy to use and shouldn't require any special training. It would have to be intuitive, so a director or cinematographer could easily program even a complex zoom.
Quite a requirements list, and there would be many more hurdles along the way.
Grokking the system -
There is an ancient story of the butcher whose knife never got dull. As he cut, he sensed how to move the blade so that it would cut through the meat, avoiding the bones and cutting without resistance. In the Zen aspect, this happens unconsciously, driven by a deeper, inner understanding.
This is how you have to think if you are going to solve impossible, intractable problems -- you have to "grok" the system you are working on, until you become as one with it. You have to study the problem, understand it, then let go and allow your subconscious to seek the solution. Designing and building the zoom lens was an exercise in the Zen of design.
Designing the computer -
Let's start at the beginning. In moving the zoom from point A to point B, there are two things to consider. You can control the speed the lens is moving, or you can control its position. The zoom control would have to do both. You have to be able to set the start and end points and how long it will take to get between them.
While it is relatively easy to control the position of the lens, measuring its speed accurately, especially at very low speeds, is quite difficult. Suppose there was a way to control the speed of the lens without actually measuring its speed? Sounds very Zen.
As a start, let's say that you move an object at a constant speed. When you do this, its position also changes at a constant rate. If you draw a graph of this, it's a straight line, showing the position at each point. If we turn the problem around and look at it from 90 degrees, you can control the speed of the lens by just moving it to the correct position as it changes with time.
Now we can throw out the requirement for measuring the speed and replace it with measuring the position, which we needed in any case. All we have to do is calculate where the lens should be at any point in time.
This breakthrough greatly simplified the problem and increased its do-ability. It's an example of how to solve an impossible problem by deepening your understanding to the point where you can restate the puzzle in a solvable form.
Now all I needed was a computer that would generate the information needed to control the position of the lens
Here's the actual computer from the zoom lens control. It's an analog computer, with twelve operational amplifiers, precision resistors (.001% accuracy), and assorted capacitors, all mounted on a card measuring two and a half by four inches and weighing less than two ounces. When used for its specific role, it will outperform a digital computer many times its size.
There are two basic types of computers. Digital computers are what's used today. They're general purpose, equally adept at calculating inventory or handling airlines reservations. As mentioned earlier, the digital computers in 1968 were giant, power-hungry beasts, difficult to program and difficult to interface with things like a zoom lens. But maybe there was another way.
The other type of computer is called an analog computer, very popular in the early days of computing, but little known today. The main use of this type of computer is to generate mathematical functions and curves that express mathematical formulas. The earliest analog computers used gears and other mechanical elements. Later models used sophisticated electronic components. While digital computers expressed their output in discrete steps, the analog computer produces its output as a smooth, continuous line or curve with no breaks or jumps.
Another breakthrough. Another piece of the puzzle falls into place.
These are the two equations solved by the zoom lens computer. The first equation uses the inputs from the computer interface to calculate the speed that the lens should move during the zoom. The second equation calculates where the lens should be at any point during the zoom.
Just what I was looking for to calculate the path for the zoom lens to follow. What the zoom lens computer would compute, in terms of calculus, was "the integral of the speed of the lens, between the start and stop point, with respect to time."
In other words, given the start and end points, and told how long it should take to get between them, the computer would figure out the correct position for the zoom lens at every point in time during the length of the zoom. Triggering the computer would begin the zoom, directing its path at every point along the way, all the way to the end. All that remained was to design and build the computer.
Easier said than done. While generating the functions required by the zoom lens were theoretically possible, it took months of design, experimenting, and tweaking to get the precision that this task required. The final result is the miniature computer shown above.
This is the motor control circuit board. Power transistors, more operational amplifiers, and assorted components combine to give an ordinary DC servomotor an infinite speed range.
Controlling the motor -
The next insoluble problem was how to give the motor that would turn the zoom lens an infinite speed range. This would be necessary if a director wanted to do a zoom that lasted for minutes. Motors that had enough torque to drive the lens had a very limited speed range, and that's what you found in the zoom controls that were available at the time. Unfortunately, I needed to use the same sort of motor in order to get enough torque to drive the lens.
It was quite a puzzle, so I did a lot of reading and watched the world around me for something that would yield a clue. The first clue that I got was from an obscure manuscript on servomechanisms, buried in a discussion of control systems for radio telescopes. The next clue came to me as I was watching how the machinist, building some of the parts for the system, accurately positioned a vise on the table of his drill press.
Using these clues, it took some Zen-like thinking to turn them into a real-world solution. In the end, it involved adding an unconventional bit of circuitry to a standard motor control system. I used an oscilloscope to tune the system until the desired result was achieved. (The exact solution remains a secret to this day.)
The final result was a control system for DC servomotors that had no low-speed limit. The longest and slowest zooms were now possible. This is what made it possible to shoot the opening scene of The Godfather.
This is what the Angenieux 25 - 250mm zoom lens looks like with its enclosure containing the drive motor and the position sensor that reports the current position of the lens. The window on the side of the enclosure displays the lens's current focal length.
Working with the lens -
Some modern zoom lenses come with a gear already mounted to the lens and that makes it easy to attach a motor to drive the zoom. The lens I was working with was meant to be turned by hand and that was it. So the first thing I needed was a way to put a gear on this lens so that I could couple the motor and the position sensor to it.
I searched the catalogs of precision gears and located a gear about half an inch bigger in diameter than the collar around the lens that I would attach it to. I had the machinist cut a hole through the center of the gear just slightly larger than the lens's collar. The gear was then split at one point and a screw went through at this point. The gear could be slipped over the lens's collar and the screw tightened to hold it firmly in place.
The motor and position sensor were both attached to this gear with gears of their own. The position sensor needed something called an anti-backlash gear that removed any play between the two gears for the highest accuracy.
Although the zoom collar on the lens turned easily, the grease in the lens had high viscosity, making it hard to turn at faster zooms. I took the lens to Marty Forscher of Professional Camera repair and he replaced the original grease with a thin silicone lubricant which allowed the lens to turn more easily. It had the additional benefit of maintaining this low torque requirement in cold weather that you would find in outdoor shooting.
This is the level of detail that you have to get into when you take a systems approach to design. Every part of the system has to be considered as a separate element and as part of the whole.
These are all of the pieces that make up the zoom lens system. The T-shaped control unit contains the computer and has the interface that programs the system. The large box contains the batteries, the motor control electronics, and the battery charger. The zoom lens is shown surrounded by the box that contains the motor, the drive gears, and the feedback elements. The small window on the side of the box has a printed tape that displays the current focal length. The small box with two knobs can be used instead of the control unit to manually control the zooming. Connecting cables are not shown.
Packaging the system -
The final zoom control had three main units.
First, there was the zoom lens itself, with its attached motor inside a soundproof enclosure.
Then, the motor control circuitry and the batteries were in a welded aluminum case occupying less than a cubic foot of space. The batteries could power the system for about 15 hours.
The most important piece was the handheld control unit that was used to program and control the operation of the zoom lens. This unit also held the analog computer described above. The control unit was built into a T-shaped handle that made it easy to grip. The main interface was a set of knobs and switches that made it easy to describe the required zoom. A set of four pushbuttons controlled the starting, stopping, pausing, and resetting of the zoom.
The three units were connected together with a set of cables, Each connector had a different number of pins, making it impossible to hook the system together incorrectly.
This is the case that contained the batteries that powered the system, the power supply circuitry, the battery charger circuitry, and the motor control circuitry. Note that the sockets for the cables that interconnect the system have a different number of pins so that things couldn't hooked together incorrectly.
The interface -
My motto is, "Build complex toys and simple tools." The zoom control would be useless if its interface was too complex and difficult to use. I needed an interface that could be used by anyone, after just a brief instruction. In addition the interface should be designed to mirror the way that the person using it understood their job. Its controls should exactly emulate the way that a director or cinematographer would think when setting up a zoom.
The best thing versus the right thing -
This is a philosophical issue that helps to guide understanding and choices. For instance, the "best" computer was obviously some giant digital system that would provide a more general solution to this or any other problem. It's a seductive trap that many designers fall into. The solutions listed above were, in part, made possible by rejecting this sort of thinking.
Revolution versus evolution -
Although there were breakthroughs that helped to set the path, the actual process, from beginning to end, was a set of tiny steps, at each point focused on improvements, tweaks, and tunings that would yield a small additional benefit. Every day, the system and its component parts were examined again as if for the first time, with an open mind, ready for some new insight. This examination never stopped during the years that it took from the very first steps to the finished design. And even today, I still think about the design and what sort of improvements remain to be made.
Once you understand this sort of Zen approach to solving problems, you will be like the butcher whose knife never got dull.
What the finished system looked like
The computer and its interface, all in one neat, handheld package. Seen from this angle, you can see its shape.
In 1970, this looked like something out of Star Trek.
A head-on view of the T-shaped handle containing the computer and its interface.
Interface and use -
This was one of the earliest examples of using a simple interface to hide the complexity and power of the computer system it controlled.
The main programming of the zoom was done with a bidirectional toggle switch and two knobs, as can be seen in the picture above. Looking through the lens, the director or cinematographer would first push the toggle switch to the left and use the left-hand knob to set the framing for the start of the zoom. Then he would push the toggle switch to the right and use the right-hand knob to set the framing for the end of the zoom. By doing this while looking at the camera's groundglass, the framing for the start and stop could be set quickly and precisely.
Above the knobs was a set of numeric thumbwheel switches. Using these, it was a simple matter to dial in the exact time that the zoom would take. Another knob dialed in the softness of the start and end of the zoom.
In just a few seconds, you could set up a precise zoom, framing the start and stop, and setting the exact time the zoom would take.
There were four square pushbuttons, vertically aligned on the handle. These controlled the running of the zoom. The Start button started the zoom. The Stop button cancelled the zoom. The Pause button paused the zoom, in case an actor hadn't hit their mark yet. Pressing the Pause button again would continue the zoom. The Reset button set the zoom up for another take.
And that was it. A piece of computer equipment that you could learn to use in just a few minutes, thanks to its intuitive interface.
Discovery Technology -
I had to think up a name for the corporation. You have to register the name with the state corporation commission and the name has to be unique. After a number of tries, I came up with the name Discovery Technology -- an interesting name and also a pun. In those days a business also had a cable name for receiving overseas cables, usually a shorter version of the corporation's name. Our cable address was "Disco-Tech."
TV commercials -
This was the first thing that the zoom control was used for. Commercials took advantage of two of the control's strong points -- precision and repeatability. Both are crucial requirements in shooting TV commercials, where a 30-second commercial has to take exactly 30 seconds. The zoom control allowed the commercial's director to quickly set up the required zooms and fit them to the exact timing dictated by ad agency's storyboard.
One commercial that took advantage of both the slow zooming capability and the precise repeatability of the zoom control was for the gubernatorial campaign of Nelson Rockefeller. As part of his campaign, he had proposed a new set of drug laws that would put drug dealers in jail for life. How to show this in a 30-second TV commercial? The photographer build a set in his studio that looked like a jail cell. The drug dealer (actually the account's art director) was posed carefully on this set and the zoom was set for a pullback that lasted almost the full 30 seconds.
The first shot, with zoom, was of the drug dealer at his current age. Then the subject held that pose while makeup was added to age him a little. Then another zoom. This was repeated seven times, with the makeup getting more extreme at each take. In the editing of the commercial, sections of the different takes were spliced together with the result that the subject aged fifty years during the 30-second pullback zoom. This would have been impossible to shoot without a computerized zoom control.
We'll always have Paris
- In April of 1971, my fiancée and I took the zoom lens to Hollywood to exhibit it in a trade show of motion picture equipment. We had made arrangements to be married there as well -- an elopement in true Hollywood fashion. At the show we met a charming Frenchman named Claude Chevereau. He said that he'd heard that we'd just gotten married and asked if I would like a honeymoon in Paris in June. I thought he was kidding and said yes, at which point he handed me two airline tickets.
So we rushed to update our passports and in the second week of June we headed for Paris. Claude met us at the airport and helped to smuggle the zoom lens into France (we would be leaving it in France for rentals there). A few days later, it was my birthday and we celebrated with Claude, his wife Evelyne, and some new friends at a Hungarian restaurant until the wee hours of the morning.
My French was primitive, but somehow I was able to work with the French technicians in attaching the zoom control to a lens that belonged to Claude. In Paris, Claude and Evelyne sponsored a salon for us to demonstrate the zoom lens to French cinematographers and directors. Afterwards, we went out for drinks with the French crew. They taught us some French and I taught them some American expressions. My favorite was "mazel tov," which I explained means the same as "bonne chance." I had visions of these French guys, in their future interactions with Americans, saying "mazel tov," and what sort of reaction they might get.
Here we are at the Film 71 show in London. This shot is from the August 1971 issue of American Cinematographer magazine. The Angenieux 25-250mm zoom lens is mounted on an Arriflex 35mm motion picture camera. The rectangular box housing the motor and gears is mounted on the lens. The T-shaped control unit, with its knobs and switches, is seen in the foreground. The computer is actually inside this handle. Not shown is the box containing the motor control circuits and the rechargeable batteries that powered the system.
Off to London
- After Paris, we hopped a plane to London, where we exhibited the zoom control at the "Film 71" International Film Technology Conference and Exhibition that was held at the Royal Lancaster hotel. The picture above was taken at that exhibition.
The most interesting point of our London adventure was meeting Prince Phillip who, it turns out, visited most of the trade shows held in London. David Samuelson, owner of London's largest cine equipment rental house, brought Prince Phillip over to our exhibit and explained the equipment we were showing. I shook hands with the prince who asked if we were considering opening a subsidiary in London. I said yes. Anything is possible. Later, the shots of the two of us were on the evening news in London.
Claude Chevereau and David Samuelson asked if we wanted to bring our equipment down to Rome to demonstrate it for Federico Fellini, the famous Italian director ("La Dolce Vita"). I had to decline as we were almost out of cash, a decision I've always regretted.
- The zoom control was used in a number of movies. One of them was Doug Trumbull's science fiction classic "Silent Running." It was also used to make a bunch of movies in France.
- "Focus-pulling" is a highly regarded aspect of cinematography. It's one of the main jobs of the assistant cameraman on a movie set. There's a lot written about focus-pulling, and lots of tools have been created to support it. But zoom-pulling does not exist. There are no zoom pullers.
Zoom lenses aren't used to their full potential because there isn't a way to program them, or to make them move slowly. There is no repeatability so you can make multiple takes. After 60 years, the lack of tools has kept the zoom lens an unevolved cinematography tool.
The opening scene of the Godfather is a demo of what a zoom can do. Francis Ford Coppola and Gordon Willis originally didn't want zooms. But they ended up doing something brilliant. Using a zoom, they made what many consider the most important scene in the most important movie.
The first and the last
- My zoom control was the first piece of computerized equipment designed especially for motion picture use. It was the very first piece of computer technology used in the shooting of a major motion picture. It was the first to have an interface that could be programmed directly by the camera crew. And, very likely, the first battery-powered portable computer.
Today, the use of computers in making movies is common. In fact, some movies are made entirely in the computer without ever exposing a single frame of film.
Given that, you would think that someone, in the last fifty years or so, would have duplicated my zoom control in a more modern, or in an improved form. But no one has. If you use your favorite search engine, the only computerized zoom control you'll find is the one that I built, all of those years ago.
So I was the first, and I was the last. Maybe that's why, in the years since then, no one has shot another "Godfather."
In March, 2022, the computerized zoom lens was submitted to The Academy of Motion Picture Arts and Sciences for a Scientific/Technical award. This is the fiftieth anniversary of The Godfather, so maybe this achievement will finally be recognized.
Article from Reuters about the fiftieth anniversary of
American Cinematographer Magazine
Copyright 1957-2023 Tony & Marilyn Karp