[mag cover pic]

From the January/February 1986 issue of Graffitti

Reaching for the Edge:
A Conversation with Con Pederson

by Philip Denslow

A chrome robotic female reclines next to a table above which float metamorphic vegetables and utensils, She languidly stretches and tells us that here on Jupiter in the year 3000, cans are still the preferred method of food storage. Her movements are uncannily sleek and seductive. This is a commercial for tin cans?

Con Pederson is the primary animator-designer-programmmer-special effects person at Robert Abel & Associates. The techniques used for this spot represent the newest development in the company's pursuit of high-tech commerce. Since the early 1970's the company has been introducing new imagery and technique to the highly competitive area of TV advertising, clients such as ABC, Whirlpool, Seven-Up, Levi's, TRW, and AT&T have been represented using innovative animation effects.

In this instance, however, even with the amazing abilities of the computer, the movements of the robot were actually generated with the aid of that time-honored animation technique, rotoscoping, A real person was filmed while making the desired motions, and that film was projected onto a computer monitor. Then a simple armature version of the robot-to-be was manipulated on the monitor using a program called "Direct" until the positions matched. In this way, keyframes were made and the computer completed the motions by generating inbetweens.

The motions were used to move the fully rendered robot. The rendering was done by giving the computer a mathematical description of lots of little facets, called polygons, that are joined together to make bigger shapes. This ,process of converting drawings into numbers is called digitizing. The computer is also told how to describe these surfaces in terms of color, light sources, and reflection and refraction of the light.

Con Pederson began writing science fiction at the age of fourteen, publishing several stories. He later attended UCLA where he studied art and archeology. Pederson made three animated films and wrote an authoritative paper on petroglyphs. Then he went to work at the story department at Disney, writing for their "Man on the Moon" and "Mars and Beyond" programs being produced for television. When that series ended, he continued with graduate school and got a job at a small but pioneering production company, Graphic Films.

At Graphic Films Pederson worked with fellow UCLA alumni Robert Abel, Ben Jackson, Robert Swarthe, and Colin Cantwell. Douglas Trumbull and Jim Dickson were also there. All these men would make their mark in the animation/special-effects industry. Graphic Films was involved in making films for NASA when Stanley Kubrick asked them to do development work on his fledgling project, "2001". Eventually Pederson and others left Graphic Films and moved to England to work on the film.

After returning to Los Angeles, Robert Abel asked Pederson to join him in the beginnings of his production company, Their first assignments were animated logos for ABC Televison and Whirlpool Corporation. The techniques developed for "2001" were then applied to commercial advertising. In the dozen or so years since, the technological changes incorporated into their work are remarkable.

Most of the images produced by Pederson and the others at Abel & Assoc. were achieved with a computerized camera stand. These motion-control techniques were very successful, but the company wanted to preview motions that couldn't be seen until the job was almost completed. They found a solution with the Evans and Sutherland flight simulator. This device could present a simple wireframe version of what was to be photographed later by the motion control camera. This test of the motion could be produced quickly, and shown to a client for approval.

Soon they discovered that by using colored filters and multiple exposures, the images produced by the E&S system could be used themselves. With additional improvements in rendering and controlling computer generated images and movement, they have achieved the ability to make things like the "sexy robot" come to life.

Con Pederson talked about the "sexy robot" commercial on January 18, 1984, two days before its debut during the Superbowl game.

PD: When you rotoscoped the model, were you looking at two views?

CP: In this case we had one view, We could have had another camera set up but we gambled on the fact that we had fixed points for the robot and the thing would then carry from one view to another. We could have used a second camera on certain shots and used the other angle to check what the robot was doing. On the other hand, we should be able to add something to it, not be stuck with it. But in this case what the model did, the consistency of her movements, helped. It was different than 2-D rotoscoping,since the computer was smoothing the motion from keyframes, keeping weight and mass consistent.

PD: I really liked the knee joints with the brass hinges at the back.

CP: Well obviously since she's bi-symmetrical, you digitize the left half of everything and switch it over, But she has all this rendered stuff all over her that has to be treated as separate objects. There are hundreds and hundreds. You look at a list of all the parts and all of those are separate files and each file consists of thousands of polygons.

It's not that we're trying to pursue reality, but you need the capacity to execute reality to have a safe base to go on. Because even if you're just doing a tin can, a common thing like that, you find that you can save a lot of money by doing it as a rendered object. Then you must treat it as a real object and do all this subtle stuff to it in terms of rendering it to make it look as real as possible. The reason is that en-route to doing the kind of art that exceeds reality, that does something better than reality, by that I mean you can exaggerate things and you can leave things out, it's nice to be able to do what reality does and then do it better, instead of not being able to do what you want because you haven't been able to achieve realism.

You can always justify doing something simple, if you want to. But once you have the power to do reality, then you have a choice. I think then you begin to appreciate simplicity as an addition to reality. Then you can start editing and leaving out things that you don't really need. Whether there are things you don't want or things that are in the way, just simplifying parts of it emphasizes some aesthetic criteria you wouldn't get with an ordinary shot. I get more fun out of doing something that's interesting and isn't real at all.

PD: What are you working on or would like to work on to make things more sophisticated?

CP: We can do things like ripples, which was in a TRW spot. That's a little program Roy Hall, one of our programmers, wrote. It's called Ripple. (laughs) It was written just for that spot. It didn't take long. In the same spot there was a little guy on a weather vane. A penny farthing bicycle rider with his legs pedaling around. That was our first use of the Direct program, a kind of debugging stage.

We don't have motion blur yet. It's difficult when you aren't doing photography to have the blur local to the object and not affect the background. The subtlety of graduated focus is what makes things that are three dimensional very effective. When you simulate focus in a simplistic way, the Disney multi-plane camera for instance, those things are never quite satisfactory because the flatness is truly flat instead of the graduation of depth, which is what you get in reality or with a model object.

One of the things we do here that interests me, when you compare our work to some other places, is that we put in a lot of little refinements: the type of motion we use, the subtlety of the motion, as well as the quality of the design. We push that. A lot of the others develop surfaces that are pretty simple, not necessarily messed-up enough.

There's a proliferation in all directions in capabilities of the software, If you want to see something, it's just a matter of saying: "Well, what is it that I want to see?" It's becoming easier and easier because there's more than one way of doing it. Every time you get the machine to do something, it gives you more freedom to do the next thing.

Especially in the early days of computer graphics, people discovered things by accident. They didn't know how they got what they saw. They didn't understand what. it was at first, but it was really interesting. Those are fun discoveries to keep making. It's like you want to play with things all the time just to cause an accident. Even if you can't recreate it, it gives you an idea.

PD: What do you think it would take to get to the point where you wouldn't have to rotoscope a real person in order to get lifelike images?

CP: Well, there are people who for a long time have been trying to make mathematical models for the way human bodies move. This is interesting for a lot of people. Personally I'm not very interested in that, but I know it's being done. Since we do commercial work - we're not an institution or research oriented - it has no value to me because it's easier to go out and shoot someone in real time.

There wasn't much animation in this spot really. We did some animation but we didn't have time to put it in there. It was just a matter of what we had time to build up in the system before we ran out of computation time.

PD: How about the animation of moving around the room?

CP: Well, that's just motion, speed and motion control.

PD: So you differentiate moving objects around from drawn animation?

CP: I don't consider that animation. It looks like animation, it walks and talks like animation, but I call it articulation because what you're really doing is moving, translating fixed pieces. If you draw something more than once, if you create it as a unique thing, then you have animation.

PD: If you're creating something like a wisp of smoke, if you can define that as an object and have it relate to a mathematical depiction of air moving through space, would it cease to be animation for you?

CP: I think what we have here is a kind of ambiguous area. That's definitely more like animation than articulation would be, but then again it's not entirely animation. It's closer to it. But I think about that as event animation. We're dealing with something that doesn't quite qualify as character animation, but in fact we are changing the form, changing shape, altering the whole thing. To the extent that it becomes increasingly less susceptible to mathematics, then it becomes more like animation. I accept the classic definition of animation as bringing something to life. Normally we think of animation as creating living things, When we animate a cloud, that cloud is behaving like a living thing, not some inorganic matter.

In that area there is a new class of programming called particle systems, where you deal with vapors and fluids, such things as flames, water, smoke, and gas, those things that are formed of particles. Most are generated with random numbers, systems of fractal mathematics that relate to things that are broken up into a pattern, a set of rules, which gives things a template, a recursive template that's always being used at every level. One blade of grass will look the same as the others, not exactly the same but similar. There are many kinds of things you do with particle systems. You can make them look quite different.

Now I think that is animation. That's the beginning of animation, because at least conceptually you're changing form. You take on more characteristics of what's involved in character animation.

A simple test of it to me is that if I have to read in a whole new bunch of data or put on a whole new cel, it's the same thing. It's a frame by frame event. As opposed to a cel that pans or where the data doesn't change, its relationship to the world around it changes. It moves, it doesn't change. No matter if it's a piece of paper, a cel, or a pile of data, a bunch of abstract numbers, if it changes internally, it's animation. If it simply moves, it's not. If you can generate all this by machine, that's fine.

For instance, you normally have to break things down to smaller and smaller units of time, starting with a large unit of time. That process of breaking time down into smaller and smaller units is such a principle of animation, but it's only an arbitrary method of describing the change that's taking place. So if you're using mathematical methods to change from one thing to another, then it's still that word, animation. It's a good enough word to use.

Graffiti: published by ASIFA Hollywood. For more about ASIFA activities call (818) 842-8330.