GENE: How were the color separations and superimpositions programmed?
JOHN: The basic imagery was computed on an IBM 7094 digital computer at UCLA before we had any type of on-line graphic display equipment. So I ran the computer for seven or eight hours and took the digital tapes to General Dynamics in San Diego where they had a Stromberg-Carlson 4020 Microfilm Plotter. Initially I specified how many movies I wanted and how long I wanted each one to be. The program indirectly specified color based on the form or position of a figure. The output went onto three plot tapes which were converted into three pieces of black-and-white film. These pieces were used to mask primary colors in a contact printer. There was a piece of black-and-white film to represent the color red, a piece to represent the color green, and one for blue. Then I processed that footage with the contact printer for the colors specified.
A fascinating aspect of this film is that it traps the viewer into expecting mathematically logical transformations by developing in that manner for several minutes, and suddenly the forms explode or behave quite unpredictably. Once this effect has been fully explored, the solar system fades into a fish-eye image of people's faces and other representational imagery distorted, however, almost to the point of nonobjectivity. This sequence is printed in high-contrast, bas-relief positive-negative color reversals, in the manner of Pat O'Neill's 7362. In addition, the images are speeded so that a frenetic, visually distorted atmosphere is generated, suggesting extra-terrestrial creatures or anthropomorphic entities. The whirling multicolored geometrical images move across this bizarre background as though one were peering into a new dimension of existence. Dockstader's organized sound reaches a crescendo of chaotic dissonance as the final images of the film fade and disintegrate into nothingness. The sense of dynamic kinetic activity has been so powerful that this abrupt halt leaves the viewer suspended and breathless.
STEHURA: In writing programs in computer languages such as Fortran I've worked with about five parameterized models with which you can specify designs numerically. One is the "mosaic" scheme, which is the style developed for the Beflix language. You're building things with squares. Your basic figure is the square and you're building patterns and shading things with squares. The result is a mosaic pattern with chains of alternatives. The second scheme is the field model which I used for Cybernetik, in which you set objects or points in space and by controlling the strength of fields you produce image forms. A third scheme is a mathematical model of your arm as it would be used to draw figures. You define angles, specify arcs and curves, and work within those parameters. The fourth scheme I worked with is based on the deflection principle. Your mathematical model is patterned after a room or enclosure into which a ball is fired at high speed, bouncing from wall to wall. You plot paths of the trajectories, angles of deflection, distances traveled, the shape of the environment in which the projectile is moving. All this is simulated mathematically and was interesting because it presented form as the space between objects or containers. Finally there's the scheme I call "masking," which is similar to the idea of mattes in conventional filmmaking. The basic idea here is that you don't have a positive figure you're drawing, but you have masks or shapes which hold back light. You define a form and its motion, and you use that form to contain or exclude another image. It's like a cutout or translucency. You can treat computer graphics in that way.
GENE: Where has all this experimentation led you in terms of using the computer as an artistic tool?
JOHN: I discovered that working with program languages to produce graphics is rather hopeless. They're really designed for playing with numbers. A general problem with computer languages is that you get into simulating reality. That's the trip physicists and meteorologists are into. It's close to their way of thinking and their problems, but I think it's a waste of time in computer graphics or music. My explorations in computer language led me back to conventional animation, back to the paintbrush— a computerized paintbrush. So the next level, after playing with these language or parameter systems, was to establish another level of control over the computer. The idea of building a metasystem or a control system to control control systems appeared very interesting to me. o over the last four years I've developed a metalanguage which I call "Model Eight" since it's the eighth approach to these systems. It consists of a set of operators to work on one-, two-, or three-dimensional patterns: sensory patterns, music, drawings, motion, and so on. It's not a computer language which is operated one point at a time, but rather functions nonsequentially on large blocks of data. My idea was to develop a language which would synthesize all the schemes I mentioned earlier so that, for instance, in terms of work involved, whereas it took a couple of years to devise "Model 5.3" to make Cybernetik, my feeling was that it could be done in a couple of hours.
GENE: What sort of input-output situations are involved?
JOHN: Well, the two input systems I've found most advantageous are the optical-scanner and the light pen, drawing directly on the CRT. I have three modes of operation with the scanner: first, a point-by-point scan like a television scan starting at one corner and moving across, and you get a list of intensities which describe a picture, or just certain areas or colors. You can label it and manipulate the whole thing or just that part. Second is an isoline type of scan, which is what you see in weather maps: circles within circles which indicate certain degrees of intensity and so on. Then, third, there's the situation in which you start out with an isoline approach but produce lines which fill certain areas, and that output can appear on the CRT or be further modified.
GENE: What about drawing with the light pen? I understand that some artists, like Norman McLaren, have found this rather un-satisfactory.
JOHN: Well, the light pen is a crude drawing instrument, it's true. You can't do many subtle things, the resolution is low, and the way you operate you're always stopping, waiting for "INTERRUPT," for the computer to accept your line, or the accuracy always seems to be off, but it does have certain advantages related to large-volume image production. One of them is that you can input information that's more specific to the way the computer's operating. You put in a point or a line at a time, and by remembering what you're doing you can control a lot of image transformations. Representational forms are almost impossible to program with computer languages, and are extremely difficult for a computer to process. But by taking the alternate route, by drawing representational forms with the light pen you've given the computer graphic information which can be simply transformed according to simple motion and shading procedures. If you want to draw a dragon, for example, and have it transformed into a person, you simply draw the head and type in "HEAD" and then you draw the head of the man and label it "HEAD" and the computer operates on it to do the transformation. You can label portions as you draw to control the flow of the transformation. You can transform anything you can draw into anything else. And in this way you bypass much of computer language specification.
GENE: What relationship does "Model Eight" have to all this? What is the control situation?
JOHN: Drawing is a specific operation just as scanning or projections. With the language aspect you can specify a fish-eye projection or a projection on a certain plane and you supply the parameters. Now, after I've passed an image through a simulated fish-eye projection, what I want to do is start shading the forms in a different style. So you could call on a surfacing operator which will fill in your image with colors or mosaics. Now, my language isn't fixed. One "word" doesn't mean one fixed thing in one fixed context. For example, you have a mathematical model of an arm movement and you tell the computer to swing the arm in a 360-degree arc to define a circle. The basis of this operation is a sine wave to produce the smooth circular form. The fact that you have a sine wave is specified, even if by default. The output of this operation is a circle, a set of points. And as far as I'm concerned it's a wave form just as legitimate as the sine wave. So you could run this form back into the same particular operator and tell the computer to use this form— not the sine or cosine, but this form it has just described. The same recursive form applies to the other operations. For instance, you could take projections of projections, use an object as an element to shade a surface and so on.