Bruce Cornwell

1974

Focal Press

A system for the production of computer generated motion pictures involves facilities and skills which differ somewhat from those associated with conventional motion picture production. Fewer steps in the process require direct control by the designer - hence his work is simplified. There are no provisions for him to make last-minute adjustments as the image is photographed. The computer follows instructions that are submitted only at a prescribed step in the process. Although this gives the designer less ease of control, the whole system can deliver a processed film within an hour, if not minutes, from the time that the last design or director's decision is entered in the computer program. The designer can therefore view a scene in its final form each time before he rethinks it, if this is necessary. Thus the system is unusually attractive for visual or filmic experimentation, or even for a number of conventional film production applications. But by far the most significant characteristic of the medium is that there are virtually no bounds to the variety of images a computer can draw.

The computer is an extremely adaptable tool, and the ability to design can be transferred to it quite easily. How can this be achieved?

Fig 2.1: Lines constructed at random by the computer from Lines in a Plane; Bruce and Katherine Cornwell

PROCESS AND APPLICATION

In its most readily accessible form, the computer requires that images be described in numerical or mathematical terms which might involve shape, size ratios, direction of movement, velocity of movement, and so on. This description is then translated into a series of concise instructions utilizing a common programming language. These instructions are probably entered into a computer by being keypunched on to the familiar IBM cards, and run as a batch mode program (along with a lot of other programs). If the run is successful, the computer generates a very extensive and compact sequence of binary instructions which in turn drive a microfilm plotter that exposes a series of images on to motion picture film stock. When this film is processed the movie is ready for viewing.

Let us now consider how the computer can be adapted.

Some digital computers are used solely for one application, such as for the control of a manufacturing process. But an identical computer can perform a wide variety of jobs, on an hour to hour basis or simultaneously, with so little delay in response that each user has the impression that the computer is exclusively at his beck and call.

PROCEDURAL INSTRUCTIONS

In order to operate, each computer has one or more sets of language procedural instructions, or compilers, which are stored permanently in the computer, and which are revised possibly once or twice a year. Each compiler enables the computer to accept instructions in one language. The type of language used depends upon the general type of task to be performed. For instance, the computer language known as Fortran is used for the simulation or the solution of scientific problems, but Cobol is more adaptable for accounting problems. Another type of language which should be approached with caution is an assembly language. This is a low-level language that allows you to come more closely to talking with the computer on its own terms. It has a certain appeal because it is very detailed and precise. The use of this language bypasses the need for software, which is not, as it may sound, a simplification in the production of films. Assembly language involves many repetitious instructions and is useful primarily in writing software packages. People skilled in assembly language are often eager to apply their talents toward writing a package that will precisely fill a particular need, such as that of motion picture production. However, if one wishes to learn to paint, it would be foolish to engage a chemist to create the pigments, even if one did wish one's paintings to be unique in the world of art. Only if one cannot find access to any graphics software package, is the long road of the assembly language justified.

In much early use of computers it was felt by non-mathematicians that ideally the computer should be addressed in a language that was simple for the operator to use, approximating to English as closely as possible. In fact, the language of mathematics prevailed. The nature of the animated motion picture, when each visual is broken down into specific steps, lends itself to mathematical description, and so we find that Fortran is a reasonable language for both the programmer and the computer. Although other scientific languages are used, Fortran is currently the most widely adopted language in the United States for the generation of computer motion pictures.

Visuals are broken down in the following way. Any visual element in the field of the camera aperture can be located as a position on an x-y plane. When the visual stays in a fixed position, this length of hold is entered as a number of frames. When an element is moved at a uniform velocity from one location on the screen to another, the amount of movement per frame is arrived at by dividing the total distance by the number of frames. Obviously the symbolic shorthand of Fortran is ideal to handle this sort of process. However, the choice of language should be resolved in conjunction with the search for access to a computer facility, a software package, and possibly a microfilm plotter.

A software package is the collection of detailed instructions, stored within and used by the computer. Many of these instructions are the solutions of mathematical formulae, so that an elaborate problem can be solved by a very simple request in the program. As computing technology has developed, the software packages have gradually become more sophisticated so that more and more elaborate responses can be called out with a simple instruction. Software packages have also become increasingly specialized, and one can shop among them or even have one specially designed; the experienced programmer can of course write his own.

To create computer films it is desirable, though not essential, that the computer has software packages of both graphic and movie instructions. Whichever of these packages is used, it will enable the computer to accept more sophisticated instructions than the mathematics and logic which the language can deal with. These software instructions may seem familiar to the non-mathematician, for they are often initiated by giving the computer a specially selected key word, such as write, zoom, draw, frame, line, circle and so on, followed by the letters or numbers which are abbreviations for the particular conditions required. All of this is co-ordinated within the Fortran, or whatever compiler language the software is written around.

To use such a package you no longer need to know the formula for a curve, such as a circle. You simply have to call out in the appropriate symbolic language, circle, and lo! a circle will be drawn. An average twelve-year-old can undertake this type of work, albeit with less speed than the experienced programmer. If you have lost the mental agility and self-confidence you had at twelve, or if you do not have a flair for logic and accuracy, the process of learning a language like Fortran may take six or even twelve weeks. There are, of course, many books on the common programming languages, so that you can study alone, but you must have access to a computer so that you can practice as you proceed.

The more direct approach towards making computer films, if you are more interested in visualizations than in computers and think of both the computer and the programmer as tools, would be to seek the co-operation of an experienced programmer who has worked with the language required.

GRAPHICS SOFTWARE PACKAGE

The graphics software package extends the capability of the computer to perform repetitious tasks in creating instructions to be transmitted directly to the plotting device, or recorded on to magnetic tape for use with an off-line, separately located, plotter. Such a package usually comprises several boxes of punched cards, or a single reel of magnetic tape. It is prepared for a specific make of computer and for use with a particular type of compiler (such as a Fortran or an Algol compiler). A graphics software package enables one to create single images. A succession of images with slight variations between each, or an animated motion picture, can be created by using a standard graphics package with the available compiler language to accomplish incremental variations. Graphics packages are widely available for creating graphs, mechanical drawings, perspective drawings, schematic diagrams, logical diagrams, and original graphics for printed circuit board manufacture. Any of these might be used to create various realistic or non-objective images.

Despite their physical portability, software packages cannot easily be swapped around even among similar computers. The operating system of a large digital computer is a very sophisticated collection of routines by which the machine can keep accounts of its usage, detect errors in the user's programs, inform users, by printed messages, of the location and nature of these errors, among a multitude of other jobs. This sophistication is usually a source of problems when one installs a new software package in a computer. To overcome such problems may require anywhere from a day, to months, of work for a system analyst. It is simpler to find a computer that already has a graphics package in operation. This package might not be kept in the computer, but stored on a magnetic tape and then read into the computer whenever a graphics job is run. So, instead of selecting the computer first, it is wiser to seek a working graphics package already installed on a computer.

Graphics packages vary widely in what they enable the programmer to accomplish with ease, and in the type of plotting device their outputs can drive. The conventional graphics packages that have been written for the creation of static images may have one or more of these features:

  • Capital alphabet and numerals, one size only.
  • Capital and lower case alphabets and numerals, one size.
  • Either of the above in a range of sizes.
  • Straight or curved lines as in a two-dimensional plane.
  • Simplified commands for drawing x and y axes.
  • Simplified commands for drawing grid lines as on ruled graph paper.
  • Options for drawing dashed or dotted lines.
  • Options for drawing lines of different widths.
  • Perspective drawings in various projections achieved by locating each point or line in three elevations.

At least some of the above routines and some of the following routines comprise the essential features of the various motion-picture software packages.

  • Routines for gradually moving images into and out of the plotted frame giving the effect of a panning movement.
  • Routines for gradual scale changes giving the effect of a zoom.
  • Shield and window routines which enable an image or an area to either matt out or to matt in another image or portion of an image.
  • Routines for the storage of images within the computer memory so that recalling the image from time to time, or even from frame to frame, is simpler and more economical on computer time.

Fig 2.2: The generation of a perpendicular bisector from Lines in a Plane.

Fig 2.3: The generation of a perpendicular bisector from Lines in a Plane.

The characteristics of a software package can have a marked influence on the type of images that a designer can create. However, an imaginative designer can often come up with remarkable images regardless of how limited a range of routines is available in his package. A striking example is the Beflix language which can create very intricate and dynamic mosaics, both abstract and informational, through the device of plotting only one font of a capital alphabet and numerals with punctuations. The Beflix is a quite versatile language, but some of the graphic results might be accomplished, albeit with more effort, using a package that is limited to alpha-numeric output.

If it seems necessary to select the computer before the software package, it is important to assess the core-memory capacity of the machine. A computer with a Fortran, or comparable compiler must store the graphic or movie software with in the machine while it is in use. Various packages require different amounts of storage, but about 128,000 bytes is a likely amount. The core memory capacity can limit the complexity of the image that can be drawn within a single frame, or that can be stored for re-use in subsequent frames. The manner in which a computer describes an image is very similar to the scheme used in children's activity books where a picture is drawn by connecting numbered points with a continuous line, except that there can be several separate sets of points, or a number of separate lines, in a single visual. These points are not displayed but are written in computer language as instructions to the plotting device, which draws the connecting lines, or vectors, which are in turn exposed onto the motion-picture film stock. As a rule of thumb, we cay say that the more line vectors that are needed to make up a single frame, the greater the internal memory required of the computer. Incidentally, the more vectors there are per frame, the longer the time required for the computer to do its work, and for the plotter to draw the frame.

Beyond the minimal needs of core storage, there can often be economies effected by working with the large, relatively faster machines, such as the IBM 360-65. On the larger machines the usage charges are normally much higher, but the amount of time consumed for a particular run may be reduced by a yet greater factor.

TIME FACTOR

A major factor in the selection of a computing facility involves turn-around time, which is the amount of time that an attempted computing run is out of one's hands until the results are received. On virtually all runs the programmer must examine the printed output messages from the computer before requesting that the computer output tape be put on the microfilm plotter, which will create the actual motion-picture footage. During the initial months of working with a software package, from 60 to 90 per cent of the computer runs will have a printout that indicates that a tape was not written. In fact, the programmer will often put in test runs to determine if the computing procedures are operating correctly but will request within the program that no tape be written. If one is working with an on-line plotter, the procedure is quite similar. To the film-maker, the fact that a combined computer-plotter final run might take less than one day, including turn-arounds for a thousand frame run of intricate animation, makes it quite absurd to worry whether computer turn-around takes five minutes, two hours, or even three days if it is handled through the post.

However, the programmer's point of view, whether he is a novice or experienced, is very different. The amount of minute detail required in the writing of a program causes him frustration if he has to wait days to learn that one misplaced comma, or one data entry keypunched in the wrong columns of a card, had caused the computer to totally reject doing any work with the program except to indicate in the print-out the programmer's error or errors.

There are several ways of reducing turn-around time, and they vary in cost.

  1. Although tapes are frequently sent through the mail, it is generally more expedient to send the punched card deck to the computer centre by messenger.
  2. It is most desirable to work with a computer centre where the dispatcher is co-operative in estimating waiting time until the run will actually go through the machine.
  3. At commercial computing centres, you can usually secure a higher priority of run for an established price. This might mean that you can have the output in as little as fifteen minutes from the time of delivery. If you ask the computer to compile but not to write a tape, the price difference for a higher priority may be trivial even when compared with the savings of making a short run on the lowest priority time, such as an overnight run. At commercial centres the night rates are lower, and the turn-around can often take a number of hours if the job happens to fall into line behind a backlog of mammoth runs, which are usually deferred to the night hours.
  4. Time-sharing teletypewriter terminal access to a computer offers greatly reduced turn-around time, but involves considerably increased expense. These units incorporate a keyboard for feeding the program to the computer and a printer which gives a written record of the input as well as the printed response from the computer. Some more recent units incorporate a visual display for reading (essentially an oscilloscope) which allows you to read the printed input or output without paper print out. On the other hand, some units do have facilities for making a paper print of such a visual display, which offers paper output much like that of a teletypewriter, but at a higher number of words per minute. However, all of these time-share terminals offer the attractiveness of direct line access to a computer from your office, studio, home, or wherever there is a telephone line. Some of these units are quite portable and as easily carried from place to place as a piece of luggage. However, they can only work with a computer that is specially designed for time-sharing. When a graphics package is incorporated in a time-share computer, the output tape will usually be delivered to the plotter by other means than a telephone line. It can take many hours to transmit the contents of one reel of computer output tape over an ordinary telephone line. Moreover, in addition to the increased expense, the use of most time-shared facilities means you can not have your program on punched cards. Punched cards are quite convenient and economical for storing, editing, revising and re-ordering programs.
  5. A further refinement toward reducing turnaround time is the use of an interactive graphics terminal, such as the IBM 2250 or the Digital Equipment's PDP-9 with a model 370 light pen, which allows the programmer to specify images both with keyboarded instructions and with light pens. These units allow you to examine frame by frame stills and can show from a working program, (before it is translated by the computer into plotter driving instructions) a sequence of such stills that will be a reasonable facsimile of a movie. The economies of this type of facility depend greatly on the type of film images desired. If the images require many arbitrary human decisions, this route can offer tremendous savings of the programmer's time, and will allow the designer to actually get his hands on to the computer. But even here the designer should have a clear idea of what he wishes to accomplish. An interactive graphics terminal can become a very expensive scratch pad for doodling.

If therefore you can develop the ability to construct images in mathematical terms, the relatively slower means of submitting decks of punched cards to a computer for batch process is still a very feasible route from the creative point of view, and particularly for the novice designer of computer films. The use of conventional computer languages with punched cards or via keyboard input may seem to inhibit one's creativity, but as one gradually feels the power of logic and mathematics, freely drawn images with the light pen begin to feel very constrained. Neither system need exclude the incorporation of the other within a single production or even a single frame of film. The inherent flexibility of the digital computer allows for diversification of approach without any modification to the equipment at hand.

Fig 2.4, 2.5: A pair of tesselations that alternately generate each other, from Tesselations, Bruace and Katherine Cornwell

THE PLOTTER

For the film producer, the most important machine component of this system is probably the plotter, the device that actually draws and photographs the visuals. The simplest of computer-driven plotting devices incorporates an electro-mechanical system that moves a ballpoint pen over a sheet of paper. Of these X-Y plotters, the CalComp line is widely used. Some of the earliest computer-animated films were made by using the X-Y plotter to draw successions of images on paper, which were in turn punched as animation cells, or registered under the camera by using the marginal perforations already on the paper, and then shot on a conventional animation stand. For each variation in the visual the computer would produce a new drawing, just as in conventional animation. These techniques have been almost totally replaced by the use of the microfilm plotter.

The microfilm plotter translates strings of X-Y addresses into driving voltages for drawing the point-to-point vectors on the face of an oscilloscope tube, which is a tube similar to the television picture tube. Enclosed in a darkened area with the oscilloscope display is the microfilm, or motion-picture recording camera. This camera has no shutter. The moving point of light on the face of the scope draws the vectors one after another until all the elements of the image within one frame have been displayed. The image has a very short decay time, so that the moving point of light does not create lines behind it that linger on the face of the scope such as are commonly seen on radar sets. Rather, the microfilm system works photographically in much the same manner as when one opens a still camera shutter on a dark night and allows the lights of moving automobiles to describe lines within the picture. When the specified vectors have been driven on the scope for a particular frame, a computer output instruction causes the plotter to advance the camera to the next frame of film.

Because the image vanishes instantly, there is no way for the camera to shoot a number of frames of a single image from the face of the scope, in the manner that a hold-frame can be shot in other forms of motion picture photography, both live and animated. In computer animation, if a visual is to remain on screen for one second, or 24 frames, then the visual will have to be drawn 24 times. Although this may seem to be inefficient it is not a liability when compared with the X-Y plotter. The typical rate at which the vectors can be drawn on the scope by the microfilm plotter is about 200 per second. Whereas a mechanical X-Y plotter can draw only about five or ten vectors a second.

Cost efficiency for most computer users is usually in proportion to the speed of the operation. The X-Y plotters are very inexpensive to purchase compared with the microfilm plotter, but if you are buying time on a machine owned by a computer service bureau, it usually costs less to use the far more expensive but more rapid microfilm plotter. This is a paradox of working with computers: the apparently simpler way of doing something is not necessarily the least expensive. X-Y plotters are simple machines but not only are they slow, but their output necessitates the extra step of photography on a conventional animation stand.

The term microfilm plotter can be misleading when one is seeking a facility for computer-animated film production. Many plotters can create only printed characters of a fixed size and are essentially a substitute for the computer output printer. Other microfilm plotters handle only unperforated film as is used in microfilming, although some of these plotters can be fitted with an optional camera which will handle perforated film. For generating motion pictures, the ideal machine is a microfilm plotter with a vector capability which is fitted with a pin-registered camera mechanism. Only this type of mechanism can ensure a sufficiently accurate registration and therefore steadiness of the image from frame to frame.

The computer generates a magnetic tape output which must be read by a second tape drive unit connected to the off-line microfilm plotter. This is more efficient than having the plotter operated on-line, or driven directly by the computer, for three reasons. Plotters usually run more slowly than computers, and the cost of computer operation is mainly determined by the time it takes to complete a job. A second factor is that most computers will have many jobs that do not require a plotter as an output device, so if the plotter is independent, it will be possible for it to serve many computers in a large geographical area. Finally, if the plotter you normally use is not working it is often possible to proceed with work at hand by sending the output tapes to another computer facility that has a similar or compatible plotter.

Unfortunately, there is no way of editing a computer tape. The tape has seven or nine parallel tracks of binary information, which consist of nothing but machine-translatable ones and zeros. The only degree to which the tape can be edited is to request the plotter operator to run a portion of the tape only and even this is awkward; you should allow a 50-frame leeway in your request, for on many plotters the operator has no way of monitoring the process for visual cues. Furthermore, there is no means of previewing a tape as there is on a film editing bench. Only when a tape is mounted on a plotter will it create images. Also, if a tape for some other application is inadvertently mounted on a plotter, no images of any sort will be forthcoming. The plotter will indicate to the operator that the tape has nothing but errors.

The consideration of computer generated tapes as a means of driving the plotter leads an inventive person to speculate on other possibilities. Could one create a hybrid operation by substituting a video tape recorder, or a video playback unit for either the computer or the plotter, in the hope of achieving either some economy of cost, or of effecting a novel film or television output? Unfortunately, the answer seems to be No. Although both the microfilm plotter and the television image are generated on the face of a cathode ray tube, the two systems could hardly be less compatible. Whereas the television image is created by a methodical scan on the picture area, much as your eyes move down this column of type, the vector plotter follows a multi-directional path, much as an artist would move his pencil over the surface of his paper. Moreover, on television the shades of grey or of colour are varied continuously throughout the range. The vector plotter has only a limited selection of line weights available, and some plotters have only one line weight.

THE IMAGE

The use of the term vector should not lead one to think that this system is limited to drawing only straight lines. If one were to lay a piece of tracing paper over an S curve and then proceed to evenly distribute about 30 dots along the path of the line, these could be considered to be vector end-point addresses. If the artwork that had been used to position the dots is removed and each adjacent pair of dots is connected with a straight line, the sequence of lines will be a reasonably good approximation of a curved line. A circle can be drawn with sixty or so equally spaced points, or as few as eight or twelve points if it is to be projected on a very small screen.

Instead of plotting a vector, a single X-Y address can cause the beam to display a single point of a diameter that is approximately equal to the line width. Thus a series of points can describe a curve. But this is far more expensive in computer time than drawing a number of short vectors end to end.

Shading in a solid area can be done by plotting parallel lines, uniformly spaced and quite close to each other. But this technique can require quite a long computing time, depending on the complexity of the area to be shaded. For the computer to locate by trigonometry the end points of 200 parallel vectors to shade a large circular area, the computer time will be 12 times, as long as would be required for it to draw a congruent circular outline constructed with 33 vectors. Moreover the plotter will spend about twenty or thirty times as long drawing the shaded image, for the ends of these vectors are not connected sequentially as in the outline circle, and these vectors are also longer.

The graphical capability of the computer driven plotter has a much higher resolution than a television image. The most primitive plotters have a plotting area that can be addressed with in 1/1000th of the height or width of the area. The resolution of United States television allows only about 550 × 600 increments of position. Of the more recent microfilm plotters, the Stromberg DatagraphiX 4060 has approximately 3000 × 4000 addressable locations and the Information International FR-80 approximately 12,000 × 16,000 addressability.

However, resolution is not necessarily related to addressability. The optimum image that the FR-80 will give is a line width about 1/8000 of the width of the 35 mm camera aperture. There are five selections of line width available under control instructions that drive the plotter. When the 16 mm film shuttle is used on this machine, the image is electronically scaled down to fit the 16 mm shuttle aperture, thus using only the central portion of the fixed size plotting area. Consequently the ratio of line weight to film frame size is greater than in 35 mm usage.

Since the microfilm plotters have only limited, or no variation of line intensity, it is customary to use high contrast black and white film stock in the cameras, in order to achieve maximum resolution, or sharpness, of image. The exposed film is then processed as a negative, giving a black line image on a clear background. Or it can be reversal processed, giving a clear line on a black background. The latter is usually preferred. The use of high contrast film means that the different weights, or intensities of line will register on film as widths of line. This is rather like a draftsman's pen, for the line width cannot be changed while drawing a line.

Fig 2.6: A dragon curve is constructed by sequential folding of segments repeatedly, from Dragon Fold, Bruce and Katharine Cornwell.

Fig 2.7: Rotating dragons lock noses, from Dragons Fold.

Fig 2.8: Dragons lock tails.

Fig 2.9: Nose-to-nose and tail-to-tail the dragons form a space-filling curve, from Dragons Fold.

It has been suggested that there must be a gold mine of experimental potential with microfilm plotters, by using colour film with filters, or dissolves, defocus, multiple exposures and so on. Some people have even suggested extending the scope electrical leads so that the images can be photographed directly on the optical printing bench.

The other alternative is to generate the high contrast black and white footage on a standard microfilm plotter and then take it to an optical bench, or other printer, for all the special effects.

The latter seems far more feasible since the output of a quality colour film image requires careful calibration and set-up procedures, if not exposure tests, and occasional re-runs, all of which are very uneconomical to perform on the plotter. A microfilm plotter could cost as much per hour as an optical printer must earn in a day. It seems unreasonable to put an extra work load onto a very expensive machine when that work can be handled with even more flexibility on a less expensive one. With the present rapid advances in electronic equipment design, plotters become obsolete within about five years, which makes it less practical to load one with extra components that cannot be used on a stand-alone basis when the plotter is discarded. And, finally, there are advantages in the use of a separate optical printer for black and white work which will be discussed later.

Fig 2.10: A computer simulation of aircraft movements on the proposed Dallas-Fort Worth airport. An engineering study by Tippets-Abbett-McCarthy, Startton, New York.

COLOUR

Now that colour is becoming available in various computer display and plotting devices, including microfilm plotters, the use of an optical printer in conjunction with black and white plotter output may soon be obsolete. How soon is a matter for speculation.

These advances in colour were motivated by the fact that most graphs, drawings, maps, charts and even printed text, can convey information more deftly if certain visual elements can be colour coded. The colours themselves need be no more exact in quality than those available in ballpoint pens. If the future use of computer-driven plotters and display devices follows the present trends, these colour standards will serve the vast majority of needs and colour capability will probably be incorporated in most machines available at commercial and educational institutions.

It is interesting to note that once colour capability is added to a plotter system it becomes a very easy matter for the computer to call in these changes, as compared with the much more complicated procedures required for the storage and accessing of type fonts in various styles and shadings, or of lines that might be dotted, dashed, or ruled in a number of different ways. This is the reverse situation to that of a printing press where a whole range of fonts is available, but the total job cost is multiplied by the number of colours desired.

The assessment of the plotting device will also be influenced by the selection of the graphics package. In general, one can take output from any one computer to a number of plotters, but it is rather complicated to obtain a software package and get it working on any particular computer. Some microfilm plotters can only be driven by instructions of one format. For instance, the Stromberg DatagraphiX 4020 plotter can accept output only from packages designed expressly to be used with it. The Stromberg 4060 plotter which has greater resolution and finer increments of plot location (addressability), in addition to five steps of controllable line width, can accept output from packages incorporating these commands - or simply the more limited 4020 software output. The Information International FR-80 plotter, which has yet greater addressability and resolution of plotted lines, has its own input format for which packages certainly will be created or adapted. However, it can also accept computer output tapes in both the 4020 and 4060 format. The FR-80 can also accept tapes that are intended to drive certain X-Y plotters (pen on paper output) such as the CalComp line.

Fig 2.11: Variation on construction of dragon curve.

Fig 2.12: Here forms may be produced in colours.

Fig 2.13: Variations of space-filling curves

Probably the most common microfilm plotter with vector capability is the Stromberg DatagraphiX 4020. Although this machine has modest resolving power in terms of line width compared with aperture size, it is very adequate for a vast amount of experimental, if not production, work with computer generated films. Its one peculiarity is a square plotting area. Normally its cameras are fitted with lenses that centre this plotting area in the rectangular film aperture so that it fills the frame vertically when projected. This offers no problem when working with designs that are to be restricted to the central portion of the frame, but it does not allow lines to bleed off the sides of the frame. There is, of course, no way to discern the edge of the maximum plotting area on the processed film if no lines are chopped by bleeding that edge. If you are working with images that must touch, or bleed, all sides of the plotting area, it is quite simple to fit the camera with a lens which has a 33 per cent longer focal length than normal. This will cause some loss of the total plotting area and a corresponding loss of resolving power. The other alternative is to plot all final runs on 35 mm film and make an adjustment in plot size and placement on an optical printer by making a subsequent print of the original film.

ADAPTION

In the event of your gaining ready access to a vector plotter which does not have a camera for perforated film, and there being no budget for obtaining such a camera, there are ways of adapting conventional cameras to this type of equipment. (N.B. Cameras prescribed by plotter manufacturers usually sell for three to four times the cost of comparable motion picture cameras for professional use.) First, the camera under consideration should have an animation motor drive so that the film will be advanced only upon command of the microfilm plotter. The typical plotter will signal the advance with a 28 v DC pulse of 29 ms. duration. Secondly, the camera shutter should be either removed, or phased so that it closes only during the film advance. As stated earlier, different frames require different amounts of time for the plotter to draw. Then comes the most difficult task of all. The plotter circuitry must be modified so that the delay in plotting the next frame is sufficient to allow for the film advance. The cameras usually provided for plotters advance a frame of film in about 1/50 sec., and resume plotting in little more time. This modification could probably be accomplished by the manufacturer's engineer who maintains the plotter. At least his co-operation, and probably that of his superior, would have to be obtained.

If the adaptation of an ordinary animation camera seems too complicated, another alternative is to obtain a data recording camera such as the Model 207 Flight Research camera manufactured by Geotel, Inc. This is a pin-registered 35 mm camera which is actuated by a 28 v DC pulse and has a sufficiently rapid film advance. The price is quite in line with comparable motion picture equipment.

CONSISTENCY and RE-RUNS

The microfilm plotter creates images whose size and rectangularity are controlled by electronic circuits. There is no assurance that the same magnetic tape plotted on two separate occasions will produce identically positioned images on the respective films. One precaution is to hold all tapes related to a single production for one continuous run on the plotter. If separate runs are necessary, the operator should be requested to take care that the plotting area is aligned with his visual references. Even then, he may not be able to adjust the plot outline to perfect rectangularity. If there is a possibility of separate film runs being combined in optical printing, it is a great advantage to have every plot start, if not also end, with a matrix of plotted grid lines that will simplify visual alignment.

A number of plotters have what is called forms capability - a means of storing an image which can be plotted or photographed on any frame when so requested by a statement in the program. In the Stromberg DatagraphiX machines this is a photographic slide (usually high contrast film) which is illuminated by an electronic flash and photographed on to the microfilm before the frame advance is initiated. This feature was incorporated into plotters to save the computer and machine time necessary to generate ruling or grid lines and column headings which might have to be repeated on hundreds or thousands of similar frames of data. For the most part this is not a very useful motion picture tool; however it can be used to display a static background needed for an entire scene or film if its pictorial content is too complex for multiple reprocessing. In this case one could use a continuous tone film, a hand-rendered transparency (measuring about 3 × 4 in.) or even sheet colour film in conjunction with colour film in the recording camera. The latter would probably be possible only if one had a close working relationship with the management of the plotter facility.

Some of the newer machines such as the FR-80 have replaced the photographic slide with a system of storing and recalling vector and alphanumeric images from within a small memory integral to the plotter. This obviously precludes the use of various graphic backgrounds, but if a static, line-drawn background can be of use, there will be less tape for the computer to generate, and hence, less computing time. The simplest rule to keep in mind when estimating the saving of computer time is to assume that the computer requires approximately equal time to write the instructions for each vector.

If you are to have plotting done at a commercial facility, you may find considerable price variations among companies that are supposedly in close competition. Price is usually based on the machine time required for a run as indicated by a time clock built into the machine circuitry. As the number of vectors, or points, within a frame is increased, the plotting time is increased almost proportionately. If no information is plotted, the machine clock still logs the time used for advancing film in the camera. For some reason identical machines do not always run at the same rate when plotting from identical tapes, and dissimilar machines rarely run at the same rate, even when they can accept identical tapes. The difference can be as much as 33 per cent.

In addition to the variations in pricing for machine hours among commercial establishments, there are variations in set-up charges. These are the charges usually made for handling each order, regardless of how many reels of tape are included. Some establishments, however, charge extra for mounting each additional reel of tape. It actually takes a good operator no more than 90 sec. to unload one reel and mount another, but even if the machine is only idle for a short time this can amount to a money loss worth reckoning with.

A final warning for those who plan professional motion-picture production with these facilities, commercial or otherwise: the microfilm plotter is really a scaled-down, specialized digital computer with an analog output device, and, like most other digital computing equipment, on some days it just does not operate. Even the best of plotters can be down for two or three days at a time, and many are taken out of operation for half a day every two weeks for maintenance. So scheduling should be approached with more caution than is needed with the usual motion picture laboratory.

When working with commercial facilities for both computing and plotting the combined charges for these services can demand that you consider using an optical printer to freeze-frame print elements that hold fixed positions over a number of frames, or entire frames that may remain unchanged for a length of time in the completed film. In fact, the optical printing instructions can be plotted along one edge of each frame to be repeated in the optical printer, by sacrificing a minute amount of picture area. This idea is already incorporated into the Beflix software package which requires the output to be optically printed for particular applications.

RESEARCH FOR FUTURE SYSTEMS

Work that has been done with computer generated graphics and motion pictures at the University of Utah, and other places, points to the future feasibility of outputting images made up from areas of flat tones, ranging from white through a wide range of grey steps to black. The computer causes the plotter to create these grey tones under program control. The tones are phased in discrete steps rather than continuously as in ordinary photography, so there tends to be a slight edge effect where two different grey areas abut, even though they may be adjacent on the grey scale. In this type of application the computer transmits vastly more information to the plotter than a comparable system working in vector mode plotting. Consequently it is far more expensive in terms of computer time. In the conventional computer-microfilm plotter configuration there is a practical division of labour between the computer and the plotter: the computer gives only the end point locations for each vector, while the plotter is responsible for illuminating all the points between those end locations.

With the University of Utah system the computer must direct the plotter to sweep in every point on the entire screen, with the possibility of omitting some areas where there is no image. This puts the greater burden on the computer and so little on the plotting device that it is possible to improvise a plotter configuration from a Tektronix oscilloscope coupled to a modest slave computer that receives its plotting information from the main computer.

Ultimately, this or a similar system will be refined to the point where a continuous grey scale will be possible. But if one is to request 200 steps of grey instead of 20, and every point in the screen is to have a grey level specification instead of strings of adjacent points having identical grey levels, then this represents an enormous amount of work for the computer compared with the more usual present vector plotting requirements. Nonetheless, the real beauty of this system is the possibility of three colour separations with a colour control that would be in line with present motion picture standards.

As computer hardware and software technology advance there may well come a day when continuous tone colour-separated images will cost as little as the range of current monochromatic vector images. However, these same advances will surely make vector plotting proportionately more economical, and certainly far more accessible. Also, as continuous-tone plotting capability becomes more economical there should be increased use of optical scanning to input photographs or motion picture footage into the computer as digitized information. This should open up a whole new range of creative possibilities for manipulating these images that cannot be even thought of in the context of present photographic mechanical-optical systems.

A final variation of computer generated films to be considered are those produced by analog devices. Computer Image Corporation of Denver USA has done much to develop the technology and to encourage the production of films by this technique. Briefly, hard line images are generated by a variety of electronic circuits that modulate the lines to give shape, dimension, or texture. In turn these basic image units are animated in real time by control of knobs affecting dimensionality, or are controlled by output from a harness worn by an actor or a dancer. The harness gives the computer angular information relating to positions of limbs and body. Again, the resulting images are created in real time on a cathode ray tube and are photographed at 24 fps. The Computer-Image system can also accommodate the input of high-contrast black and white photographs or art work for similar manipulation. This system is probably one of the greatest labour-saving devices since the invention of hand animation, and certainly a vast range of effects can be achieved. The creation of compatible images with digital computers seems far more cumbersome by comparison. However, the digital computer is capable of simulating, or drawing, a graphic analogy of anything that can be described mathematically, which makes it rather like an extension of the mind - unlike the analog system which is an extension of the hand, or body. There is no logical answer to which route, analog or digital, any one artist should follow. The artist works with both mind and hand, so the ultimate device for artificially generating motion pictures may not yet be conceived.