This paper considers some of the design problems encountered in the production of computer generated motion pictures which simulate real events. The particular film studied is an airfield simulation study plotted from data input with a GPSS software package. Various optical printing techniques are examined which supplement the computer generated images. It is suggested that one might use innovative scaling devices and abstract symbols in order to effect a more adequate overall view of the action without undue crowding of symbols. However, the degree to which the film can compromise with accepted realism must depend upon the audience for which the film is intended: Because the film is such a powerful communication tool further study of the effectiveness of design elements must be made.
Within the spectrum of computer generated motion pictures that convey explicit information there is a wide range of films that simulate real events. These films are not strictly mathematical or statistical graphs, nor are they attempts at photographic like representations of actual scenes, but they are partially both. To justify their generally high cost, these films must communicate successfully to viewers on several levels, and hence, the design problems are far from trivial.
Basically, we are considering. films that are created by, (a) accessing digital computer data that relates to complex movement simulations, then (b) re-ordering and interpolating this data to give continuous movements with appropriately designed symbols within a scaled area plot, then (c) a final computer run translates this output into microfilm plotter driving instructions, and, finally (d) this output is either fed on-line, or off-line via magnetic tape to a plotter with vector capability which photographs the resulting animation with a motion picture camera.
The computer generated motion picture as an output from a digitally computed simulation study gives the user a more complete understanding of what the computer does with the input data than he can get from the tabular numerical output alone. The programmer who writes a simulation program, or one who at least puts values into the parameters of some previously written program, probably understands just what the computer is doing in the program. But this process can leave the personnel responsible for effective decision making in a passive role, and often unsure of the validity of the simulation. Black boxes that give numerical conclusions may be intriguing, but they are not always convincing. With imaginative utilization of flat images that can move, an oscilloscope display, or a motion picture, the communication gap between the computer and the decision maker can be substantially narrowed. And because the visual is a powerful tool, it is important to realize how modifications of the design format can affect decisions.
Film is slower to produce than is a visual on the oscilloscope display terminal, for in addition to the plotting time, it requires time for the developing, so the display terminal is ideal for debugging. However, once a useable series of images is realized, the logistic advantages of recording the visuals on film become obvious, simply from the standpoint of inexpensive repeatability, transportability, storage and recall. Moreover, these computer films can be optically printed to provide a composite of related images.
An elementary technique of optical printing is to combine a dynamic image with a static background (Figure 1). A sample frame from the film of aircraft ground movement on the proposed Dallas-Ft. Worth airport illustrates the use of the computer to generate the dynamics of the moving planes, shown as plane symbols, the occupied parking areas at gates and aprons where planes are shown as squares, synchronized with the numerical clock in each frame of film. The rest of the film image, the runways, terminals, and labels, was photographed on a motion picture animation stand from manually prepared artwork. Only a short length of such background is needed, since the optical printer can print repeatedly from one frame.
The engineering company Tippets, Abbett, McCarthy, Stratton who produced the Dallas-Ft. Worth film used this study to show and to compare two separate conditions. The basic computer simulation from which the film was derived examines an expected typical heavy-load hour of airport operation circa 1975. (This airport has not yet been built!) The first run through is with the gate space allocation as suggested by the airlines involved. The second run through is with the same arrival-departure schedule, but with modified gate allocations and apron configuration. The film runs at about ten times as fast as real life so an hour of airfield operation requires six minutes of screen time.
These two runs might be combined by optical printing so that the viewer could make a continuous comparison of the results. A color print can be made from the black and white computer film giving red symbols for the less efficient run and blue symbols for the better run. By precise registration, the symbols that are unaffected by the change should print a coincident image that will approximate white - by the addition of two near-complimentary colors.
This technique of multiple exposures of separate runs on a single length of film could combine as many as eight or twelve outputs of the basic display sequence. For instance, when the movements of a stream of vehicles are initiated by randomness, it is common for the programmer to make a number of preliminary runs with variations of the random generator start numbers in order to find the run that seems to have typical characteristics. This is then presented as the typical study, and is translated to film output. A more valid output might be achieved by taking a diverse series of runs and combining them by successively overprinting them onto a continuous tone motion picture stock so that the intensity of the image has a mathematical correlation with the density of coinciding vehicles.
A film composed of such a stack of images that give a variation of photographic density becomes a tool for abstracting a mass of information. Whereas, the more usual use of film is to create a realistic representation, with a discrete movie-like display, which approximates a view from a helicopter.
As one works with computer films there emerges a persistent conflict of objectives. First, there are films in which one tries to exactly create, or to re-create, certain realistic conditions, such as in a film which will anticipate a pilot's experience in guiding an airborne or spaceborne vehicle. Such a film can involve very complex input devices, software, and output devices. On the other hand, effective computer films can be dynamic x-y plots which are simply mathematical graphs with an extra dimension, time, as real time or screen time. In neither of these two types of film is there much conflict as to what is real and what is abstract.
However, a film such as the Dallas-Ft. Worth airport simulation can raise a number of questions. It might be argued that this film should approximate the output of an aerially suspended motion picture camera, pointing straight at the earth from directly above the center of the airfield. With this goal in mind, the planes are drawn as closely to scale as film plotter resolution will permit, all planes will have noses and tails, and congested aprons are drawn to the same scale as those runways where planes seldom get within thousands of feet of each other. But does such an exact replication necessarily make the visual presentation more clear?
One reason for the production of this particular film was to show exactly what was going on inside the General Purpose Simulation System with this specific input data. GPSS is an elaborate piece of IBM software which was modified to go through the thousands of detailed steps to describe the movements or locations of hundreds of aircraft within an elaborate piece of geometric diagramming, the planned airport. Normally, the GPSS gives a tabular summation of operating efficiency, with ratios of cost, time, facility utilization, and so on. This output data is not the work of a concise mathematical formula, but, rather, it is the result of the computer keeping tabs on a detailed run through on the data fed into the GPSS. This run through is never seen as a graphic movement unless an auxiliary program is tied in with the GPSS to continually translate it into plotter instructions, which in turn give frame by frame display of all data as scaled locations of the selected symbols. The format of any film visualizing the GPSS execution, becomes a series of arbitrary decisions. These decisions are often tempered by a consideration of the target audience for the film.
If the audience is to be solely the operations research people who are concerned with the validity of the GPSS output, their critique is likely to be confined to whether or not the points that represent instantaneous location of craft move in a proper manner. However, if the audience is the civic leaders who personally raised the funds to build this lasting monument, this airport, then one must be prepared to embellish the film with refinements that will make the film more impressive but that are not necessarily relevant to the GPSS execution. These seemingly slight additions can often cause more problems than the skeleton program which gives the display of moving points. Although the film may be made more realistic in appearance, care must be taken not to compromise the clarity and useability of the film for the operations staff.
Other alternatives are possible. If the computer is only required to plot points instead of symbols, there will be an obvious economy of computing and plotting expense. From this film a composite of two images could be exposed on the optical printer. One would be sharply in focus, giving a precise location for every point on each frame, and the other image would be the same points but generously defocused and of less exposure so that a disk of light surrounds each reference point (See Figure 2).
A viewer can readily detect the movement of a halated point even if he is not looking at the precise point. The scene of the Dallas-Ft. Worth film (Figure 1) displays about 18,000 feet east-west, with the planes drawn approximately to scale. If one is following action in one quadrant of the screen, action can be taking place in other areas that will not be seen except by repeated screening. From the practical point of motion picture design, these symbols are too small. This is said with all respect to the high resolution output of the FR-80 microfilm plotter at Computer Micrographics in Los Angeles and the meticulous optical printing of Elliot, Unger, and Elliot in New York City.
An alternative to enlarging the symbols until they are out of scale would be to make an inconsistent scaling of the airfield. This calls for a cartographic projection that would approximate a photograph with an extremely wide-angle lens as sketched in Figure 3. This would give maximum spread to areas of greatest congestion and activity and would combine a realistic symbol with an adequate overall view of the airfield. However, whether or not such an unconventional image would detract from such a film is debatable. Often people who pay for films are much more apprehensive about image side effects than are people who design films.
In general, if symbols and diagrams are going to have maximum effectiveness as tools of communication, they should be designed for optimum clarity. If they are part of moving images, simplicity in design can often result in a more pleasing image as well as in economy of computing and plotting. Static diagrams can be gradually refined as they are designed. The results are obvious and viewer impact is fairly predictable with conventional information. Conversely, in the design stage, there is hardly such a thing as a predictable motion picture film. Motion can be tempered by many factors. Often the motion of a simple shape is more attention getting than the identical motion of a complex shape - even though the latter may more closely resemble some sort of reality.
Although computer films have a vast immediate potential as verification of GPSS type programs, and as an effective presentation of simulations, there are a number of factors that work against the fullest development of such films. Too often they are designed to present the most elaborate film for the large sum of money that had to be approved for their production. This can work at cross purposes with effective design if polished reality is to be the major criterion of a quality image. So little has been done with dynamic symbols, or motion graphics, that there should be a means for experimentation to be carried forth without concern for the polish or realism of the finished product, if these are irrelevant to the validity of the communication, or to find out whether they are relevant.
Although the problems that can arise in handling vehicular symbols are great, a problem of an even higher order of magnitude arose in another project. How does one depict the flow of passengers through the complex of an airfield terminal area? Moving dots, one per person could be very expensive in computer and plotter time, if one wished to keep track of about 1,000 people in a ten-gate area, and keep them from bumping into one another. The first bit of realism that might well be discarded is the no-bumping. Let the dots bump, or overlap, as long as the program keeps track of the total density of people per unit area and does not exceed that. And a second thought, if an area is full off moving people does it really make any difference in the facility utilization study in which direction the people are moving? Such areas might be shaded in a solid cross hatch. Even in sparsely used areas, the direction of movement may be inconsequential. Now the problem is to suggest something like density contours to simply indicate people-density which changes with the net change of people within the units of area. But contours have the drawback of giving similar images where there are sizable areas of uniformly heavy density of uniformly light density. The answer to these questions must be visually logical, and mathematically logical if the computer is going to process the scene with any degree of efficiency.
A final factor which might influence design is whether the output is intended to be screened many times by the operations research staff, a few times by the executives involved, or if there is just to be one gala showing before the financial backers.
Beyond these design problems is the fact that simulation films in this area can bring the decision-making people much closer to a full understanding of what the computer program is doing with the operational side of this simulation. This is especially true with airfield simulations if the viewing executive has actually flown commercial craft or has sat in a control tower. Now he can talk with the operations research team directly in terms of what the program is doing on screen. For him, it is now a projection of reality. Similarly, discussions between operations executives and airfield construction engineers can become much more objective when differences must be resolved. When a parameter is changed, the film will show what the effect will be.
It is precisely because computer films can present data in a visual language that is equally clear to viewers from different levels and with different perspectives that makes it such an effective communication tool.
1. Halas, John, (Ed.), Computers and the Art of Animation, (Hastings House, London, 1971 ).
2. Sih, Peter, "Aircraft, Ground Operations Simulation", Proceedings of the 4th Annual Simulation Symposium, March 1971.
3. "Computers, are they giving you the best for your money?", Engineering News Record, October 8, 1970.
Bruce Cornwell is an independent producer of films and is now specializing in computer generated films. He has designed and produced some thirty animated films in mathematics. He was producer and consultant to the Calculus Film Project of the Mathematical Association of America. He was Research Associate with the Computer Animation Film Project of the Polytechnic Institute of Brooklyn, was Director of Film Production (Computer Generated Films) for General Computing Corporation, and is on the faculty (Information Processing) at the New School, New York. Recently his films were singled out for a special showing at the national meeting of the Mathematical Association of America, and they have been shown at festivals and events in Edinburgh, Venice, Geneva, Brussels, Buenos Aires, Padua, and Pardubice.