A discussion is presented of the research and development involved to produce 35- and l6-mm color film as a direct output from CDC 660 computer runs.
Since July 1968, experimentation has been carried on at the Los Alamos Scientific Laboratory to determine the feasibility of color film output from the CDC 6600 computers. The goal was to produce magnetic tapes on the computer which would generate color film on the Stromberg-Carlson 4020 film plotter (hereinafter abbreviated SC-4020). This film would then be processed by Group D-8 (Graphic Arts) as generated, without additional tinting or washes. (Color film has been produced at other installations by tinting black and white runs.) Further design goals were that color film would require no more computer time than an equivalent black and white run and that sharpness and clarity be maintained. Previous work in using direct color film output required overstriking any given plotted point to achieve good results (see Section III).
This preliminary report summarizes work done to date, hardware development, and the advantages of color film output. The prints shown in this report as Figures 1 through 10 are taken, by means of intermediate negatives, from 35-mm color film generated on the SC-4020. No special touching up or tinting by hand was done.
There was an initial period of hardware research and development, with collaboration between Group C-8 (Computer Engineering Support) and Group D-8, concerning filter types, lens characteristics, film type, and cathode-ray tube intensity. Runs generated on the CDC 6600 were used to produce SC-4020 film output. The processing variations were done by Group D-8. The present operating system was developed during this initial period.
After this initial phase, interest in color waned, due to the necessity of overstriking. In March 1969, the question arose as to the possibility of color film output for MAG4, the magnetohydrodynamics (MHD) code for Sherwood (see Appendix A). We decided to look into it, provided that only a single strike per point would be necessary. Replotting the same point many times to achieve color output was prohibitive in computer time.
An interesting film strip had been generated during the initial research period (see Figure 1). Each frame on the film strip had been washed with a different color. On these various colored backgrounds, vectors were drawn, each vector having a different number of overstrikes. It seemed to show that if the film could be presensitized by sweeping the background, only a single strike (as for generating black and white film) would be necessary (see Section IV). Various tests were made using a blue background sweep, but normal developing by Group D-8 proved unsatisfactory. Therefore, the film was developed by a one-stop and two-stop forced processing. The one-stop forced development yielded the best results (see Section IV). An 8½ × 11-inch paper print made from 35-mm film is shown in Figure 2. This print represents a Sherwood plasma run, with magnetic fields, in the so-called tearing mode.
The development of a totally new, in concept, color filter device (see Section III) markedly decreased the time required to change color filters on the SC-4020. Using another version of the MHD code, a production run was made to represent hot gas expanding into air. A print of this run, which used a green background sweep, is shown in Figure 3. Tests were also made with a red background, but it proved to be unsatisfactory.
These runs still left us with two problems. There is a total of four colors on the filter device (red, blue, green, and white). If one of the colors is used to sweep the background, only three are left to do the plotting. Also, the time to process the magnetic tapes on the SC-4020 increased due to the background sweep. Thus, the ideal situation would be a single strike per plotted point with no background sweep. Thus, the generation of color film would require the same amount of time on both the computer and the SC-4020 as an equivalent black and white run.
New filters were installed on the color filter device which permitted more light to pass through. Also, the lens aperture was set to an f-stop of 3.5 (see Section III). The resulting film was developed several ways, with the two-stop forced processing yielding the best results. About the same time, new routines were added to yield better results in drawing large characters on the SC-4020 through the use of short vectors instead of plotting dots.
In August 1969, two production runs were made utilizing the new techniques. The first run (see Figure 4) depicted hot gas expanding into air. The second run (Figure 5) was similar except the air was partially ionized, yielding a lower density. As can be readily seen, material separation and imbedded contour lines stand out in color. A black and white print of the same run is given as Figure 6. Two short 16-mm movie films were made from these color runs for viewing through a projector. A polar coordinate run, using a blue background sweep, of expanding gas is shown in Figure 7.
The advantages of color over black and white film are many and varied. Color has a distinct impact that black and white does not have. Wherever color film or prints have been shown, the reaction has been startling and favorable. Color also has utility as an analytical tool. Figure 8, which strikingly depicts material separation, was generated by a computer run that represents the density profile of three isotopes across the face of a reactor fuel element. Each isotope was given a different color that readily distinguishes each material. It would be difficult, if not impossible, to generate such a plot using only black and white.
As shown in Figure 5, contour lines in the expanding gas are more clearly visible. For comparison, Figure 6 is a black and white print of the same frame. The difference is evident. Note, too, that material mixing cannot be seen on the black and white print. However, the two colors of the two materials yield a third color on the print to dramatically illustrate material mixing, Color adds another dimension to computer generated output.
What is possible using color film output? No one can yet say, but some idea can be obtained by referring to Figure 9 and Figure 10. These are demonstration runs to illustrate color film output. Several people are currently using color output from the computer and many other examples could be shown. Research is continuing and we hope to improve on what has already been accomplished.
The technique of producing microfilm copies of computer-generated information by photographing it while it is displayed on the face of a cathode-ray tube (CRT) is commonplace. Until recently, however, essentially all such microfilming was in black and white despite the advantages that might be gained from the use of color.
In 1967 Sandia Laboratories, in cooperation with Stromberg-Datagraphics, Inc, initiated a research program to explore the possibility of obtaining color film output from a Stromberg-Carlson SC-4020 Printer/Plotter.
[C J Fisk, Cathode Ray Tube Plotting, SC-BR-68-146, Sandia Laboratories, Albuquerque (January 1967)]
A new phosphor mix, containing two parts P2 Red and one part each of P22 Green and P22 Blue, was used in the production of a new CRT for the SC-4020, inasmuch as the standard mix lacked sufficient red. In addition, a color filter wheel, driven in one direction by a stepping motor, was placed in the light path between the CRT and the camera. Kodak Wratten filters, 29 Red, 61 Green, and 47B Blue, were installed in three of the quadrants of the wheel; the fourth quadrant, with no filter, permitted transmission of white light.
It was thus possible to position a filter, produce a display in color, and photograph it on color-sensitive film. However, because of the sharp cut-off characteristics of the filters, reduced intensity of the new CRT, and limited speed of the film, it was not possible to obtain adequate exposure except by overstriking, i.e., displaying the same information several times.
This method produced satisfactory output but required an average of seven overstrikes, greatly increasing the amount of computer time and SC-4020 time required to generate a frame. SC-4020 time was further increased by the interval required for the filter wheel to rotate and come to a complete stop. To move to an adjacent filter position, 90° away, required 200 msec and, because rotation was unidirectional, 6oo msec was required to select a filter 270° away. This characteristic required care in programming to avoid excessive motion, and additional software was needed to monitor the position of the wheel because only the no filter position could be selected automatically under program control.
LASL became interested in the feasibility of modifying its SC-420 to produce color output and initiated a study of the Sandia method and other approaches.
A CRT with the new phosphor mix was ordered from Stromberg-Datagraphics, Inc, and research into the implementation of a faster filter positioning device was begun.
The first avenue of experimentation involved, as a basic mechanism, a stepping switch driven by a rotary solenoid. The device operated well and reduced the filter rotational. positioning time to 50 msec. As an additional advantage, by using one wafer of the switch to indicate position, each filter could be associated with a unique operation code and thus reduce software requirements.
Limitations of the system involved quality of mechanical components and a rotary solenoid relaxation period of 65 msec. (Output prints from this system are shown in Figure 1).
Further research used an ICON-Fujitsu Model 109 heavy-duty stepping motor, capable of rotating the color filter wheel by one position and stopping it in a stable condition in 50 msec. The disadvantages of this stepping motor were its high power requirements, the need to cool associated electronics, the complexity of the circuitry to generate the necessary slow-fast-slow pulse train to drive it, and the additional circuitry for position monitoring.
The use of three separate solenoids, one for each filter, was then studied. A device consisting of a filter attached by an arm of 20-mil magnesium to the backplate of a Ledex rotary solenoid was constructed for evaluation. The schematics for the solenoid drivers and the filter control logic are shown as Figures B-3 and B-4 in Appendix B. Actuation time was 18 msec and relaxation time was 25 msec. A timing chart is shown as Figure B-2 in Appendix B. Attempts to reduce the relaxation time by increasing the tension of the solenoid return spring were unsuccessful. Because this device was much faster and simpler than previous mechanisms, it was a prime candidate for further design effort.
To check reliability and durability, circuitry was built to actuate the solenoid 50 times per minute and the mechanism was operated continuously 240 hours for a total of 720,000 actuations. Subsequent inspection showed neither damage nor appreciable wear.
The solenoid employed in the LASL system is a Ledex 28-V Model 810-360-330. Drive circuitry is designed to furnish full power for actuation and to provide reduced power to hold the solenoid energized without overheating. To match existing signal levels and facilitate maintenance, color control logic utilizes standard SC-4020 printed circuit cards. A simplified block diagram to show the interrelationships is Figure B-1 in Appendix B.
Three existing, but unused, operation codes, Select Camera One (Octal 41), Select Camera Two (Octal 42) and Select Both Cameras (Octal 43), have been diverted to select the red, green, and blue filters, respectively. A fourth operation code, Octal 40, cancels filter selection and permits passage of white light to the camera.
Because each filter-positioning solenoid operates independently, the control logic is such that selection of a filter simultaneously releases any previous selection. Therefore, during the same interval that one filter is being positioned, the one selected earlier is moved out of the light path. The control logic introduces a delay interval during filter selection sufficient to accommodate the 25-msec solenoid relaxation time plus 5 msec to compensate for filter arm oscillation following motion. Thus, the time required to select any filter is 30 msec and filters may be selected at random under program control.
A two-position switch was added to the upper Customer Engineer Control Panel. In its COLOR position, the filter controls are enabled for runs using color film. The NORMAL position disables them and causes the filter-select operation codes to be ignored should they occur during runs with black and white film. This feature also allows program check-out on black and white film prior to making color runs.
The solenoid housing is constructed of 0.25-inch thick aluminum to inhibit vibration and is mounted on the pellicle (mirror housing) just below the camera. The solenoids are arranged in a semicircle below and adjacent to the camera lens (Figure 11). In this retracted, or relaxed, position the filter arms are covered by the top of the solenoid housing assembly (Figure 12). Thus protected, the color filter assembly may be permanently installed in the camera cabinet. To make a color run, the operator need only load the camera with color film, place the NORMAL/COLOR switch in the COLOR position, and proceed as in a black and white run. Upon completion, a filter remaining in the select position is automatically reset when the operator manually advances the film prior to removing the take-up magazine from the camera.
The need for overstriking to provide sufficient exposure of the film has been eliminated by opening the camera lens aperture to f/3.5, by the use of new filters (Kodak Wratten filters 25 Red, 57 Green, and 47 Blue), and by the forced developing process used by the LASL Graphic Arts Group (see Section IV).
Further experimentation with filter selection mechanisms is planned to investigate the possibility of significantly reducing positioning time. It is also thought that film developing may be simplified by additional modifications to increase light transmission. These include a higher quality lens, removal of the pellicle assembly, and use of broader bandwidth filters. The feasibility of adding more filters, other than the basic colors, is also being considered.
On May 19, 1968, Group D-8 was requested to participate in a series of tests to photographically record in color, through appropriate color filters, an image displayed on a specially coated white phosphor cathode-ray tube of the Stromberg-Carlson 4020 Printer/Plotter. Kodak type EF daylight color film, a fast film with reasonably high resolution, was selected on the basis of manufacturers' published specifications and results of independent film testing previously performed by Group D-8. EF daylight color film has a resolving power rating of 80 lines/mm and a normal ASA speed value of 160; it is compatible with forced processing to achieve ASA speed values of 1250 and even higher.
For convenience in handling in a small hand-line processing system for pilot testing in the Group D-8 laboratory, type E-4 chemistry was substituted for the recommended ME-4 chemistry used in production-type continuous processing machines. Processing times for normal and one-, two-, and three-stop increases in film speed were established and found to agree quite well with results obtained from ME-4 chemistry in production machines.
Using the color separation filters Wratten filter numbers 29, 61, and 47B and a lens aperture of f/5.6 on the SC-4020 camera, a series of exposures versus processing times for the EF film were made. Normal processing required various numbers of overstrikes per color to achieve a suitable image.
Forced processing required fewer overstrikes. The goal, of course, was to achieve single-strike exposure capability. (A strike is defined for photographic purposes as any particular point in the CRT display where the electron beam, as directed by the computer, has energized the phosphor on the face of the tube to create a visible light emission representing a single bit of information which may be recorded by appropriate photographic means. Overstrike is to repeat one or more times the same pattern as the first.)
Because of the spectral emission of the CRT phosphor, spectral sensitivity of the color film, and transmissions of the color filters, a different number of strikes (or overstrikes) was required to properly expose individually the red, blue, and green records on the film.
The results of these tests are as follows:
Normal processing: 10 red, 7 blue, 3 green overstrikes (see Figure 1) 1-stop forcing: 5 red, 4 blue, 2 green overstrikes 2-stop forcing: 3 red, 2 blue, 1 green overstrike 3-stop forcing: 1 red (marginal), 1 blue, 1 green strike
It was decided that three-stop forcing, although possibly usable, was beginning to degrade both the image and the background with an undesirable loss in quality. Since the photographic color medium is subtractive in nature, it was decided to wash (i.e., to expose) the entire background of the display with a dark-blue light exposure. The computer-controlled image display was then exposed through the appropriate color filters to the film. This resulted in a dramatic recording of the desired color separations employed, provided that the film received a one-stop forced development.
Further tests also proved, however, that a background wash in any color of choice (see Figure 2) restricted the full rendition of color because of the subtractive nature of the film. Further investigation of the system suggested that a maximum aperture of the lens - f/3.5 instead of f/5.6 - and a change to filters of greater transmission but slightly broader in band width (Wratten filter numbers 25, 57, and 47) would permit single-strike exposures with no background wash to be acceptable with only a two-stop forced processing. This proved again to be exceptionally good (Figure 4, Figure 5, and Figure 8).
Realizing that some programs could use a limited range of color rendition, we determined that certain records displayed with an appropriate background wash could be exposed and processed with only a one-stop force in processing with extremely pleasing results (Figure 3 and Figure 7).
At the moment, it would appear that photographic requirements would be to record the computer-generated displays in color on a master film in a 16-mm motion picture mode, and supply one or more release type print films to be shown in motion picture projectors or to record a master 35-mm film for frame-by-frame projection analysis and subsequently to supply 8½ × 11-inch hard copy color prints of selected frames. Some alternative methods of color recording and reproduction have been investigated as follows:
Suggested additional considerations for trial with 35-mm EF film would be:
Evaluation shows that the best and most compatible system would involve, for 16-mm motion picture projection f11ms, recording on 16-mm EF film developed as a positive and forced two stops in speed with release printing done on reversal type color print film.
However, the best approach for hard-copy paper prints would be to record on 35-mm EF film force-processed for two stops as a positive and to make an enlarged color intermediate negative and then print (type C negative process} on a professional color paper.
Because an intermediate negative must be made for paper prints and because of developing costs, 8½ × 11-inch color paper prints are relatively expensive. However, as the number of prints from a single intermediate negative increases, the cost per print is drastically reduced.
Because this report represents a team effort on the part of LASL personnel, the authors would like to make special mention of the following contributions:
Sandia Laboratories Albuquerque and Clifford Fisk for their cooperation in our early efforts.
Roger B Beauchamp, Group C-1, and the SC-4020 operators for their special handling of color film.
Ann M Solem, Group C-2, for Figures 9 and 10; Arthur W Walker Jr, and A H Lochebay, Group C-2, for the special SC-4020 color routines.
Robert M Frank, Group C-4, for the new large character headings.
Thomas Gardiner, Group C-8, for assisting in design of color device, Daniel J. Torres and Freddie Salazar, Group C-8, for hardware implementation assistance on SC-l4020.
Billy R Claybrook, Group D-8, for f11m processing; G Norman Lindblom, Group D-8, and his staff for intermediate negatives and paper prints.
David S DeYoung and Carl R Shonk, Group J-10, for results which yielded some of the prints.
Barry K. Barnes, Group K-l, for Figure 8.
Richard L Morse and Clair W Nielson, Group P-18, for developing magnetohydrodynamics equations.
Many of the illustrations in this report were generated with a magnetohydrodynamics (MHD) code on the CDC 6600 computer. The MHD code uses a variation of the particle-in-cell (PIC) technique.
Consider a two-dimensional computing mesh in R and Z, fixed in space and divided into rectangular cells (see diagram). Particles, representing some type of fluid, are allowed to flow through the mesh of cells. The problem may be considered to be in two phases. Phase 1 computes the cell quantities and Phase 2 executes particle transport, with updating of some cell values if a particle crosses a cell boundary.
The object is to study the effect of magnetic fields (cell values) on fluid or plasma (particles). At the same time, the density profile of the fluid has a definite effect on the magnetic fields. Thus there is interaction. The magnetic field values are Aθ, Br, and Bz. The quantities carried with each particle are its velocity components and position. Other cell values are various sums that are needed.
Typical number of cells vary from 1,500 to 10,000. Typical number of particles vary from 13,000 to 1 million. More than one type of fluid can be present in the particles, and by varying the initial setup, different kinds of studies can be made.
Appendix B contains the schematic, logic, and timing drawings for the color filter device attached to the SC-4020. They are included to allow other organizations to install a similar device.
Figure B-1 is a simple block diagram to show the inter-relationship of the SC-4020 system and where the filter device fits in.
Figure B-2, the timing chart, illustrates how the system block relates to the elapsed time to select a color filter.
Figure B-3 is an electrical schematic showing what circuitry is needed to drive the filter solenoids.
Figure B-4 is the most important of the group. It shows the overall logic of the entire filter device. This diagram coupled with Figure 11 and Figure 12 in the main text would enable an installation to construct the device.
