During the first third of the 20th century the cathode-ray tube (CRT), in the form of Braun’s oscilloscope, began to be widely used by electrical engineers to observe the waveforms produced by their circuits. Television had been slowly developing since the 19th century “electric telescope” and other devices which had all involved some sort of mechanical scanning system. Baird’s experiments in the UK, circa 1926, used a large disk with a spiral of holes in it, based on the Nipkow disk,1 [Fig.50 and Fig.51] which rotated before the scene to be televised. The mechanical scanning system had huge problems of resolution, frame rate and light level requirements [Fig.49], so electronic means of scanning an image were sought.
In 1927, one Philo T. Farnsworth, from Idaho in the US, began investigating an electronic television system, however it was a display device based on the CRT that became the first successful, fully electronic television receiver.
This approach, which became the television set that we are so familiar with today, was developed by the Russian born, American inventor working at RCA, Vladimir Zworykin, who in 1929 demonstrated his “Kinescope”[Fig.52]. As he later put it:
“The advantages of the kinescope in television over other means of reassembling the picture of the receiver are: the use of an inertialess beam, easily deflected and synchronized at speeds far greater than required for television; a sufficiently brilliant fluorescent spot which may be viewed directly on the end of the tube eliminating the restricted viewing angle usually present in mechanical scanners; noiseless operation; and the outstanding feature of the flexibility of the cathode ray tube itself.”2
It is only now with the development of LCD and Plasma flat-screen displays that these important features of the CRT have been outweighed.
The CRT was also used by the British, during WWII, to display images from their newly invented radar systems so that they could track German bombers in sorties over Britain. [Fig.54]
The very first computer generated electronic image was made on the Williams-Kilburn tube, developed for use in the Manchester University Mark I computer in the UK, immediately after the war (see the next section for the story). Modifications of this system were later employed in the Whirlwind system at Massachusetts Institute of Technology (MIT).3 The use of CRT’s in radar displays led to their adoption in the SAGE 4 air defence system, an outgrowth of the Whirlwind project, and this then led to the “vector” or “calligraphic” displays used in the first commercially produced computer graphic systems.
Computer imaging begins
The history of the memory storage systems used in computers is intimately tied up with the development of graphics displays and the kinds of graphics that could be displayed. By the early 1950s the need for computer graphics was immanent, particularly for scientific data visualisation. The computers to do the data reduction and the calculation of mathematical functions were being developed at the time (late 40s and early 50’s). The basis of the technologies to do computer graphics were also there but at this point were very primitive. The first computer-produced CRT-displayed images were made using oscilloscopes which mirrored the contents of memory (be it the Williams-Kilburn tube electrostatic store or the acoustic (mercury) delay line store) producing in effect a sort of raster (grid) display. One example is the oscilloscope used for program monitoring in the CSIRO MkI (also known as CSIRAC). Despite its having only a 20 dot by 16-line display, the potential was recognised and in 1957 fulfilled in the work of Dick Jenssen. We will discuss Jenssen and the use of the CSIRO MkI on another page. In the first computer conference held in Australia in 19515 several delegates mentioned the potential usefulness of graphic representations of the data produced by scientific calculations. For example the British mathematician who had done much to bring the computer work of the US to the attention of the British, and vice versa, Douglas Hartree in a discussion at the conference noted that
“difficulties arise in using digital machines when the input is irregular or discontinuous. Analogue machines are convenient when the results of the calculation are needed in graphical form. However it would be quite possible to develop some form of digital-to-analogue convertor to produce a curve on paper or on a cathode-ray tube, from data given in digital form.”6
This comment clearly points to the use of CRT displays as vector, or calligraphic, display devices that was adopted from the use of radar screens in MIT’s Whirlwind computer over the early 1950s. In that same discussion, Trevor Pearcey who had worked with Hartree in the UK and was, in 1951, building Australia’s first computer, commented (with reference to crystallography) that
“organic chemists require summations of triple fourier series for electron distribution of a molecule. They give a vast table of data and require results in graphical form, which can only be done by hand. Often in the data the signs are not known and test summations have to be made to find the signs. A fast method should be available. Cathode-ray tube representation of the output of a digital machine could be used but analogue methods seem more desirable.”7
At the next computer conference, at the Weapons Research Establishment (WRE) at Salisbury, South Australia in 1957, the crystallographer A.S Douglas, who was visiting from the UK mentioned displaying crystallographic results of calculations on a Williams-Kilburn tube. But he noted the difficulty of getting good straight lines if they were off-axis, and the difficulty of photographing vertical straight lines with the CRT camera, because they were too thin.8
Although, today, computer graphics are made and displayed on a vastly more sophisticated version of the original Williams-Kilburn tube “bit-map” display, the real development of graphics took place on vector type displays that Hartree foreshadowed. In these devices a list of what are, effectively, electron-beam moves are converted from their digital representation to an analogue form to then drive the deflection coils moving the beam as per the instruction list. The notion of this “display list” is behind the first consistent, well described, computer graphics drawing package, Sketchpad, which was specifically designed for interaction between the operator and the computer. Ivan Sutherland developed it at MIT’s Lincoln Laboratory in 1963 as his Ph.D. thesis: Sketchpad: A Man-Machine Graphical Communications System.9 The Sketchpad system utilized a “light pen” (a product of the SAGE project10), with which one could interact with the graphical display, manipulating images to control their forms. It also allowed graphic software to develop around the concept of the drawing primitive: lines, rectangles, polygons, arcs and circles, etc., which could be linked in sequences to produce a drawing. Sutherland gave computer graphics a consistent basis and many large corporations4 began to use computer graphics to design products and to simulate the design’s behaviour in data visualisation. Although graphics are now generally displayed on raster displays the approach that Sutherland took has continued into modern day graphic design forming the approach taken in Computer-Aided Design (CAD) systems, Illustrator and Draw packages or Flash graphics today.
The Display Monitor
In order to understand the importance of the display technology in constraining the development of computer imaging we need to follow the process of its incorporation as a display system used in computing in general. Thus, we now need to look in somewhat more detail at the electronic visual display, based on the CRT, since by the mid 1960s it becomes of primary importance in the display of electronic images. Basically there are three stages in the development of display devices that depend on the CRT. We will start with
1 The electro-static storage display because it was the first specifically computer applied CRT device, being a memory system, and as such forms the basis for the display of bit-mapped images from numbers stored in a computer memory.
We will then follow on to the two classes of display on which computer graphics were substantively developed. These are:
2 Vector type displays such as the oscilloscope, the radar display and early computer displays, and
3 Raster type displays such as the television screen and modern computer monitors.
We have covered the initial development of the Cathode Ray Tube (CRT) in Chapter 4 section 2, but we should now detail its construction. A CRT is a glass vacuum tube, roughly in the shape of a bottle. At one end (the neck of the bottle) is a piece of metal filament (the cathode) that can be heated by an electric current, thereby producing a stream of electrons. The interior surface of the bottle is coated with a conductive material called aquadag and serves as the anode towards which the electrons are accelerated by the high voltage between it and the cathode. The interior surface of the base (the face-plate) of this bottle is covered with a material (a phosphor) which produces light (photons) when stimulated by a focussed electron beam. Two specially placed pairs of magnets focus the beam. If the two pairs of magnets are driven by a changing voltage the focussed beam can be driven around the inside of the tube and will appear to paint a moving line or curve onto the face-plate, which can now be seen to serve as a screen.11 [Fig.15] This is the basis of the oscilloscope, the electro-static storage tube, the vector display, and the video monitor.
1: Electro-static storage tubes – an early form of volatile electronic memory.
Immediately after WWII,12 between 1946 and mid-1948, a new form of computer memory, an electro-static storage tube based on the CRT, was developed in Britain. In 1946, Fred Williams, while working at the Telecommunications Research Establishment (TRE) had discovered that he could store a single “bit” on a CRT screen using the secondary electrons produced in the phosphor by the CRT’s electron beam. Tom Kilburn then took this single electron spot technique and began to work out a way of generating multiple spots in an array across the tube faceplate which allowed up to 1024 dots to be independently turned on or off as a form of memory. In 1947, Williams was appointed as head of the Electrical Engineering department at Manchester University and took the development system and Kilburn with him. While there Kilburn finished the 1024-bit store, for which the Manchester Small Scale Experimental Machine (SSEM, or the “Baby”) served as a test-bed. For a little more on, and a picture of, the SSEM see Chapter_4_4]
Let’s have a look at how it worked. A series of pulses, a controlled distance apart, intensify the electron beam; long pulses (dashes) indicating a “1” and short pulses (dots) indicating a “0”. By placing a metal screen over the face of the CRT some of the secondary electrons could be absorbed and their current amplified, thus reading the tube. For the tube to remember the states that had been applied to it the memory location had to be regularly refreshed via a recirculation circuit that used the amplified signal from the “read” screen to trigger the initial electron beam intensifying pulses to lengthen for the duration of a dash if the location stored a 1, or to simply pulse for the dot period if a 0. If a 1 was to be written to a location that was currently a 0 then a pulse from the computer lengthened the beam intensifying pulse and if a 0 was to be written to a location that currently held a 1 the pulse lengthening was inhibited by the signal from the computer.13 The clue to understanding this is to recognise that if the recirculation were cut off the tube would be continually refreshed with all 0s by the beam intensifying pulses. By knowing the position of the electron beam on the screen through the timings of the beam pulse waveform and the vertical sweep waveform applied to the CRT’’s deflection plates – which effectively acted as the memory address counter, a memory location could be marked with either an active (“1”) or an inactive (“0”) state. [Figs.16 & 17] The regeneration process meant that, at the appropriate moment – as the “address counter” reached the particular location of interest, the signal could be tapped and placed on the data buss (then known as the “Digit Trunk”) of the computer for use as a value in calculation, or displayed on another oscilloscope so that the operator could monitor the state of memory and thus the progress of the program in the computer.
The Williams-Kilburn tube, as it should have been known,13 was not only a form of computer memory but, since the waveforms that governed where the memory locations would be, it formed what was, in essence, a television raster. By judicious switching on and off of locations in the memory, ie, by storing the right numbers in the right locations, it could, coincidentally, provide bit-mapped graphics in the same device. Thus was the first bit-mapped image produced, as Kilburn demonstrated in his progress report to the TRE in December 1947. The images of Fig.18 show that the storage was also extended to 2048 bits, and are the first known digitally generated images, which, by the way, also possesses that other major function of digital images, public relations, as their motivation.14 The SSEM evolved into the Manchester Mark I and the Ferranti Mark I and later machines developed by Ferranti (eg, the Atlas). Meanwhile the news of the Williams-Kilburn tube type of storage had reached the US and John von Neumann’s group at the Institute of Advanced Studies (IAS) at Princeton, New Jersey.
They sent Julian Bigelow to Manchester in June 1948. Subsequently, Williams visited the US in 1949 and the tubes were incorporated into the IAS machine and some of its derivatives including the IBM 701 and of particular importance for us, the ILLIAC15 of which the SILLIAC was a copy. (see Chapter 4 section 4) On machines that used these tubes for memory, such as the ILLIAC series, a separate Williams-Kilburn tube, without the pick-up plate in front of it, was run in parallel with the storage version so that the operator could keep track of the program’s progress in the machine.
2: Vector displays are those in which the beam is drawn around the screen of a CRT by varying electronic voltages (ie, waveforms) applied to its deflection plates. This is the basis of an oscilloscope. When driven by, say, an audio oscillator these waveforms can produce the kind of display known as a “Lissajous figure”, and this is how the ABC TV logo was first made.
In an oscilloscope the two pairs of magnetic plates are placed at right angles to each other so that one pair when driven with a voltage deflects the beam of electrons in the vertical (y-axis) while the other also when driven with a voltage deflects the beam in the horizontal (x-axis). [see Fig.15] Thus if a waveform, say a sine-wave tone, is applied to the vertical deflection plates and a sweep signal (a triangle wave whose period is consistent with the time it would take to sweep the beam across the tube face) is applied to the horizontal deflection plates a waving line showing the shape of the vibration of the sine-wave over time will be painted onto the screen and will remain there for the duration of the persistence of the particular phosphor used in the tube. Supposing one also applies a varying waveform (not the sweep waveform) to the horizontal deflection plates then Lissajous figures will be displayed.
The work of American artist Ben Laposky, in the 1950s, provides a paradigm example of work based on this possibility.16 The Lissajous figure is also used in the electronics lab to test oscillators against a master oscillator when tuning them for example. It was this use of Lissajous figures that led the members of Optronic Kinetics to the Lissajous in making visual displays with a TV picture tube modified to receive the outputs of their theremin in an installation work at the Fine Arts Workshop of the University of Sydney in 1969. See Chapter 5 section ?.
Not only can the oscilloscope be used to monitor electronic waveforms in real-time as they vary across its inputs, but it can also be used to display images from a computer. There are two versions of this approach. The first was the use of an oscilloscope as a way of displaying the contents of the Williams-Kilburn tube memory device and the second was by directly driving the deflection plates with electronic waveforms generated from the computer output, turning the beam on when a line was to be drawn and off when the beam was to be relocated for the start of a new line. This latter is known as the vector display and became the prime means of generating computer graphics in the 1960s and 70s.
The vector display also forms the basis of a class of computer graphics displays in which strings of numbers from the computer are translated, by a digital-to-analogue converter, into voltages which drive the deflection coils used to control the position of the beam, and thus the image on the display. The computer only has to produce the data about the line that it is drawing at any one moment which meant that the vector display allowed the computers of the day, being quite slow compared to today’s machines, to display, reasonably quickly, the results of its calculations as (moving) images on a screen. [Fig.19] The drawback of the vector display is that it can only draw lines; filled areas and useful variations in brightness are not really possible. Vector displays, while not the first CRT type computer display – oscilloscopes have that distinction, were the first to draw with sufficient resolution to produce detailed images. We will see this in some of the work of Stan Ostoja-Kotkowski (see Chapter 5 section 2) as well as much of the work produced by Doug Richardson on his system. (See Chapter 6.) We also saw them in use in some early video games such as “Asteroids”.
3: Raster displays drive the deflection coils with a specific waveform which sweeps the electron beam from side-to-side, and then progressively down the whole surface of the screen. The image is produced by varying the current on the cathode of the CRT, thus modulating the brightness of the electron beam and thereby the scanned image. By this means grey-scale images and images in which areas are filled can be drawn. [Fig.20]
The use of the raster monitor (the TV) was seen as a means of getting cheap display devices but depended for any substantial use on large amounts of memory and really had to wait until such memory was cheap enough to use in this way. In imaging with the raster monitor the image is stored in memory as a “bit-map” which represents the sequence of values of the brightness of the image at every point (pixel or picture element) in the horizontal line being drawn in the raster. Because of this, and the problems that memory access was slow compared to the display time of a frame of video and the cost of memory was very high, raster displays did not become computer monitors until the early to mid 1970s. A computer video display could, in the early days, be 256 (28) pixels across by 256 (28) lines down and more than one bit per pixel for a shaded image so they took a lot of memory, in this example 216 or 65,536 bits, which was more memory than some computers had for their programs and working data. We now generally know raster displays as the ubiquitous television and computer monitor. Their initial use in the making of artworks was as video monitors in the production of video art.
As has been mentioned above the raster display is a grid and in computer memory it is represented, notionally, in bytes (locations) of memory, each representing a point on the display, ie, one byte per pixel (in colour there are three bytes used per pixel, one each for the red, green and blue channels of the image). Sequential locations then represent sequential positions in a line of video so that the image appears as a map in the memory more or less in the same relationships as it appears on the screen. This is what is called the ‘bit-map’. Now, as you will remember, I have referred to this before. It is the way that the Jacquard mechanism is organised on the weaving loom and this is why many people claim that the Jacquard loom is the first “computer” graphic device. see Chapter x, section xx. The grid approach also turns up in the Williams-Kilburn electrostatic memory storage device and very interestingly the grid turns up in painting as a line of development in modernist art which we shall come to when we look at the work of David Smith and Optronic Kinetics, chapter X, section xx.
Computer graphics can also be recorded to film. However the images still have to be presented to a monitor for checking at least. The first computer animation recorded to film was made by Edward Zajac at Bell Telephone Labs in 1963. He simulated the motion of a satellite in space on the computer and then filmed it frame-by-frame off the computer display. This approach was also taken by Ken Knowlton and his collaborators, such as Stan Vanderbeek and Lillian Schwartz also at Bell Labs. Others in the US include John Whitney who collaborated with Jack Citron at IBM in New York state, from 1966, and Charles Csuri, 1967, who was working at Ohio State University.17
In Australia the first recorded sequences of computer graphics to film as animations were made by Doug Richardson in 1968. We will follow up his story in Chapter X section XX. However he too used a film camera controlled by the computer to take frame-by-frame exposures of a computer screen.
What I have discussed in this section forms an important aspect of our digital heritage, portions of which you will find in the many museums of science technology throughout the world. I will now return to the development of computing machines over the 20th century.
1 Regarding the modifications to the Williams-Kilburn tubes see, Goldstine, Herman H., (1972) The Computer from Pascal to von Neumann, Princeton, New Jersy: Princeton University Press, pp.309-10. Regarding the Whirlwind see Kidwell, Peggy A. and Ceruzzi, Paul. E. (1994) Landmarks in Digital Computing, Washington, DC: Smithsonian Institution Press, chapter on The Whirlwind, p.68. See also < https://web.archive.org/web/20070313233738/https://www.mitre.org/about/photo_archives/whirlwind_photo.html > for a collection of photographs.
2 Lee, J.A.N. (ed.) (1983) Annals of the History of Computing: SAGE Special Issue, vol.5, no.4, Oct. 1983. SAGE used a very sophisticated display system based on radar screens and a “light gun”, the precursor to the “light pen”. There is also a paper on Whirlwind (II ?) with a photo including graphical displays.
See also <https://web.archive.org/web/20050915075615/http://www.mitre.org:80/about/photo_archives/sage_photo.html > for a collection of photographs including an image of an operator using the “light gun” at the display console.
3 Zworykin, V.K. (1936a) “Description of an Experimental Television System and the Kinescope” in Television: Collected Addresses and Papers on the Future of the New Art and its Recent Technical Developments, vol.1, 1936, New York: RCA Institutes Technical Press, p.158.
4 See Maloff, I.G. and Epstein, D.W. (1938) Electron Optics in Television, New York: McGraw-Hill, p.34, for a brief description of Nipkow’s mechanical scanning system. See also Schoenherr, Steven E.(2004) History of Television, at http://www.tvhandbook.com/History/History_TV.htm,
5 PCACM Proceedings of a Conference on Automatic Computing Machines held at the CSIRO Division of Radiophysics in the grounds of the University of Sydney, 1951.
6 Hartree in PCACM, 1951, p.55.
7 Pearcey in discussion in ibid, p.56.
8 Douglas, 1957, pp.119-1 – 119-11.
9 Sutherland, 1963.
10 Lee, 1983.
11 Beginning with Boeing and General Motors.
12 One description of the technology as it was being applied in the early days of TV is in Maloff, 1935, p.337.
13 Most of the mathematicians and engineers who had been working on various secret projects, especially cryptography and radar in Britain and in ballistics in the US, were released immediately after the war and although they were not allowed to talk about it, used much of their wartime experience to set up computing departments in many of the Universities in the UK and the US.
14 Williams and Kilburn, 1949.
15 It did tend to be called the Williams Tube by the Americans who used it a lot (and see the SILLIAC story).
16 Kilburn, 1947. Progress report to TRE issued 1st December 1947.
17 Goldstine, 1972.
18 see Franke, 1971, p.60ff. and Laposky, 1976, pp.21-2.
19 see Franke, 1971, pp.93-97.
Return to Contents