The object of this paper is to attempt to predict future developments in hardware and software related to all forms of computer graphics. This prediction may be successful for two reasons. The path of development may be so clearly defined that prediction is merely the extension of this path or the ideas presented in the paper may be sufficiently powerful that the paper influences the future and therefore verifies its predictions. These predictions and speculations can however fail for many reasons. Consider attempting to predict the future of art at the time Rembrandt was alive, would the works and style of Picasso or Dali have sprung to mind and if they did would one have been prepared to present such ideas.
So many different graphics systems, methods and specialized application areas exist today, that it is difficult to make a sensible projection for the future. In what direction will computer graphics develop, and in what direction should it develop?
In order to be able to answer this, it is necessary to go back to the essential aspects of computer graphics. This paper analyzes these and presents the basic ideas of computer graphics in a generalized way. Its purpose is to stimulate thought and discussion on future development, rather than to give concrete suggestions although a few conclusions are drawn.
Line drawing systems, interactive displays, text setting, animation, picture processing, pattern recognition, they all seem to be very different in both methods they use and application areas they are used for. However, they all have one thing in common: they deal with pictures.
Computer Graphics is a set of all the means to deal with pictures that can be seen by man and/or machine.
Man sees a picture when he becomes aware of a stimulus in a certain part of his brain {visual perception). He then analyses and interprets it.
It is known that interpretation of pictures depends on cultural environment and education. Also it is not always possible for man to analyse a certain picture: it might be blurred, or too complex. These are two reasons why also machines must be able to see pictures.
A machine sees a picture when it can analyze it and abstract from it features necessary for the task involved. In addition a machine can store a picture, this, man usually cannot do.
Many different types of pictures exist in a wide range of complexity. They can be static or dynamic (with continuous or stepwise motion). Perhaps the most important distinction is according to their purpose and complexity:
Pictures must somehow be input, specified, stored, manipulated, transmitted, output, etc. So far pictures cannot be treated as a whole. Thus, a picture must be decomposed into parts, such that these parts can be treated separately.
Picture creation is done by an output device. Logically it consists of two parts: the picture compiler and the actual display. The picture compiler receives input data defining the picture (including dynamics) and produces from this a generalised display list, from which the actual display device can build the picture.
The picture compiler also can receive local control input to affect overall picture generation. This local input is only for manipulating the picture as a whole {like turning over an object in ones hands, or moving through a holographic image), and has nothing to do with the actual definition stage of the picture.
Existing graphical devices have both picture compiler and display in varying amounts of complexity. Plotters receive pen-increments, displays receive vector commands, or even transformation commands.
Starting with pen and paper moving relative to each other under computer control one has the digital incremental plotter in all its forms and accuracies. These units are becoming faster in operation and the inherent mechanical problems of control and overshoot, acceleration against inertias and inkflow to the pen are increasing. The pen has in some devices been replaced by light or electron beams and the paper by some photosensitive material or the drum of a XEROX type copier.
At first sight the natural extensions appear to be in the direction of bigger and faster and perhaps cheaper. With the increase in the use of raster scan C.R.T. displays there will be an increasing need for hard copy units attached to such devices. It is thus envisaged that the XEROX type copier will be developed for this purpose in order to produce large high quality colour prints.
If this development trend results, what will become of ink jet type displays? A future is seen for these devices in the area of commercial art. Thus sizes will be increased. Research into the chemistry of the inks could well result in the ability to produce textured surfaces and the possibility of creating instant old masters! These types of display units could also become mass production devices by becoming substitutes for litho techniques for producing relatively small quantities of high quality output picture material.
An extension to these concepts could be the display of pictures by eroding multilayered material. Consider the use of a simple type of scraper board having a black top surface, white secondary surface followed by red, blue and green surfaces. A particular coloured line would be produced dependant upon the depth of a scribing tool. Much effort has been devoted to producing the illusion of three dimensional form using two dimensional displays. Perhaps the culmination of this effort is the work at the University of Utah, where the viewer who wears a special head-set, is presented at each eye with an image from a small C.R.T. By detecting the viewer's position the images are updated and the viewer can stroll around a virtual image.
Such a tool is ideal for research purposes and could play a vital role in psycho-analysis in that effects upon the mind could be created without recourse to narcotics. It is however not the sort of tool that would be used by a company board meeting discussing a new style of automobile.
One solution to this problem is to use a simple on-line machine tool and cut forms in plastic foam, chalk or other suitable media. Such a device is a hard copy unit and does not provide interaction, but it does provide a starting point for speculation. Surface production for a specific class of surfaces could be obtained by having a matrix of say, 1m × 1m which consists of rods of 0.5 mm diameter at 1 mm centres, with the possibility that the whole could be covered by an elastic membrane. Thus by pushing up the rods, representation of a surface can be obtained. This device could also be used in the dynamic sense to simulate vibration of surfaces and three dimensional wave motion. Another possibility for three dimensional display is the use of an electro-chromatic gel in the form of a block with electrodes attached on two sides. Surfaces would appear as surfaces of colour within the block. Interaction could be by a thin wand inserted into the block. Withdrawal of the wand causing the gel to close up. Such a concept is within the realms of possibility in that the Kerr cell uses electro-chromatic effects to produce a shutter for ultra-high speed cameras.
Another exciting possibility is the development of computer generated holograms. If such images are to be produced in virtually real time, new techniques of processor design will be required and parallel processing through the use of banks of microprocessors will probably be employed.
The ability to incorporate movement and interaction into computer displays came with the first refresh displays, and in essence these displays have altered little in their fundamentals.
Features that were implemented on a software basis are now incorporated as hardware and the introduction of microprocessors will ensure that such moves will continue as part of the development process. Colour is achieved with these displays using either phosphor layers of different colours which are excited by electron beams of varying intensity or shadow masks as in a colour television receiver. In order to give displayed output body there is a natural move toward raster scan systems and it is envisaged that the addition of video recording to such displays will be a natural development. The coupling of these two devices would provide a more flexible system for producing animation.
The rapid playback and editing facilities accorded by video recording of colour raster scan displays should produce a large impact on the area of film making. Another possibility would be the development of faster CRT that make it possible to draw 50-100,000 vectors at a 60-HZ refresh rate. This would give the possibility of drawing and manipulating reasonable complex pictures or making real-time computer animation.
High speed video recording and play back to other displays could be also used to provide a different level of multiplexing on systems having many users.
The related developments with Computer Output on Microfilm will be dependent upon developments in film technology with the present speed of processing colour film being a holding factor.
New applications of the physical properties of matter could be used in new displays. Examples of this could be large area liquid crystal displays and displays constructed from a matrix of micro light emitting diodes. Finally one should not preclude the possibility of direct interaction with the human brain.
To summarise the ideas concerning picture display.
Display devices will probably never be really cheap. One always wants the picture to be better, or to be produced faster.
A picture can be produced by:
The latter possibility seems to have enormous potential, but will probably not be feasible in the near future. Existing devices are all of the first category. They either create a 2D or 3D object that is illuminated by ambient light (plotter, NC-machine), or they create the light rays themselves (CRT-screen, microfilm, hologram).
The specification of a picture consists of primitives and of statements as to how these are interrelated, transformed, etc. including the dynamic aspects.
Usually pictures are specified literally by assembling them from primitives. Conceptually it is much easier, however, to define a picture top-down, though suitable instruments do not seem to exist at present.
The primitives used depend on the type of graphics system involved: a line drawing, a surface drawing, or a body drawing system.
Most systems today are line drawing systems, and the other two are usually simulated by line drawing systems with more or less good results.
When a picture is input as such, it is also decomposed into primitives, e.g. digitizer, 3D measuring tool, raster scan input. Information essential to the task at hand is retained, though some information is lost in the process.
Input devices generate commands to control a process, such as the definition of a picture (indirectly by modifying a database). Input devices are logical devices. Their physical implementation might take various forms. They have nothing to do with output devices, though they might share some hardware or a communication link.
The command inlet to a process (either interactive or not) is an input stream (a file). Commands are structured or not, and are accompanied by data in various forms: text, numbers, arrays, etc. Commands are generated on a lower level by input tools.
Input tools - again an abstraction can be classified according to the type of data they deliver:-
There has been an unfortunate preoccupation with the light pen as an input device and this may account for the relatively slow development of input devices compared with displays. The mouse or tracking ball or other related potentiometer-activating devices have found uses related to specific displays. The tablet in its various forms is however giving the user the natural freedom experienced with pencil and paper. A tablet is ideal for drawing-type self expression but is not the complete answer for all forms of interaction. The touch wire device originally developed for rapid interaction in air traffic control systems has an obvious place in normal input work for interacting with display menus. An extension to this is a proximity switch keyboard for the input of alpha-numeric data. Thus input devices attached to a display will be matched to specific functions rather than attempt to produce an all embracing single device. Having established a more than possible base for growth there are two major input areas that require more elegant solutions than are available at present. These are the rapid input of mass drawing data and the input of data related to three dimensional shapes, it is speculated that the work of J. Radon that resulted in tomography could be developed such that an object is placed upon a turn table, multiple projections taken and converted such that a complete three dimensional representation is loaded to computer store. The combined use of light and X-rays could provide for the collection of internal as well as external object data.
If a representation of an object is to be obtained from an engineering drawing drawn in an orthographic projection the problems are not trivial. They will be solved and drawing scanners having this capability will become common place.
Because of the volume of data and speed of processing required, these types of input units will have their own computing systems based upon microcomputers.
To summarise the ideas concerning input
The use of raster scan or related systems in reconstruction with picture processing, scene analysis and 3-D reconstruction will become one of the most important input devices to graphic systems in the future.
Techniques will be developed for computer recognition and refining of sloppy and incomplete drawings and models without the user having to be more explicit or categorical than he would be in communication with one of his colleagues.
Simple, user-friendly, adaptive command languages will be developed. They will make it possible to adapt the guidance, the commands and the error handling to the individual user's experience, skill and habits. Analogous to normal computer peripherals, graphical devices will become as self-contained as possible. They will have their own (more advanced) picture compiler. This will probably be a microprocessor, preferably programmable (local or remote) to accept picture descriptions of different standards.
Looking at these aspects from the viewpoint of a general user, there will be required a vast effort to produce graphics systems in which all of this related processing is transparent.
The object of this paper is to provide a stimulus so that in any subsequent consideration of standards relating to graphics there is sufficient elasticity or open endedness to deal with the future.
The predictions made have not allowed for some discovery to be made in chemistry, physics or human biology. In the final analysis, the end processor in any graphics system is a human and it must be admitted that some work taking this into account is in providing stereoscopic output or depth modulation of displayed pictures. Picture is a universal language of communication and any means developed to produce pictures must also possess this attribute of universality.
