In 1963 DAC-1 computer aided design program is released. Begining in 1959, the General Motors Research Laboratories appointed a special research team to investigate the use of computers in designing automobiles. In 1960, IBM joined the project, producing the first commercially-available Computer Aided Design program, known as DAC-1. Out of that project came the IBM 2250 display terminal as well as many advances in computer timesharing and the use of a single processor by two or more terminals.

		The DAC-1 was unveiled in 1964 at the Joint Computer Conference, the same forum that Sutherland had used for Sketchpad a year earlier. Where Sketchpad's repertoire of curves was limited to arcs of circles, DAC-1 could reproduce the flowing lines favored by automobile stylists, even thought such curves follow no simple mathematical formulas. But because a car's shape must be drawn more precisely than a draftsman can work freehand, DAC-1 contained no provision like Sketchpad's for creating a design on the screen from scratch. In stead, the system required that a designer either explain the desired shape in a program and feed it into the computer or submit conventional engineering drawings to be read into memory with a digitizing camera.
Working demonstration of the DAC-1 graphic console with Cadillac.

Graphics - The display software performed the function of replacing points, lines, surfaces, and text by a series of blanked or unblanked vector endpoints that were assembled in an in-memory display buffer. If necessary, clipping computations were performed to limit the image to the defined size of the display window. After the vector end-point display list was formed, a channel program was started that simply looped on the contents of this display buffer, refreshing the image on the display screen of the graphic console. No work was done on hidden-line or hidden-surface removal simply because the user community was accustomed to looking at all the hidden construction lines on engineering drawings. Since all of the applications for DAC-1 were with stylists, product engineers, or tool engineers, this never became a problem. Years later, the technical illustration people became interested in this technology, at which time hidden-line and hidden-surface removal became a more important problem. With the drawn in the computer, DAC-1 became more interactive, employing an electronic stylus, much as sketchpad used a light pen, to permit the operator to enlarge the drawing and change portions of it. GM shied away from describing the system as being able to perform in real time, but in fact, DAC-1 reacted quickly.
		Frames from a 1964 demonstration film show the capabilities of DAC-1. With the touch of an electronic pencil, the designer of this Cadillac could enlarge a circles area and rotate the model.

		All graphic images in DAC-1 were created from points, lines, surfaces, and text. Since the system could display only a wireframe image, the representation for a surface was a series of section lines through the surface and the boundary lines of the surface. Remember that all of the geometric data were created in three dimensions. Therefore, a format was defined that contained the projection transformation (orthographic or perspective), the window location, and the font or line format information. A drawing was defined in terms of a display list containing names of formats, points, lines, surfaces, or collections of geometric elements. What was of more importance to the use of graphics could better be described under the category of human factors problems. Project personnel became concerned about this problem and did some early evaluation of user attitudes as a joint study with IBM human factors personnel.
1964 DAC-1 Scale expansion, rotation, and partial views in a design exercise.

Because most images contained too many vectors for the hardware refresh rate, there was a tendency toward flicker in the image. As a result, there were many concerns about the visual perception of the users. This became an important problem as soon as graphics terminal access became more readily available. What types of vision problems would be experienced by a user who sat at a graphics terminal for eight hours? Some of the solutions were obvious: improve the resolution, reduce the flicker, and eliminate background reflections. These still are valid concerns today, although visual images are much cleaner and color seems to have made a big impact on image legibility.

The other human factors issues related to the input devices such as the electronic pencil, the function keyboard, and a card reader. The only use that was ever made of the card reader was to process an accounting record when logging onto DAC-1. The function keyboard was more easily replaced by functional icons displayed on the screen. Alphanumeric input could have been more readily prepared on a normal typewriter keyboard, instead of the graphic console keys, which were awkwardly arranged in 6 x 6 alphabetic order. Finally, as was mentioned earlier, the electronic pencil was used only as a pointer to pick items or icons that were displayed. Sketching was not practical, and even the fact that the operator had to raise his arm to screen level to make an icon selection became objectionable to the users.
		DAC-1 graphic console from a brochure ad.

In 1964 at the Joint Computer Conference where DAC-1 was unveiled, Thurber Moffett recalled that people lined up two hours ahead of time to get in to see the DAC-1 demonstration. The technicians operating the machine "showed a structure rotating on the screen," Moffett said. "Nobody had ever seen that before." Around this time other companies, including Boeing Aerospace, IBM, McDonnell Douglas, General Electric, and Lockheed, developed similar CAD systems.

		Graphic console - During the development of DAC-1 there was so much preoccupation with reading and writing drawings that the graphic console (or operator’s console) almost became an afterthought. Little did the development team realize that interactive graphics consoles would become the dominating method of design and that the need to “read” drawings was of little consequence. In fact, it was probably the case that IBM supplied the console more for operator use than as a design terminal. To be sure, the graphic console did come equipped with both input and out-put graphics capabilities. The graphic console could display wireframe drawings or system messages. However, the complexity of the image was very limited by two factors: 1) the display buffer for the vector stroke image was held in main memory (which already was very limited) and 2) the vector drawing rate was such that the display would flicker for images with more than a few thousand vectors.
In November 1963, a full-size rear deck (trunk lid) for a prototype Cadillac design was created directly from mathematical data.

Frequently, the screen turned into a wheat field. The operator immediately knew the display buffer had been corrupted, since the vector display of computer instructions looked like random vectors. The graphic input medium was an electronic pencil whose position was sensed by a conductive coating applied to the screen of the display unit. However, by 1965 it became apparent that designers could not easily sketch on a vertical display screen, and the mode of sketching a design was quickly abandoned. The human factors of raising an arm to point or sketch on a vertical screen were very objectionable.
		Trunk lid prototype.

There also was an abortive attempt to use the electronic pencil to draw or print characters on the screen with automatic character recognition, but this technology was also abandoned because of human factors and problems with the recognition algorithms. In 1968 the development team had an opportunity to visit Douglas Engelbart at Stanford Research Institute, and everyone was very impressed with the graphic input possibilities of a mouse. Since graphic input at a console was not productive, how was the graphic console used? First, it served as an operator’s console for controlling the image-processing functions of the photo scanner/recorder. Second, the development team quickly realized that computer-aided design was really a series of geometric constructions that could be initiated and controlled from a graphics console. The console was used to input commands via the alphanumeric keyboard, and, of course, the console could be used to input parameters or select alternative operations via function button selection.

		Utility routines were developed to provide multiple choice, text entry, and query capabilities. In fact, the mode of operation was to program specific application-defined functions for each button on the keyboard. The keys themselves were fitted with a transparent overlay such that the labels could be changed for each new application by replacement of the overlay. One button might be assigned the function of computing the intersection of two lines. Depressing this button would cause a program to prompt the user to “pick” the two lines from the display from which to compute the intersection. Other buttons were assigned more generic functions such as Left, Right, Up, Down, or even Yes and No. Thus, computer-aided design became a process of selecting function button operations in a desired sequence and supplying parameter values for each function.
First production was 1965 Cadillac trunk lid.

Within a very short period of time, the development team realized that the buttons (today we call them icons) could just as easily be programmed to appear on the screen and be selected via the electronic pencil. Specific regions of the screen were dedicated to specific functions, as in a menu area or a graphics display area. So began the development of a menu- and window-based design environment.

3D Mathematics - As might be expected from the nature of the applications, the DAC-1 system made a substantial commitment to the development of new mathematical modeling techniques. Creation of a three-dimensional mathematical model of vehicle geometry was in the forefront of many minds. Mathematical operations to create and manipulate points, lines, and surfaces were a major part of the system. However, the two topics that received the most attention from the standpoint of mathematics were free-formed line and surface approximation. Early on, it had been assumed that most of the input to the system would come from drawings and from designer sketches. When both of these modes of input became less viable, conventional coordinate measurements taken from physical models became the starting point for the design activity.
		3D network of surface feature lines.

These coordinate measurements were defined in a Cartesian coordinate system and related to vehicle body position by an arbitrary origin and orientation. Physical models were digitized as a network of intersecting point-set space curves lying on the surface. The network of surface curves was chosen very carefully to capture surface “flow” lines or any apparent feature lines that represented surface discontinuities. The process of converting the network of point measurements to a single mathematical representation was then formulated in three phases.

		Photo scanner/recorder - The photo scanner/recorder was combined into a single subsystem and packaged in a single enclosure. The purposes for which the photo scanner was intended were perhaps the most ambitious portion of the joint project with IBM. The entire system was extremely complicated and, at one time, the IBM engineers estimated space utilization inside the scanner/recorder enclosure at over 300 percent. (Light paths for the scanner and recorder were sometimes reused three and four times by bouncing images off multiple mirrors.) The digital electronics represented state-of-the-art technology, and the performance specifications for the analog elements of the system pushed the very limits of available technology.
The photo scanner recorder console.

Very complicated lens optics were used to focus the light onto a 35-mm film image and subsequently onto light-detection circuitry. Special distortion-correction circuitry was added to the system to compensate for nonlinearity in the positional accuracy of the CRT spot.

Software was developed by GMR personnel to calibrate and dynamically compensate for both optical and electronic distortions within the field of the image. Mechanical failures would continually dog the hardware, which contributed to a high failure rate and abnormally low availability time. For example, the 35-mm film mechanisms frequently scratched the film and, instead of analyzing data, the program logic became confused by distortions in the field of view. Shortly after the hardware was moved to the GM Technical Center in Warren, Michigan, a supply line for moist air was left on all weekend. The temperature dropped, and on Monday morning the development team stood and watched IBM bail water out of the electronic enclosures. The purpose for which the photo scanner had been justified was to “read” engineering drawings.” As the project evolved, there was less and less need for this technology simply because engineering drawings were not needed as an input medium to the system.
		DAC-1 operating systems layout. The System configuration with the IBM 7090 computer.

However, even without the need to use drawings as input, the analysis and understanding of the information content of an engineering drawing are complicated problems. Even if one makes a lot of assumptions about standards such as labels, dimensions, title blocks, and so on, the topology of a drawing can be ambiguous and extremely difficult to understand. Fortunately, the CAD/CAM applications for which the DAC-1 system was intended found other methods for entering digital information into the system (IBM cards or mylar punch tape).

		It was also the ambition of the DAC-1 team to input three-dimensional measurements taken from models of vehicles. Structured lighting patterns were projected onto full-size clay models of concept vehicles. Alternatively, designers were requested to mark the clay models with tape or to scribe lines on the model to highlight the important “features” of the design. Two stereo photographs of the model were then transferred to 35-mm film, and continuous-tone images were digitized using the same photo scanner. Close-range photogrammetric measurement techniques were then used to correlate the two images to compute the three-dimensional coordinates of the feature lines on the surface of the model. Computationally, the techniques worked well, but the nonrepeatability, the unreliability of the scanner, and the distortions in the optics made the system unusable. Years later, other systems were built to perform the same function using glass plates and lasers to obtain much higher digitizing accuracy.
Left and right photo-grammetric stereo images for the scanning of 3D clay models data.

Termination of the DAC-1 project - In 1967 Edward Cole was GM’s president. He had observed the beginnings of computer-aided design at GMR. He made the observation that this was becoming an increasingly large undertaking. Therefore, he decreed that DAC-1 was no longer a research project and that the responsibility for further development and application of this technology should be transferred to the Manufacturing Development Staff. Accordingly, the DAC-1 development group at GMR was broken up, and key members were transferred to carry on this work under new management. Other members of the DAC-1 team transferred to other divisions of the corporation, including Marketing Staff and the GMR Mathematics Department. A smaller team of engineers and scientists, reporting to Edwin Jacks, remained without a project.

1994 The Origin of Computer Graphics within General Motors by FRED KRULL