There has been a long-standing need in certain computer graphics applications for human figure simulations, that as descriptions of the human body are both accurate and at the same time adaptable to different user environment.This article describes a progression of figure simulations development by the author over the last 20 years an attempt to meet this need. Emphasizing graphic design as the proper direction for technical developnemts the work chronicled below was accomplished a computer graphics system framework in witch the human figure was, but a single element of the overall research effort. The figures accompanying of human figures the four stages of this progression of human figures simulation. A number of the simulations and applications were firsts at the time meeting their original objekctives and simulating other researchers, who often commanded greater technical resources.

"First Man" and "Second Man"
In 1959, the author's most fundamental work leading to human figure simulation was the development of computer perspective animation a project proposed to and authorized by Verne L. Hudson, chief of preliminary design at the Wichita Division of the Boeing Airplane Company.2lt was Hudson who probably coined the term computer graphics.The first human figure simulated by means of a com- puter may have been the "Landing Signal Officer" illustrated in Figure I.

LSO was a fIXed database we developed to give the location and scale of the landing signal officer on the CVA-19 class aircraft carrier for a motion picture that simulated cockpit visibility during landing.3 Viewed from the flight path, the figure was a 12-point silhouette with the lines representing two- dimensional edges only. For queueing theory studies, however, 30 straight-line connected points defined the figure in a block form, and the lines represented the rightangle block edges. These figures were moved down the aisles of simulated commercial aircraft to predict passenger movement. This first stage culminated in the production of the "First Man," shown in Figure 2. This model was composed of seven movable segments that could be articulated at the pelvis, neck, shoulders, and elbows to approximate various pilot motions. Anthropometric data already encoded in the original figure, which represented the average-sized American male (50th percentile), was digitized and mathematically transformed to smaller or larger figures. The lines now represented surface traces of minimum radiian illustrator's economical technique of letting a single line represent the likely demarcation of shadow.

Through an agreement with the Boeing Company and Computer Graphics, Inc., the author used his First Man data for a 30-second television commercial, illustrated in Figure 3, at Graphcomp Sciences Corporation for Norel. co. Modifications made to the figure database for this purpose included approximately sound-synchrono!1s lip motion as the figure spoke. Possibly the first commercial to make use of computer graphics perspective, it achieved high prime-time ratings. While simpler in appearance, the "Second Man" series-representing the second stage of this work-improved the articulation of the First Man figure by incorporating a greater number of joints and continuing to develop its anthropometric accuracy. Film sequences based on a photo series by Muybridge4 demonstrated animation of the 19-joint figure. Tests of animated motions included operating an aircraft control column, running, and high jumping, as seen in Figure 4.
 
"Third Man and Woman"
The third model-"Third Man and Woman"-which was developed at Southern lIIinois University at Carbon- dale's Department of Design, represented an initial attempt at a hierarchical figure series, with successive figures separated by an order of magnitude in complexity. Prior experience had shown that human figures at dif-ferent scale had useful applications. We developed a progression of databases using one-point figures for demographic distribution, 10-point figures for queueing studies, and l00-point figures for anthropometrics (see Figure 5). Points located the figures' center of gravity for one-point figures only. In the 10-point figures, a single line represented an entire limb, head, or torso. The l00-point figures were 50th percentile representations based on Humanscale 1/2/3, a book of anthropometry for designers.S They contained 10 movable body elements and could be animated the lines represented edges of a boxlike definition of each body element. The l00-point figures were used with databases of other designed objects in display exercises where figures seated themselves in automobiles, walked in architectural environments, or rode the escalator to a monorail station,6 as in Figure 6. To gain more detail, we needed to be able to measure points on the body surface more precisely.' Existing photogrammetric plots of human figures had considerable asymmetry, so the author redrew them to obtain a symmetrical and aligned l000-point figure that would simplify motion. The lines now represented contours con- necting points on the surface equidistant from a plane within the coordinate system (see Figure 7). As we developed and reviewed the more detailed, l000-point figures, we realized that ultimately not only surface but also internal elements would have to be de- fined. Using the same baseline perspective animation pro- gram with a slightly different hierarchy of elements, we digitized a simplified skeletal structure of the hand and animated the fingers. Here, as shown in Figure 8, the lines represented edges of a blocklike description of the bones. We had successfully achieved the progression of detail we sought with the one-, 10-, 100-, and l000-point figure series, but in each system the lines represented different elements. This variance compelled us to look for a. more; consistent presentation of the figures, despite changes in ) their level of complexity.

"Fourth Man and Woman"

This research continued in 1977 in the newly formed Southern Illinois Research Institute, where the author decided to access biostereometric tapes-a more accurate and manageable starting pointS for the fourth stage of his work in human figure simulation. At the Bellevue, Wash- ington, office called Siroco, we defined horizontal con- tours and began searching for a program to construct polygons between these contours in addition to varying the numbers of contours for figures at different levels of complexity. In writing software to restructure data from the biostereometric tapes, we developed a program that con- tained options for selecting contours, selecting desired numbers of points along the contours, and dividing the points evenly along each contour.9 A range of levels of detail was then possible. Decisions on the number of polygons were based on factors such as line weight, image size. and-particularly-viewing distance. Thus, we can generate a progression of figures that appear to have about the same mass of lines for plotters and about the same polygon size for raster displays even when the a, complexity per figure varies considerably (see Figur)
The "Fourth Man and Woman" line plots develop Siroco have been useful in testing geometry and explc the visual effects of hemispheric projection to-an art research originally demonstrated by the author in 1960's with the display of information on the inner face of a hemisphere. Hemispheric projection totally merses the viewer in a visual environment for perspective scenes or other information displays. The databases in preparing Fourth Man and Woman plots demonst correct geometry when properly projected inside a hemisphere, even though a slide in its original flat form pears distorted, as in Figure 10. In demonstrating the appearance of the Fourth man and Woman on raster displays, surface normals of e. polygon are calculated to determine surface visibili amount of hue, and the highlights necessary to render figures with shading and color. The Fourth Man a Woman figures prior to error corrections and smooth, are shown Qn the title page of this article. The facetsoft polygons can be smoothed with the additional calculated of averaged normals for each polygon vertex (see Figure 11). While the levels of detail selected are still a result of the designer's or programmer's experience, a planned feature in new software is the automatic selection of contours and polygons based on viewing distance, physical screen size, available raster resolution, and any storage limitations of the system. Integration of this work with motion systems by other researchers is also planned. 12 Both the First Man and Second Man models achieved line definition and motion at Boeing, an aerospace environment ideat for this type of research because both the technical resources and the application to demonstrable needs are closely linked and visible to participants. On the other hand, in the SIU-C academic environment the opportunity existed to explore hierarchical databases and surface treatment strategies. And at SIRI/Siroco, in a research institute environment where some advantages of both other environments exist-raster
scan, color, shading, and highlights are now part of the Fourth Man and Woman series

Acknowledgments

At Boeing, designers and programmers for the projects were Jack Gahman, Kenneth Frank, Robert Fee, James Berry, Robert Tingley, Michael WeIland, Franc Wilson, and Elmer Smith. Computing help also came from James Herrington, John Scheffler, Leroy Allen, Walter Bern- hart, Tino Lazzaretti, and John Freyman. At Graphcomp, Robert Tingley and Charlene Sawyer assisted on the TV commercial for La Roche, McCaffrey, and McCall. At SIU-C, student designer/programmers were Albert Anen, Frank Crow, Richard Rovens, John Virruso, Larry Evans, Lewis Wright, Dennis Andrews, Todd Nickle, Frank Juzwik, and Frank Stemm. Southern Illinois University at Carbondale Research and the Na- tional Science Foundation provided support, and the biostereometric tapes were available courtesy of Robin Herron. At SIRI/Siroco, interns were Russell Vaughn, Stanton Drake, Takeshi Ogawa, Craig Wittenberg~ and James Wallace. Steve Williams of Lawrence Livermore~tional Laboratory produced the color images of the Siroco data using the Hilite program by Frank Crow. NSF and private grants supported the research. Also, I gratefully acknowledge the thoughtful comments and suggestions offered me by Norman Badler and his referees during my preparation of this article.

References
1. W. D. Bernhart and W. A. Fetter, Planar Illustration Method and Apparatus, US Patent No. 3,519,997, filed Nov. 13, 1961, issued July 7,1970.
2. W. A. Fetter, "Computer Graphics, Aircraft Applications:' Document No. D3-424-I, Boeing Airplane Company, Wichita Division, 1961.
3. W. A. Fetter, Computer Graphics in Communication, McGraw-Hili, New York, 1964, p. 52
4. E. Muybridge, The Human Figure in Motion, Dover Publications, New York, 1955.
5. N. Diffrient, A. R. Tilley, and J. C. Bardagjy, Human- scale 1/213, MIT Press, Cambridge, Mass., 1974.
6. W. A. Fetter, "A Computer Graphics Human Figure System Applicable to Transportation," Transportation Research Record 657, Applications to Interactive Graphics, Transportation Research Board, Commission on Sociotechnical Systems, National Research Council, Jan. 1977, pp. 20-23.
7. W. A. Fetter, "A Computer Graphics Human Figure System Applicable to Kineseology," ACM Special Intemt Group on Design Automation Newsletter, Vol.
8, No.2 of 3 (late issue), June 1978, pp. 3-7.
8. W. A. Fetter, "A Computer Graphics Human Figure System Application of Biostereometrics," Proc. NATO Symp.. Applications of Human Biostereometrics, SPIE, Bellingham, Wash., 1978.
9. W. A. Fetter, "A Computer Graphics Human Figure System Applicable to Bio-stereornetrics," CAD J. Fourth Int'l Con/. and Exhibition on C;omputers in Engineering and Building Design, IDC Science and Technology Press, Guildford,SurT!Y,England,1980,coverandpp.175-179.
10. W. A. Fetter, ','Wide Angle Displays for Tactical Situa- tions," Proc. US Army Third Computer Graphics Work- shop, Virginia Beach, Va., Apr. 1981, pp. 99-103. II. Bui-Tuong Phong, "Illumination for Computer Genera- ted Images," Comm. ACM, Vol. 18, June 1975, pp. 311-317.
12. N. I. Badler and S. W. Smoliar, "Digital Representations of Human Movement:' Computing Surveys, Vol. II, No. I, Mar. 1979, pp. 19-38.

William A. Fetter has been the research director for Siroco (the Bellevue, Washing- ton, office of Southern Illinois Research Institute US) since 1977. In 1970, he began conducting research and teaching in the Department of Design, Southern Illinois University at Carbondale, and served for two years as department chairman. Prior to that, starting in 1959, he pioneered in the initiation, organization, and application of computer graphics at the Boeing Company. His research interests emphasize innovation in computer graphics from the viewpoint of graphic design requirements as a basic direction for technical developments. Fetter received a BA in graphic design from the University of Illinois at Urbana and has pursued subsequent studies in design. He is an associate fellow of the AIAA and a member of SCS and SID.