Faces

While I was at Atari Research, I was fortunate to meet Susan Brennan, who had done impressive early work in the algorithmic representation of the human face. Susan's ideas greatly affected my thinking about the importance of the human face in all communications; I resolved to make best possible use of the human face in my future work. Unfortunately, the best I could do with the hardware and storage available was to depict static faces, simply blitting artwork onto the screen. It was nice, but not good enough. But with Guns & Butter, I saw an opportunity to apply some of Susan's ideas. For the diplomatic interactions, I realized that it would be useful to communicate the attitudes of the various characters toward each other. A perfect application of faces! I set to work. I started with a "base face," consisting of a set of basic features, each feature consisting of a set of points connected by lines. The first facial feature was its outline, a rough oval defined by about a dozen points. There were ears, nose, eyebrows, lips, eyes, cheeklines, eye lines, and so forth (see Figure 24.9).

24.9. Algorithmic faces.

What was special about all of this was that there were no bitmaps whatsoever; everything was drawn using lines, regions, and ovals. Now, by itself, a non-bitmapped face is, as you can see, not very good, but what made this system interesting was its algorithmic flexibility. Changing an expression on a face was a simple matter of re-mapping the face points according to a set of numbers defining the facial expressions. I was therefore able, in the highly constrained hardware environment of the times, to display eight different faces with seven different facial expressions. In 1988, when I designed this system, I was way ahead of the rest of the games industry, although the work at various research lab was much, much better. I continued to develop face depiction in later work.