Today's desktop graphics technology can soundly outperform a half-million dollar 3D graphics workstation of five years ago. However, this revolution in 3D performance has had only a modest impact on people's lives. One reason is that the average person does not create 3D content -- it's too difficult with existing tools. This might explain why the most noticeable impact of 3D graphics is entertainment: the game and movie industries can afford to hire trained experts to painstakingly create beautifully-detailed scenes.

Why should it be so hard to create 3D content? After all, many of us find it easy to sketch out a rough illustration of using a pencil or chalk board. As palmtop and tablet-PC devices are beginning to proliferate, we should be able to use such devices to sketch out our ideas at a coffee shop, the way we might use a napkin today. School teachers, architects, clothing and industrial designers, and story tellers should be able to easily create illustrations involving 3D shapes by somehow sketching with a computer.

In this talk I will describe our own efforts and those of others to make such applications possible. I will argue that a key enabling technology is "non-photorealistic rendering" (NPR). This relatively new field of computer graphics seeks to leverage principles that artists and illustrators have developed over many centuries for conveying information. I will survey NPR research, and describe some of the challenges and new directions for the field.