A long-standing puzzle in perception is the question: why is it so easy for us to understand shape in line drawings, even though they do not correspond to any real-world percept? Past theories have been unsatisfactory, for example, hypothesizing that line drawings are a culturally-specific learned language, or that line drawings “trick” V1 into treating lines as step edges at object contours. I propose a new answer: line drawings do, in fact, roughly correspond to realistic depiction of a specific, plausible combination of shape, lighting, and materials, though in a combination not normally seen in nature. This theory is based on past computer graphics research in line drawing algorithms, by myself and others. I will describe a few possible variants and shortcomings in this theory. I am looking for your feedback, for possible publication in a perception venue.