Maximizing the science return on the Event Horizon Telescope project requires fitting models for spatially resolved black hole images to the data. These images can be calculated from accretion and jet theory, but theoretical uncertainties lead to systematic errors in the predicted images. In many cases, however, the images are dominated by the combined effects of Doppler beaming and light bending, leading to a characteristic “crescent” shape. I will discuss a geometric crescent model for black hole images based on these effects. The crescent outperforms other simple geometric shapes in the description of current data, can be used to simulate future EHT observations, and has implications for detecting a black hole shadow.