The Informedia Project has developed and evaluated surrogates, summary interfaces, and visualizations for accessing a digital video library containing thousands of documents and terabytes of data. This chapter begins with a review of Informedia surrogates for a single video document, including titles, storyboards, and skims. Incorporating textual elements, considering user context and emphasizing phrases over words have all led to better video surrogates. These lessons have driven the development of direct manipulation interfaces and visualization strategies for exploring news video libraries. Summarization strategies employed in the Informedia Project for sets of broadcast news video documents are discussed, concluding with the research challenges in building visual interfaces from vast quantities of image data and imperfect automatic video processing.