Prioritization strategies for video storyboard keyframes

This exploratory study investigates one type of video surrogate, storyboards, in terms of their ability to summarize and communicate the themes of arts-related videos. An HTML interface containing the storyboards, videos, and instructions was developed and run in a standard browser. Three phases—consisting of storyboard evaluation, full-length video evaluation, and their comparison—were completed by each user for three different videos. The data were analyzed for issues relating to keywords, summaries, and recognition of visual style for both the storyboards and the full-length videos. The linear sequence and narrative structure of storyboards are questioned, and a three-tiered model is proposed. The first layer consists of keyframes representing the “entity” and “action” of the videoʹs central theme, the second layer consists of “entity” and “action” keyframes with regard to background or supporting information, and the third layer is composed of keyframes representing attributes, locations, and time periods. This structure facilitates the identification of appropriate keyframes for storyboards, eliminating redundant or peripheral images, and improves the storyboardʹs ability to communicate the essential message of videos. The tiered model is motivated and supported by the user study as well as current research on video surrogates and classical indexing theory.