Abstract
Interactive digital storytelling promises to be a new artistic discipline. But what does the artwork look like? And how much control does the artist have over the final result? By searching for the right answers, we draw several approaches from diverse fields such as filmmaking, game design, autonomous agents, psychology, and narrative intelligence. We first give a definition of interactive storytelling that includes a scope between authorship and emergent narrative, and present two example projects with different emphases. Then, we introduce a multi-level concept for an experimental stage that can be used to explore interactive storytelling at varying degrees of flexibility versus predetermination. Instead of inventing the ultimate virtual narrator, we suggest developing layers of run-time engines that allow authors to work with directions on each single layer separately-from selecting high-level dramatic structures down to directing animated actors. The concept of underlying models at each level is explained in detail by the example of a story model implementation. Further, we show variations of level adjustments causing graded degrees of semi-autonomy. First results of this experimental stage are presented, and the primary future tasks are pointed out.