The Metaphorical Camera: Automation and Embodiment in Visual Narrative

From visualizations to movies to video games, a camera is often an indispensable tool for structuring narratives given a set of input data. Today, an ever-increasing number of images are gathered by public, private, benevolent, and/or malicious agents, autonomous or controlled, and it is humans who must parse them in order to turn raw data into knowledge. However, the size of these datasets renders them intractable to understanding without a level of analysis and automation: either in what to capture, or in how to capture it, or in how to present/re-present it.;Automation is increasingly ubiquitous, and it can work in various ways, ranging from adjusting small parameters in systems under human supervision, to human-in-the-loop systems, to so-called "fully automated" systems. How to best design and implement an automatic control system is extremely context-specific, especially with regard to how to integrate human decision-making and awareness into the system.;This dissertation examines the process of automation of real and virtual cameras, drawing on insights from artificial intelligence, robotics, narrative theory, and interactive systems design, and presents two contributions: an analysis of automation in camera systems, and a prototype software tool for virtual camera control. Informed by the content of Edward Branigan's Projecting a Camera, the software provides a proof-of-concept for generating automated visual narratives through camera control in virtual simulations.