Dissertation
Understanding and designing for control in camera operation
Vorschaubild nicht verfügbar
Volltext URI
Dokumententyp
Text/Dissertation
Zusatzinformation
Datum
2019
Autor:innen
Zeitschriftentitel
ISSN der Zeitschrift
Bandtitel
Quelle
Verlag
Ludwig-Maximilians-Universität München, Fakultät für Mathematik, Informatik und Statistik
Zusammenfassung
Cinematographers often use supportive tools to craft desired camera moves. Recent technological advances added new tools to the palette such as gimbals, drones or robots. The combination of motor-driven actuation, computer vision and machine learning in such systems also rendered new interaction techniques possible. In particular, a content-based interaction style was introduced in addition to the established axis-based style. On the one hand, content-based cocreation between humans and automated systems made it easier to reach high level goals. On the other hand however, the increased use of automation also introduced negative side effects. Creatives usually want to feel in control during executing the camera motion and in the end as the authors of the recorded shots. While automation can assist experts or enable novices, it unfortunately also takes away desired control from operators. Thus, if we want to support cinematographers with new tools and interaction techniques the following question arises: How should we design interfaces for camera motion control that, despite being increasingly automated, provide cinematographers with an experience of control?
Camera control has been studied for decades, especially in virtual environments. Applying content-based interaction to physical environments opens up new design opportunities but also faces, less researched, domain-specific challenges. To suit the needs of cinematographers, designs need to be crafted with care. In particular, they must adapt to constraints of recordings on location. This makes an interplay with established practices essential. Previous work has mainly focused on a technology-centered understanding of camera travel which consequently influenced the design of camera control systems. In contrast, this thesis, contributes to the understanding of the motives of cinematographers, how they operate on set and provides a user-centered foundation informing cinematography specific research and design.
The contribution of this thesis is threefold: First, we present ethnographic studies on expert users and their shooting practices on location. These studies highlight the challenges of introducing automation to a creative task (assistance vs feeling in control). Second, we report on a domain specific prototyping toolkit for in-situ deployment. The toolkit provides open source software for low cost replication enabling the exploration of design alternatives. To better inform design decisions, we further introduce an evaluation framework for estimating the resulting quality and sense of control. By extending established methodologies with a recent neuroscientific technique, it provides data on explicit as well as implicit levels and is designed to be applicable to other domains of HCI. Third, we present evaluations of designs based on our toolkit and framework. We explored a dynamic interplay of manual control with various degrees of automation. Further, we examined different content-based interaction styles. Here, occlusion due to graphical elements was found and addressed by exploring visual reduction strategies and mid-air gestures. Our studies demonstrate that high degrees of quality and sense of control are achievable with our tools that also support creativity and established practices.