Tangible and Autonomous Approaches for Virtual Film Storyboarding

Abstract

Films tell stories, entertain and emotionally engage their audiences. Creating film requires both time and money — resources which must be closely managed. The pre-production phase is a vital part of this management, both to minimise the risks of expensive changes to the filming schedule and to communicate the visual style of the film. A key part of this pre-production process is the storyboard which is the focus of this thesis. It is a document that allows efficient and creative shooting during production by representing each camera shot visually, similar to a comic-book. It is curated off-line, through multiple iterations, and although predominantly sketched on paper, is increasingly created using virtual, digital environments.

However, to produce a compelling storyboard, knowledge of creative visual grammar, and exploration of the scene are both required. The novice film-maker may lack experience of one or both these areas. Such skills gaps offer an opportunity to be exploited through novel ‘scouting assistants’ to provide appropriate image composition and therefore camera placements. The flexible virtual technology capabilities borrowed from the Virtual Production (VP) industry enable a suitable environment for such a goal. A related opportunity arises from the combination of cinematography with virtual environments, due to the incumbent keyboard and mouse peripherals not reflecting the physical attributes of real camerawork.

Concerned with improving efficiency and creativity in the storyboarding stage, this thesis concentrates on two topics. Firstly, it investigates how existing film content, when analysed, can guide an automated camera scouter to recommend new camera compositions in the virtual space. Secondly, inspired by contemporary trends in affordable VP techniques, it investigates a novel, tangible Human-Computer interaction (HCI) method for manual exploration and framing of virtual cameras.

This work has resulted in visual feature analysis of film, unsupervised clustering, and a Virtual Autonomous Cinematography Environment (VACE) in which to enact these ideas. It has also developed a tangible peripheral for virtual cinematography named Camera Chameleon (CC), with a rigorous user study to understand its impact. Combined, these contributions demonstrate methods to improve efficiency and creativity for time and experience limited cinematographers, with the advantages and disadvantages discussed.

Date of Award	25 Jun 2025
Original language	English
Awarding Institution	University of Bath
Supervisor	Wenbin Li (Supervisor), Tom Fincham Haines (Supervisor) & Yongliang Yang (Supervisor)