Researchers increasingly use Visual Novels — narrative games combining dialogue, choices, and simple 2D visuals — to study behaviour and support education. Many lack the technical background to build a game themselves, and typically rely on specialised systems built for their needs. However, these systems are costly to build, and are hardly reusable and adaptable in other projects. This thesis is motivated by a case where a researcher tried to reuse Propellant — a visual editor for creating police training scenarios, built in a previous student project — to create VNs about decision-making in drug-trafficking scenarios, but the system's architecture made it prohibitively difficult to extend and overcome the limitation of its domain-specific design.
This thesis investigates general-purpose game engines, such as Unity and Godot, as a more sustainable foundation for these VN creation tools instead. Extending general-purpose game engines with VN features is an approach that is currently underexplored in research. The research questions investigated are:
- How does the engine's general-purpose architecture shape the design and implementation of VN-related editor and gameplay features?
- How does the general-purpose editor environment influence the VN authoring experience?
First, a feature analysis was performed on purpose-built editors, dedicated VN engines, and VN extensions for general-purpose engines, to understand the state of these tools. It showed that VN extensions support the essential features found in VN engines, and that alternative player input features, such as numeric responses, represent a gap across all surveyed tools.
Based on this analysis, a proof-of-concept solution was produced by extending Dialogic, a free and open-source VN extension for the Godot game engine, with two new features: numeric input and interactive scenes ("hotspots"). The overall design of the solution shows that Godot's scene-based architecture provides a flexible foundation for implementing extensions. A usability evaluation session with the researcher behind the motivating case study suggests the tool feels like a coherent authoring environment, though some confusion with the scene system arose from their prior familiarity with a different game engine.
Accounting for the limited scope of this thesis, these findings offer a preliminary but positive answer to the suitability of general-purpose game engines as a foundation for sustainable VN editors in research contexts.
Link to the Godot project template, including Dialogic with the two implemented features: https://github.com/zemiguel20/Thesis-Project-Template
Thesis PDF