GDC 2025 Session Viewer
Game AI Summit: Techniques for Visualizing the Output Diversity of Generative Systems
Oliver Withington (PhD Researcher, Intelligent Games and Games Intelligence CDT)
Pass Type: All Access Pass, Summits Pass - Get your pass now!
Track: Programming
Format: Lecture
Vault Recording: TBD
Audience Level: All
Content generation systems for games are common, but tuning them is notoriously hard. Designers want to ensure a level of output diversity, but getting a useful view of the full range of content producible from a system is challenging. Such systems can produce uncountably large numbers of unique artefacts, each of which has to satisfy numerous constraints and subjective aesthetic goals.
This challenge of understanding the output range of a generative system is important for game designers, but it has also been a key concern for another group for the past two decades: Game AI researchers. Researchers need to directly compare novel generative systems to prior ones, and this has led to the development of alternative methods for describing and visualizing output diversity. In this talk, we will discuss the strengths and weaknesses of these techniques and how they could be used in your future development.
Takeaway
In this talk, you will learn about the current state-of-the-art in visualizing the output ranges of content generation systems, and how these researcher focused approaches can be repurposed for use in developing your own games.
Intended Audience
The intended audience for this talk is game developers who have an interest in using content generation systems. No domain specific knowledge should be required, but an understanding of how content generation is typically used and the challenges it often faces would be helpful.
