Queen Mary University of London
Available for placement
Ollie Withington is a researcher working on novel methods for visualising and comparing the output spaces of generative systems in the world of video-games. Following three years working in software development for a healthcare management firm, he embarked upon an MSc in Computing and Information systems in 2018 to boost his professional skill-set.
He fell in love with the world of AI research through the work started during his final thesis and this, combined with a lifelong love of playing video-games and thinking about their design led to joining the IGGI programme in 2020. He lives in London with his wife and young daughter and when he is not writing about, thinking about, or playing games he can often be found either in the local bouldering gym, or in the park with his family (especially on the rare days when London graces us with sunshine).
A description of Ollie's research:
While Ollie is interested in almost all aspects of game research and development, he is especially interested in procedural content generation systems and whether they can be made more consistently helpful for designers and interesting to end users. His research is currently focused on developing novel methods for visualising the possible outputs from content generation systems.
While automatic content generation as a field is thriving, there has been comparatively little work done on novel methods for comparing generators to each other, something he aims to remedy. Ollie's work is focused on using novel data reduction and projection approaches to produce information-rich 2D visualisations of the output spaces of generative systems, allowing for easy assessment of output diversity and system character. The hope is that this work could help game designers and researchers alike make more informed choices about which generators are optimal for their purposes and how to optimise them, as well as making the outputs of these systems more interesting to players. His latest paper 'Compressing and Comparing the Generative Spaces of Procedural Content Generators' was presented at the IEEE Conference on Games last autumn.