This user study is being undertaken by
Marvin Zammit as part of his Ph.D. dissertation under the supervision of Prof.
Antonios Liapis, focusing on Procedural Content Generation for digital games. The study is being carried out at the Institute of Digital Games at the University of Malta.
About CrawLLM
CrawLLM is a video game created for this study, consisting of a top-down 2D dungeon crawler with card-based combat mechanics. Large Language Models (LLMs) are used to generate different themes and visual styles for the game, and subsequently generate prompts to also generate the assets used in the games, such as the sprites used for the dungeon walls, floors, and doors, the character spritesheets, as well as the images used for the card-based combat sections. This pipeline is used to change the theme of the game completely, without changing its rules.
The survey
In this study, you will be asked to assess some assets which have been generated automatically. Please read each question carefully and select the response which you believe is most appropriate. The questions will be in relation to a set of generated themes for the game, through which we strive to understand how coherent and clearly relevant to the primary theme context are the generated assets. The survey should take between 10 to 15 minutes to complete.
The survey is
completely anonymous, and no personal or contact information will be collected. The data shall be exclusively used for research purposes, and shall be stored by the researcher only. Summarised results from the analysed responses may be published in academic publications. You may withdraw from the survey at any point you like by simply closing this browser window, without any consequence or repercussion. If you have any questions or concerns, you may contact directly the researcher
Marvin Zammit (
marvin.zammit@um.edu.mt) for clarifications. If you agree to participate in this study, please confirm your consent below, then press the Next button to start the survey.