Image to level?
Sai Teja Konda · 2021
Procedurally generated content (PCG) is becoming widespread in different domains one of which is game design. In game design specifically, PCG allows designers to relieve the burden of content creation, allowing them to focus on creating meaningful narratives, etc, in the games. Many tools exist where they can procedurally generate levels, aesthetics for levels. However, it is not clear to what extent some of these tools generate playable levels. In this thesis work, I consider the procedural generation algorithm featured by ImageToLevel: Generation and Repair and scrutinize this algorithm to understand whether these kinds of algorithms create or not only generate content but also generate coherent content that can be used effectively as playable levels in 2D platformer games. I do so by generating levels using ImageToLevel: Generation and Repair and let automatic play testers play each of these levels. The results would show if levels were playable. I discuss possible directions and possible improvements of this system based on other algorithmic approaches that account for basic criteria that constitute a playable level.--Author's abstract