Level generation based on model transformations: Tutorial at Game On 2015 Conference

Daniël Karavolos, Anders Bouwer

Research output: Contribution to conferenceAbstractAcademic



The goal of the worksop/tutorial is to introduce participants to the fundamentals of Procedural Content Generation (PCG) based on generative grammars, have them experience an example of such a system first-hand, and discuss the potential of this approach for various areas of procedural content generation for games. The principles and examples are based on Ludoscope, a software tool developed at the HvA by Dr. Joris Dormans, e.a.

Duration: 2 hours


We will use the first 30 minutes to explain the basics of how to use generative grammars to generate levels. The principles of these grammars and model transformations will be demonstrated by means of the level generation system of Spelunky, which we have modeled in Ludoscope.

Spelunky focuses solely on the generation of geometry, but grammar-based systems can also be used to transform more abstract concepts of level design into level geometry. In the next hour, the participants will be able to get some hands-on experience with Ludoscope. The assignment will be to generate a Mario-like level based on specific requirements, adapted to the interests of workshop participants.

Finally, we are interested in the participants’ evaluation of this approach to PCG. We will use the last 20 minutes to discuss alternative techniques, and possible applications to other areas of PCG, like asset creation, scripting and game generation.

Workshop participants are asked to bring a (PC) laptop to work on during the workshop, and are encouraged to work in pairs.
Original languageEnglish
Number of pages1
Publication statusPublished - 3 Dec 2015
EventGAME ON' 2015: 16th International Conference on Intelligent Games and Simulation - University of Amsterdam, Amsterdam, Netherlands
Duration: 2 Dec 20154 Dec 2015
Conference number: 16


ConferenceGAME ON' 2015
Abbreviated titleGAME ON' 2015
Internet address


Dive into the research topics of 'Level generation based on model transformations: Tutorial at Game On 2015 Conference'. Together they form a unique fingerprint.

Cite this