An Author-Centric Approach to Procedural Content Generation

Authors

Abstract

This paper describes an alternative approach for videogame procedural content generation focused on providing authors direct control on what gameplay ensues from the generated content. An architecture is proposed that allows designers to dene, beforehand, target gameplay indicators, and then generates content for an existing base-design that achieves those same indicators in actual gameplay sessions with human players. Besides providing a description of this architecture, a trial intent on giving evidence of the approach's feasibility is presented. This experiment used an altered version of `Innite Mario Bros.' level generator, built to evolve design parameters so as to achieve 3 target gameplay indicators. Employing a Genetic Algorithm in generation of new parameter values, and using 25 players to test the end results, the platform was able to generate parameters that achieved, with precision, the values for those indicators. This result provides evidence of the approach's feasibility, hinting at its potential use for real-life design processes.

Keywords

Game Design, Procedural Content Genration

Subject

Design and Experimentation of an Author-Centric Approach to Procedural Content Generation

Conference

ACE 2013 the10th International Conference in Advances in Computer Entertainment Technology, November 2013

PDF File