Searching for good and diverse game levels

Abstract

Abstract-In procedural content generation, one is often interested in generating a large number of artifacts that are not only of high quality but also diverse, in terms of gameplay, visual impression or some other criterion. We investigate several search-based approaches to creating good and diverse game content, in particular approaches based on evolution strategies with or without diversity preservation mechanisms, novelty search and random search. The content domain is game levels, more precisely map sketches for strategy games, which are meant to be used as suggestions in the Sentient Sketchbook design tool. Several diversity metrics are possible for this type of content: we investigate tile-based, objective-based and visual impression distance. We find that evolution with diversity preservation mechanisms can produce both good and diverse content, but only when using appropriate distance measures. Reversely, we can draw conclusions about the suitability of these distance measures for the domain from the comparison of diversity preserving versus blind restart evolutionary algorithms.

Constrained surprise search for content generation

“…However, it should be noted that the number of feasible individuals in the population may not be sufficiently diverse, and thus a smaller set of the most diverse results would be more appropriate both for in-game use and as a performance metric. In previous work [3], this set of "solutions" was discovered via k-medoids where k was a property specified by the designer. In this paper, such solutions are obtained by DBSCAN [18] which can return a variable number of clusters (and their medoids) depending on the distribution of data.…”

“…Often, the simplest solution is to assign a minimal fitness score and kill off the infeasible individual [3]. In highly constrained spaces, however, this is not a desirable strategy as most genotypical information is lost [11].…”

“…While the majority of PCG research focuses on creating one final artifact which exemplifies the desired properties of its type, there are several attempts at creating a diverse set of content using e.g. multi-objective optimization [2], multimodal optimization [3] and novelty search [4]. This paper is inspired by earlier work on creating sets of diverse artifacts [5], and applies a recently introduced divergent search algorithm, namely surprise search [6], [7], on the task of procedural content generation.…”

A Progressive Approach to Content Generation

“…Here, an evolutionary algorithm or other stochastic global search algorithm is used to search content space for content that optimally satisfies some evaluation function -for example, map balance or level winnability. Evolutionary algorithms have excellent facilities for generating sets of diverse content artefacts, even while satisfying multiple fitness functions [8]. However, search-based approaches are still in general much slower than constructive approaches, often too slow to be used in real time.…”

A Spatially-Structured PCG Method for Content Diversity in a Physics-Based Simulation Game

“…Moreover, it is not enough to simply automatically generate a great number of elements as one might be more interested in creating components that are both diverse and of high quality [7].…”