Towards a Generic Method of Evaluating Game Levels

Abstract

This paper addresses the problem of evaluating the quality of game levels across different games and even genres, which is of key importance for making procedural content generation and assisted game design tools more generally applicable. Three game design patterns are identified for having high generality while being easily quantifiable: area control, exploration and balance. Formulas for measuring the extent to which a level includes these concepts are proposed, and evaluation functions are derived for levels in two different game genres: multiplayer strategy game maps and single-player roguelike dungeons. To illustrate the impact of these evaluation functions, and the similarity of impact across domains, sets of levels for each function are generated using a constrained genetic algorithm. The proposed measures can easily be extended to other game genres.

Refining the Paradigm of Sketching in AI-Based Level Design

“…adding more walls) based on the new needs of a refined level. Finally, the case studies have demonstrated how even simple evaluations (such as exploration and safety) of Liapis, Yannakakis, and Togelius (2013c) can be adapted and reused with alternate pathfinding methods such as the hex-based paths of Endless Legend maps or the multi-floor paths of FPS levels, with stairs for moving up and open-air tiles for dropping down.…”

“…weapons for a shooter game, resources for a strategy game) via passable paths; for such constraints, the infeasible population of the FI-2pop GA minimizes the number of disconnected paths. Similarly, most case studies use the heuristics of Liapis, Yannakakis, and Togelius (2013c) to evaluate feasible levels. In short, these heuristics consist of:…”

Map Sketch Generation as a Service

“…Map sketches have been used in Sentient Sketchbook (Liapis, Yannakakis, and Togelius 2013a) as the target of a humanmachine co-creative process (Yannakakis, Liapis, and Alexopoulos 2014), where map sketches are generated by the computer as alternatives to human designs. Several computationally lightweight fitness functions have been defined by Liapis, Yannakakis, and Togelius (2013b) for evaluating human designs and as objectives for genetic algorithms. Map sketches represent game levels with playability constraints; these constraints are satisfied through a constrained evolutionary optimization algorithm described below.…”

“…Map sketches must fulfill certain criteria in order to be minimally playable, and must also have certain qualities in order to be useful in a game. In earlier work (Liapis, Yannakakis, and Togelius 2013b), the core playability criterion for map sketches was that all level features can be reached, i.e. there should be a path between all tiles with gameplay properties ("special" tiles).…”

“…On the other hand, exploration is evaluated based on a flood fill algorithm starting from one special tile and stopping when another special tile has been reached; the exploration score averages the exploration effort from one special tile to all other special tiles of this type. More information on the map sketch evaluations can be found in previous publications (Liapis, Yannakakis, and Togelius 2013b).…”