This article has been published at the Dagstuhl Seminar 15051 "Artificial and Computational Intelligence in Games: Integration". The original publication, along with its bibtex entry and other information can be found here.
Creativity facet orchestration aims to combine generation across the multiple creative domains that comprise game design. Literature identifies six core facets existent in games: level design, game design, audio, visuals, narrative and gameplay (involving NPCs or not) [1]. The first two facets are necessary for a game to be instantiated whereas the remaining four are optional. While there have been a few attempts to integrate more than one facet during the generative process (e.g. Game-o-Matic [2], Angelina [3]) these have been limited to mere hierarchical (linear) procedures. It is only very recently that research on computational game creativity has focused on ways in which more that two facets are interweaved during their generation (such as the work of Hoover et al. [4] orchestrating visuals, audio and gameplay elements).
We argue that to generate novel and valuable games that lie on unexplored regions of the game design space, an orchestration approach is needed to automate game generation in a truly integrated manner. We view this approach as an iterative refining process particularly suited for the generation of playable prototypes for designers to consider and get inspired from. Orchestration requires that the human designer specifies the desired semantics for a query within a (large but manageable) space of possible games. For example, a designer might request a horror game (directly affecting the mechanics of the game), with open-space-style levels (affecting level design), with a warm ambiance (affecting visuals), relaxing music (affecting audio), linear narrative and aggressive NPCs. The generative system blends available concepts (using e.g. ConceptNet [5]) with the aim to deviate from the query in the game design space (e.g. this process could involve searching for novel games from semantically annotated databases of existing games). At this point each facet operates independently, constrained by the semantic information the designer has provided. The facet-specific generator operates using the semantics of all generators, thus providing a high-level context to guide its generative processes. The result is a set of prototypical games with unconventional combinations of facets' outputs, all matching the same underlying semantics. Those games are then presented for designer consideration, along with information about their distance (dissimilarity across several dimensions) to typical games, in order to choose which are to be refined. Refinement tailors parameters of the game space and polishes facets such as visuals and audio.
[1] Antonios Liapis, Georgios N. Yannakakis, Julian Togelius. Computational Game Creativity. In Proceedings of the Fifth International Conference on Computational Creativity, 2014.
[2] Michael Treanor, Bryan Blackford, Michael Mateas, Ian Bogost. Game-o-matic: Generating videogames that represent ideas. In Procedural Content Generation Workshop at the Foundations of Digital Games Conference, 2012.
[3] Michael Cook, Simon Colton, Azalea Raad, Jeremy Gow. Mechanic miner: Reflection-driven game mechanic discovery and level design . In Proceedings of Applications of Evolutionary Computation, volume 7835, LNCS, 284-293, 2013.
[4] Amy K. Hoover, William Cachia, Antonios Liapis, Georgios N. Yannakakis. AudioInSpace: A Proof-of-Concept Exploring the Creative Fusion of Generative Audio, Visuals and Gameplay. In Proceedings of Evolutionary and Biologically Inspired Music, Sound, Art and Design (EvoMusArt), 2015.
[5] Hugo Liu, Push Singh. ConceptNet - A Practical Commonsense Reasoning Tool-Kit. BT Technology Journal 22, 4, 211-226, 2004.
This article has been published at the Dagstuhl Seminar 15051 "Artificial and Computational Intelligence in Games: Integration". The original publication, along with its bibtex entry and other information can be found here.