Abstracts Category : Other

Applications for Machine Learning in Procedural Game Art Creation: Art Directing Algorithms to Create Two Dimensional Game Art with a Stylized Handmade Aesthetic

by Sara Wade

Works cited: pages 54-57 "This thesis explores the potential for deep learning to contribute to the creation of stylized game art in the game development pipeline. The game development pipeline often requires that hundreds of hours of skilled artists time be spent on the production aspect of game art. This thesis proposes automating portions of the game art production process. Machine learning, specifically deep learning and style transfer, has the potential to liberate the artist in the game development pipeline. With deep learning, the artist may focus on developing the visual style, rather than painstakingly applying a developed style, in order to create production quality game assets." First paragraph of author abstract. Keywords: art direction, automation, game design, game development, procedural art,artificial intelligence, machine learning, deep learning, algorithmic art, style transferAdvisors/Committee Members: CHAIR: Cookson, Aram, Khouri, Jacques, Leong, Josephine.