O modelo conceitual de redes adversárias generativas na construção de assets 2D para artistas de games: um estudo de caso utilizando ativos artísticos do Axie Infinity

Abstract

The games market has experienced a vertiginous growth in recent years, making room for new game companies to create more and more custom games. This fact has generated a great demand for visual assets for game artists. This dissertation analyzes the potential of the conceptual model of generative adversarial networks in the construction of 2D assets for game studio artists. To achieve the result, a survey was carried out using the inductive scientific method and the case study, as well as direct observation of the phenomenon to generate inferences. An interview with artists served as a preliminary assessment of the model. It was discovered that the conceptual model can create 2D assets through an adversarial process that synthesizes images with real features, from a database created by an artist or developer. This means that this network model can be used by artists as a tool to create 2D assets or be used as a support for creating diversified 2D assets. Thus, after this case study using artistic assets from the game Axie Infinity, we found that the model is promising to automate the production of artists in studios.