2D Sprite Sheet Generator: uma ferramenta para geração dinâmica de animações de personagens utilizando Stable Diffusion

Abstract

The process of creating 2D character animations using conventional drawing methodologies usually takes a considerable amount of time, taking into account that each frame of the character needs to be created manually in modeling and animation software. In this context, the objective of this work is to present the 2D Sprite Sheet Generator, a tool for automatically generating dynamic character animations. To this end, Stable Diffusion was used, a deep learning model that uses text input and generates images with great detail and adaptability. Depending on user input and possible details, the tool generates a spritesheet of a 2D character as well as a sample GIF of the animation, taking less than three minutes. From a pilot study with seven professionals in the gaming area, satisfactory results were demonstrated in the use of the tool, showing excellent acceptance and good success in implementing its functionality. The main advantage of the approach is the innovation and potential of the solution, given that the market lacks tools that use artificial intelligence to facilitate the generation of 2D characters. The proposed approach addresses a critical need in the industry, streamlining the animation creation process and offering a practical solution for developers and artists.