Brittle Fracture Shape Generation of Plane Objects by Conditional GAN

Abstract

Physics-based fracture simulation can generate realistic fracture patterns, but it is difficult to use in applications that require fast processing, such as games. To address this issue, a method has been proposed that uses a conditional adversarial generative network (cGAN) to predict fracture shapes, thereby speeding up processing. However, in previous research, the inability to deal with differences in physical quantities such as the velocity and mass of colliding objects was a problem. To solve this problem, this paper proposes a method in which a force condition map generated using a spring mass model is input as a condition.