StyleGAN is a type of generative adversarial network, which consists of pitting a generator and a discriminator against one another. The discriminator learns to differentiate the real images of the trainingset from the fakes generated by the generator, while the generator learns to generate more accurate images to fool the discriminator.
I trained the model on two different datasets: Yurukyara 1, which consists of around 1200 single yuru-characters and Yurukara 2 which includes both single and multiple mascots, adding up to around 1600 images.