The third article-series of GAN in computer vision - we encounter some of the most advanced training concepts such as Wasserstein distance, adopt a game theory aspect in the training of GAN, and study the incremental/progressive generative training to reach a megapixel resolution.