Wang Changcheng uses simple scenarios from everyday life as the starting point for his creations. Through data set training and image stitching techniques, he reconstructs the “reflections” of the real world. In the creative process, he first selects a specific narrative and, based on it, collects relevant element data sets. Then, using Generative Adversarial Networks (GANs), he trains each of these elements individually. Finally, by stitching together images generated by the trained models, he creates entirely new images that are ambiguous, deceptive, yet strikingly similar to the characteristics of the original narrative.