“Virtual Reality (VR)” has been widely used in interactive exhibitions, and the application of AI video generation also provides a blue ocean for video creation. This study will discuss the cognitive behavior of painter Ms. Sandy Lee through flat painting display (2D Display) or head-mounted display virtual reality (HMD VR). Her paintings are appreciated through Ai images to generate dynamic videos, using PAD emotions. Scale, the test found that no matter which device is used, there is not much difference between video reading and the expression of emotions in the video. Due to the operation and screen range, viewers think that the HMD VR studied in this study is not a satisfactory tool for viewing. In addition, visual elements such as AI-generated videos need to be designed more carefully to allow viewers to have a better perception process. However, we still have to improve the syntax design of Ai video generation and the reading and interaction methods in VR environments, so that the dynamic video display of virtual reality Ai image generation can be as impressive as the physical two-dimensional painting display tour.