This study uses an end-to-end stereo parallax estimation algorithm using deep learning to construct a mechanism to simulate human eye vision in a virtual reality (VR) environment. By applying the hierarchical detail processing (LOD) algorithm related to human vision, the reconstruction of the virtual reality scene is realized to make it closer to the actual observation effect of the human eye. This technology is widely used in various artistic creations, providing the audience an immersive experience. In practical application, the audience experience of the VR opera “The Drunken Princess” received high ratings of more than 4 points in all dimensions. Among them, the audience’s recognition rating for the production was the highest, reaching 4.438 points. On the dimension of narrative experience, the average rating of VR opera was 4.393, compared to 4.1445 for traditional opera on the same dimension. This result highlights the potential of VR technology in enhancing the narrative experience. In addition, a survey on the awareness of VR painting in different industries found that 54% and 43% of practitioners in the art and design industries were aware of VR painting. In contrast, among non-art professions, the level of awareness was only 22.1% in the education industry and 36.2% in other sectors. This indicates that VR painting is more prevalent in the art and design. Participating in VR painting exhibitions and enjoyment by scanning QR codes was also widely accepted by the public, with a satisfaction score of 47.9167. The study found that providing VR art experiences through easy access can effectively attract and fulfill the needs of a wider audience.