The integration and application of computer technology and new media art can not only optimize the new media art presentation form, but also innovate the presentation form of art design. Focusing on the innovation of new media art design, this paper discusses and analyzes the promotion role of computer-aided creation methods in it. Based on the application of image generation technology in AI, taking the single-stage generative adversarial network as the basic model and introducing multiple intelligent modules, we design the text-generated image method based on DGF-GAN to stimulate and assist the design of new media art. The images generated by the DGF-GAN model are of excellent quality and diversity, and its FID and IS values perform optimally in the comparison experiments. Compared to the base model, the FID values of the DGF-GAN model are reduced by 6.34% and 5.83% and the IS values are improved by 1.46% and 6.60% on both datasets. In addition, the method has good training efficiency and image generation efficiency. The results show that the proposed method has a large development potential in new media art design, which can enhance the design flexibility and designer creativity, and promote the development of art creation in a more intelligent and precise way.
