Under the impetus of computer technology, the creation of digital art continues to develop, and computer-assisted creation has gradually become the mainstream of artistic creation. This paper is oriented to digital art innovation, in-depth exploration of computer-assisted art creation and its integration with the development of digital media. Through the in-depth analysis of computer-assisted art creation, this paper constructs an improved CycleGAN art pattern generation model by introducing the self-attention mechanism in the CycleGAN model on the basis of pattern generation. In the generation experiments of the improved CycleGAN model, the SSIM and PSMR values of the improved model in this paper are 0.721 and 17.563, and in the number of in-parameters, the model size, and the running speed are reduced compared with the traditional model, and the overall performance of the improved model is excellent. At the same time, the works based on the computer-aided art creation method of this paper compared with the traditional art creation works of the comprehensive average score increased by 11.40 points, further illustrating the more advantageous in computer-aided art creation. The study concludes by analyzing the path of the combination of computer-aided and digital media, and proposes a path for the integration and development of the two from multiple perspectives, which provides directions and ideas for the research on the integration and development of computer-aided and digital media technologies.