With the development of artificial intelligence, the use of generative adversarial networks in deep learning can generate good visual art based on content, and at the same time, it can reduce the problems of noisy texture, transition migration, and image distortion that exist in content-based image generation. The article proposes a multi-view image generation architecture (DrawGAN) based on generative adversarial networks and builds an AI design tool on this basis to explore the effect and analysis of the design tool for content generation based on generative adversarial networks on visual art. Through experimental testing, the model proposed in this paper, after the constraints of the relevant loss function, the generated image content is compatible with the input image, the style is consistent with the style of the target image, and the image generated by the base DrawGAN has a high degree of reproducibility. In addition, the average score of visual art images generated using the model proposed in this paper is 8.25, which is 2.26 points higher than that of images generated by traditional methods. To sum up, the DrawGAN model proposed in this paper is a valuable tool in the generation of visual art images.