Research on Intelligent Construction in Generative Adversarial Networks
摘要
Reflecting on the visualization issues brought by AI-generated design images, this paper proposes the concept of intelligent construction. A library of machine-recognizable constructed images is summarized through the study of material construction techniques. Additionally, by integrating image generation software such as Midjourney and the characteristics of Generative Adversarial Networks (GANs), a construction image generation process aligned with the design workflow was developed. Guided by this process, a specific bookstore design case was implemented to validate the rationality of the workflow. Intelligent construction, as one of the research directions of generative design, can further advance the transformation of AI generation towards three-dimensional modeling.