Creative interior design matching the indoor structure generated through diffusion model with an improved control network

AI-driven interior design generation offers promising applications. However, current AI-based diffusion models struggle to generate indoor layouts in pixel-level alignment with the indoor structure. This study proposes a new stable diffusion-based interior design workflow with an Interior Design Control Network (IDCN). IDCN ensures that the batch-generated creative interior designs based on an input image of an unfurnished room match the indoor structure. Generating innovative designs and rendering images directly with the proposed method eliminates the tedious creative design and drawing work in traditional design practices. The results indicate that the proposed method with the new design approach achieves nearly real-time design generation and modification and significantly enhances design creativity and efficiency. Moreover, the proposed method can be generalized to other design generation tasks, thereby promoting the transformation toward intelligent design.