MIX: An Image Generation System Using Image Prompts for Industrial Design

Abstract

Generative AI is increasingly being applied in the industrial design field. However, the existing AI design tools are still challenging due to complex operations and parameter settings. To solve these problems, we conducted comprehensive interviews with industry experts and proposed the MIX system. The system emphasizes image prompts and is designed to support designers for more efficient task completion. MIX incorporated both the IP-Adapter and the ControlNet control model with multimodal language models, making more natural interaction possible. It enables designers to quickly produce high-quality design renderings from target reference images. Through qualitative and quantitative user experiments, we found that the MIX system performs well in both efficiency and practicality in design development. Finally, we discuss about when it would be more appropriate to use text prompts versus image prompts in these different design scenarios. Available at https://mix.drafff.net.

Keywords: Generative Artificial Intelligence, Image Prompts, Product Rendering

DOI: 10.54941/ahfe1006167