UIST '25

3D sketching + 2D generative AI for car exterior design

Seung-Jun Lee

Jeongche Yoon

Sang-Hyun Lee

Joon Hyub Lee

Seok-Hyung Bae

[PDF]

Recent advances in 2D generative AI are finding applications in highly specialized fields, such as car exterior design. However, current 2D-centric approaches have several limitations: each viewpoint requires a new sketch; consistency across different viewpoints is difficult to maintain; and steering design development in the desired direction is challenging. To address these limitations, we propose a novel workflow that integrates 3D sketching with 2D generative AI for car exterior design. This workflow enables car designers to seamlessly transition between expressive 3D sketching, detailed 2D drawing, and realistic 2D generation. In a formal user study, professional car designers used our system to create new exterior concepts across all major car body types. They produced high-quality, view-consistent sets of 2D renderings in a short time, demonstrating diverse patterns of progressive design development.

30 sec. preview

3 min. presentation

People

Seung-Jun Lee

Ph.D. Student

Jeongche Yoon

Ph.D. Student

Sang-Hyun Lee

Ph.D. Student

Joon Hyub Lee

Research Professor

Seok-Hyung Bae

Associate Professor

Awards

🏆 UIST '25 Jury Best Demo Honorable Mention

3D sketching + 2D generative AI for car exterior design

Seung-Jun Lee, Jeongche Yoon, Sang-Hyun Lee, Joon Hyub Lee, and Seok-Hyung Bae

Department of Industrial Design, KAIST
© SketchLab All Rights Reserved.
Twitter
YouTube