Co-creation with AI in Car Design: A Diffusion Model Approach

被引:1
作者
He, Zhicheng [1 ]
Ma, Jun [2 ]
Zuo, Yuanyang [1 ]
机构
[1] Tongji Univ, Coll Design & Innovat, Shanghai, Peoples R China
[2] Tongji Univ, Sch Automot Studies, Shanghai, Peoples R China
来源
ARTIFICIAL INTELLIGENCE IN HCI, PT II, AI-HCI 2024 | 2024年 / 14735卷
关键词
Co-creation; Diffusion Model; DreamBooth; Car Design;
D O I
10.1007/978-3-031-60611-3_12
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
This study delves into the idea of AI co-creation in the car design domain by employing a unique methodology of training diffusion model to learn car design. Our primary objective is to investigate how the collaborative design process, where human designers and AI work together with the assist of pre-trained car design diffusion model, can revolutionize car design by enhancing creativity, efficiency, and overall user satisfaction. We built the car design dataset for the model training, which consists of real car photos and the design renderings together with the textual prompt of the images. Andwe invited the professional car designers to test the pre-trained car design diffusion model in the real design scenario. This research underscores the significance of AI co-creation in car design, showcasing the potential of AI technology to revolutionize the automotive industry. The findings suggest that collaborative AI co-creation approach had the ability to empower designers, enhance creativity, and redefine the future of car design. Project page https://automotive-design-copilot.super.site
引用
收藏
页码:163 / 174
页数:12
相关论文
共 22 条
[1]  
[Anonymous], Nvidia Blog page
[2]  
Avrahami O, 2023, Arxiv, DOI arXiv:2305.16311
[3]  
Catalanoc E, 2005, Car Aesthetics Annotation, P8
[4]   Can style be measured? [J].
Chan, Chiu-Shui .
Design Studies, 2000, 21 (03) :277-291
[5]  
Choi J, 2023, Arxiv, DOI arXiv:2305.15779
[6]  
Gong R., 2023, arXiv
[7]  
Hedlin E, 2023, Arxiv, DOI arXiv:2305.15581
[8]  
Kim BK, 2024, Arxiv, DOI arXiv:2305.15798
[9]  
Lee S, 2024, Arxiv, DOI arXiv:2305.03509
[10]  
Li DX, 2023, Arxiv, DOI arXiv:2305.14720