A Study on the Applicability of Generative Artificial Intelligence in the Automotive Design Process
A Study on the Applicability of Generative Artificial Intelligence in the Automotive Design Process
신진규(홍익대학교); 노재승(국민대학교)
33권 10호, 849~857쪽
초록
Since ChatGPT's introduction in November 2022, generative AI tools such as DALL-E and Midjourney have significantly impacted on automotive design. DALL-E uses Convolutional Neural Networks (CNN) to align image details with textual prompts, while Midjourney employs Generative Adversarial Networks (GAN) to produce realistic images through competing neural networks. This study explores both tools in automotive design via qualitative and quantitative research methods, using generated images, extracted prompts, and adjective keywords. This study showed DALL-E produced results closer to user intentions compared to Midjourney, supported by user surveys. The study suggests future research should explore language variations, different vehicle types, user expertise, and professional designer insights to expand generative AI's application in design.
Abstract
Since ChatGPT's introduction in November 2022, generative AI tools such as DALL-E and Midjourney have significantly impacted on automotive design. DALL-E uses Convolutional Neural Networks (CNN) to align image details with textual prompts, while Midjourney employs Generative Adversarial Networks (GAN) to produce realistic images through competing neural networks. This study explores both tools in automotive design via qualitative and quantitative research methods, using generated images, extracted prompts, and adjective keywords. This study showed DALL-E produced results closer to user intentions compared to Midjourney, supported by user surveys. The study suggests future research should explore language variations, different vehicle types, user expertise, and professional designer insights to expand generative AI's application in design.
- 발행기관:
- 한국자동차공학회
- DOI:
- http://dx.doi.org/0
- 분류:
- 자동차공학