An Empirical Study on the Applicability of Image Generation AI in Fashion Design: A Comparative Analysis of GPT, LOOK AI, and Diffusion

This study examines the applicability of generative AI in fashion design through a comparative analysis of GPT-based generators, LOOK AI, and Stable Diffusion. A dress sketch was chosen as a complex test item, and images were evaluated by 15 professional designers using a structured survey across three domains: design reproduction, visual fidelity, and practical usability. ANOVA results showed that LOOK AI achieved the highest accuracy and usability, GPT-based tools performed moderately with strengths in structural interaction, and Stable Diffusion, while creative, showed lower fidelity. By combining expert evaluation with a complex garment type under controlled conditions, the proposed method provides a practical framework for validating AI image tools, offering implications for fashion practice and education as well as potential extension to other design fields.