Exploring the Future of Fashion: AI Predicts Menswear Trends at Pusan National University
Researchers at Pusan National University, under the guidance of Professor Yoon Kyung Lee, have taken a groundbreaking step in fashion design by employing generative AI to predict menswear trends for the fall/winter 2024 season. This innovative research leveraged tools such as ChatGPT-3.5, ChatGPT-4, and DALL-E 3 to analyze historical fashion data, producing a fascinating blend of technology and creativity.
AI’s Emerging Role in Fashion Design
The study focused on harnessing generative AI to visualize upcoming seasonal trends. Using historical data available up to September 2021, the researchers aimed to predict modern menswear styles. By systematically classifying design elements into initial codes, the team distilled the information into six final categories encompassing trends, silhouette elements, materials, key items, garment details, and embellishments. This structured framework provided a robust basis for translating complex data into actionable insights for the generation of fashion designs.
The researchers created 35 specific prompts for DALL-E 3, detailing unique outfits. Central to their approach was a common template depicting a male model on a runway, allowing customization of various parameters, including aspect ratios and backgrounds. The result? A total of 105 images generated by running each prompt three times.
Success and Challenges of Implementation
In their efforts, DALL-E 3 successfully implemented the prompts 67.6% of the time. The inclusion of descriptive adjectives in the prompts significantly increased the accuracy of the generated designs. However, while many images closely imitated actual menswear collections for 2024, the study noted some shortcomings, such as a predominant inclination towards ready-to-wear fashion and challenges in representing emerging trends like gender fluidity.
The findings underscored the importance of detailed and nuanced prompts, indicating that using isolated trend keywords was not adequate for achieving a high-quality output. Professor Lee emphasized that expertly crafted prompts are essential for accurate implementation in fashion design, thus affirming the enduring relevance of fashion expertise in this technologically advanced landscape.
Predictive Analysis and Methodology
The research utilized both ChatGPT-3.5 and ChatGPT-4 to analyze past data and predict future menswear trends. This involved scrutinizing various design elements and coding them into the predefined six categories. The structured framework aimed to transform raw design data into prompts that could be effectively used for image generation in generative AI.
Each of the 35 prompts provided a unique description of an outfit, consistently set in a runway scenario for the autumn/winter 2024 season. This method allowed the researchers to customize numerous aspects, including camera angles, model appearances, and runway settings, resulting in a rich and diverse output of fashion imagery.
While DALL-E 3 successfully captured many of the essence of the prompts, the researchers identified a need for more sophisticated inputs, especially for representing contemporary themes like gender fluidity. The study revealed that generating compelling fashion images requires a nuanced approach that goes beyond basic trend keywords.
Implementation and Future Potential
The implications of this research extend well beyond academic curiosity. As generative AI models like DALL-E 3 continue to evolve, they hold the potential to revolutionize the fashion design industry. By increasing designers’ efficiency and creativity, these tools can enable a richer understanding of fashion trends among both professionals and enthusiasts alike.
This exciting intersection of AI and fashion suggests a future where not only designers but also non-experts can explore, predict, and style upcoming seasons with enhanced confidence. The research highlights the dual role of generative AI as both an innovative partner in the creative process and a bridge for fostering a greater appreciation of fashion trends.
The Visionary Behind the Research
Professor Yoon Kyung Lee, an Assistant Professor at Pusan National University, is at the forefront of this research journey. Specializing in creativity and sustainability in fashion design, her focus spans across AI, digital technology, and neuroscience. With an academic background that includes an MSA and Ph.D. in dress aesthetics from Seoul National University, Professor Lee has also made her mark in the industry by showcasing her own brand, UginiO, at notable events like Seoul Fashion Week and Prt-Porter Paris.
Her work not only sheds light on the untapped potential of AI in fashion but also reinforces the significance of human creativity and expertise in an era increasingly defined by technological advancements. To learn more about her research, visit her lab website at Pusan National University Fashion Design Lab, or explore her ORCID profile at 0000-0002-5118-3789.