AIGC-DRIVEN INTANGIBLE CULTURAL HERITAGE: INNOVATIVE DESIGN OF BIAN EMBROIDERY PATTERNS BASED ON DIFFUSION MODELS

This paper presents a theoretical and experimental exploration of how artificial intelligence generated content diffusion model technology can be applied to the preservation of Bian embroidery, an Intangible Cultural Heritage. The study first analyses the artistic characteristics of Bian embroidery, focusing primarily on colour and patterns. Subsequently, a database of colour and pattern elements was constructed and translated into prompts, whilst standardizing prompt variables. Using three AI diffusion model software programmes: ChatGPT, Gemini and Midjourney; innovative Bian embroidery patterns were generated. The experimental results indicate that all three models are capable of reproducing the visual elements of Bian embroidery. However, differences exist between the models: ChatGPT’s output more closely resembles traditional styles; Gemini tends to modify new pattern elements; whilst Midjourney generates patterns with greater visual expressiveness. This demonstrates that artificial intelligence generated content technology can facilitate the preservation and innovation of intangible cultural heritage, offering a practical method for integrating traditional embroidery with generative technology, and providing new insights and references for developing innovative media applications within the field of intangible cultural heritage.