AI-Generated Oriental Fantasy in Fashion Imagery: Digital Reimagining of Traditional Chinese Dress

Abstract

This study focuses on the fashion expression of oriental fantasy-style clothing in AI-generated images, drawing on the ‘visual grammar’ theoretical framework proposed by Kress and van Leeuwen. It treats AI-generated images as visual meaning systems with language-like structures, systematically interpreting the cultural expressions and stylistic compositions within the images. Using the visual social media platform Rednote (Xiaohongshu) as the data source, this paper collected and analysed 62 AI-generated fashion images of traditional Chinese clothing. The study establishes a coding system based on three visual dimensions: pattern elements, color combinations, and material textures, to identify the key visual elements that constitute the oriental fantasy clothing style. It further analyses how these elements work together to form a unique visual language, and summarises and systematises the style standards and visual preferences of AI-generated images in the fashion representation of traditional Chinese clothing. This study provides an empirical case for the visual reproduction of Chinese fashion culture through AI technology, offering a new perspective for cultural research in the interdisciplinary field of fashion and technology. Additionally, this paper constructs a visual element coding system for traditional attire applicable to AI-generated images, expanding the methodological tools for visual culture research.