Answers are grounded in this paper and cite their sources.
Dunhuang, a historic crossroads of the Silk Road, preserves in its Tang Dynasty murals a visual culture in which Central Plains dress conventions were reshaped by influences from the Western Regions. This practice-based study examines how generative artificial intelligence (AI) participates in the contemporary re-creation of this heritage through the case of doll clothing, a collectible culture that spread from South Korea to China and has become closely tied to young consumers’ renewed interest in traditional Chinese aesthetics. Drawing on Dunhuang mural imagery made widely accessible through digitization, the project combined a comparative generation experiment across AI platforms and text-to-image, image-to-image, and conversational methods with semi-structured interviews with five practitioners. A proposal inspired by the Flying Apsaras was developed into a physical prototype using Shu and Song brocade, both recognized as intangible cultural heritage, together with hand-applied beadwork. AI accelerated ideation and broadened aesthetic exploration, and reference-based generation proved most effective in preserving Dunhuang motifs and mineral-pigment palettes. However, the tools often mixed Dunhuang elements from different periods, confused them with Japanese or Southeast Asian patterns, and rendered fine textile detail unreliably, revealing an incomplete grasp of Chinese historical aesthetics that required human cultural judgment to correct. The study argues that AI functions as a new medium through which historical traditions are re-created for contemporary audiences, while historical knowledge and living craft traditions remain essential to their faithful transmission.
No takes yet. Share an insight, caveat, or question.
Zhou et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: