From Symbolic Decoding to Embodied Reconstruction: A Cross-modal Translation Study of Chu Cultural Heritage Based on Generative AI (AIGC)

Abstract

In response to the issues of style deviation and semantic loss in the generation design of traditional cultural heritage, this study aims to explore an innovative design paradigm of translating from “physical symbols” to “digital embodied experiences”. Firstly, based on the theory of embodied cognition, this study constructed a theoretical model for the embodied translation of visual elements of Chu culture, clarifying the cognitive mapping mechanism of cultural genes in digital reconstruction. At the technical implementation level, a visual style transfer workflow centered on the Stable Diffusion and LoRA models was developed. Through parameterized training and constraining the cultural consistency of the generation space, it solved the problems of aesthetic consistency and style deviation in the digital generation of traditional patterns. Through a series of design practices in “Delta.Studio”, this approach demonstrated its effectiveness in immersive interaction, mobile application, and cultural product development. AIGC technology not only efficiently expands the visual expression dimension of intangible cultural heritage but also realizes the dynamic activation of cultural genes through the closed loop of “human intervention - model training - multimodal output”. This study provides a replicable methodological reference for the protection and innovative design of cultural heritage from the perspective of digital humanities.