Cultural Gene-Constrained Reconstruction of Miao Patterns

Abstract

As a core carrier of intangible cultural heritage, Miao pattern encodes creation myths, collective memory, craft knowledge, and identity. Yet digitization often separates visual form from cultural context, while unconstrained generative AI can introduce symbol misuse and decontextualization. This study proposes a cultural-gene-constrained framework that integrates parametric translation with generative reconstruction and immersive communication. Field records from Qiandongnan, including 56 pattern carriers and craft-process documentation, are organized into a structured gene library covering morphology, color, texture, process, and meaning. A culture-constrained prompt lexicon and parameter rules are coupled with Stable Diffusion and ControlNet to regulate generative variation, while an interactive environment links generated variants with contextual narratives and user feedback. In a 50-person recognition test, 46 participants correctly identified the generated patterns, corresponding to 92.0% (95% CI: 81.2–96.8%; exact binomial test against 50% chance level, p < 0.001); 44 participants correctly recalled the associated cultural meanings (88.0%, 95% CI: 76.2–94.4%). A separate 60-person three-condition user study compared static images, AI-generated dynamic video, and immersive interaction, but the available manuscript does not retain the raw eye-tracking variance required for valid inferential testing. Therefore, those results are reported descriptively rather than with fabricated p-values. The contribution is a reproducible pipeline linking cultural-gene constraints, parametric translation, AIGC generation, and immersive ICH communication. Its parameterized structure is designed for adaptation to other ethnic and regional pattern systems, while community participation, consent, attribution, and benefit-sharing are treated as governance conditions for responsible digital inheritance. The study thus shifts digital safeguarding from image preservation toward context-aware, community-governed cultural regeneration, while explicitly identifying branch coverage and longitudinal evaluation as priorities for future validation.