DIGITAL TWIN RECONSTRUCTION OF MUSICAL INSTRUMENTS IN DUNHUANG MURALS: AN ENGINEERING WORKFLOW GROUNDED IN PHYSICAL REPLICAS

Abstract

The Mogao Caves murals at Dunhuang preserve a dense iconography of music and dance from medieval times, yet the two-dimensional carrier and age-related degradation hinder quantitative understanding of instrument morphology. This paper proposes and validates a replica-based, 3D-capture-centered, visualization-oriented workflow for digital twin reconstruction. Using historically reasoned physical replicas as targets, we perform multi-sensor acquisition (structured light, photogrammetry), accurate registration, and high-fidelity meshing; we then conduct semantic retopology with lightweighting in parallel to physically based texture repainting. Representative samples—including the gourd lute (huluqin), large gourd lute, petal-edged ruan, guqin, bent-neck pipa, stick pipa, and round-bodied large ruan—are used to establish a traceable data lineage and metadata schema, demonstrating portability and dual effectiveness for exhibition and teaching. Results indicate that the method preserves mural-style cues while markedly improving geometric consistency and interactive legibility. The workflow provides an engineering-grade, reusable path for digital conservation, scholarly comparison, and public outreach of Dunhuang instruments.