An Intelligent Generation Method for Kizil Rhombic Mural Patterns Based on Diffusion Models and Its Application in Digital Fashion Design

Structured geometric patterns play an important role not only in cultural heritage preservation and digital design but also in engineering applications where periodic layouts inspire functional surface architectures and electromagnetic wave manipulation. The Kizil rhombic murals, characterized by distinctive lattice configurations and rich visual semantics, present significant challenges for intelligent regeneration because conventional generative methods cannot simultaneously preserve stylistic authenticity and geometric regularity. This study develops a conditional diffusion-based framework that combines latent diffusion modeling with ControlNet-guided structural constraints to generate high-fidelity Kizil rhombic patterns. A dedicated high-resolution dataset is established, and the model is optimized to jointly capture chromatic characteristics, artistic motifs, and rigid rhombic-grid organization through Canny edge conditioning. Quantitative evaluation using Fréchet Inception Distance, Inception Score, and Structural Adherence Score demonstrates that the proposed approach achieves superior visual quality and structural consistency compared with existing generative methods. The generated patterns are further integrated into a digital fashion design workflow for virtual garment creation, confirming their practical applicability in creative industries. Beyond cultural heritage regeneration, the proposed framework provides a data-driven strategy for generating structured periodic patterns with potential reference value for geometry-oriented design problems in electromagnetic surfaces, antenna-inspired pattern engineering, and intelligent visual manufacturing.