Constructing an AIGC Pipeline Workflow with Dataset and Model for Cloisonne Cultural Product Design

Abstract

Cloisonne enamel, a Chinese intangible cultural heritage, faces procedural complexity, stylistic stagnation, and disconnection from modern consumer aesthetics. A critical gap persists: no existing study combines a culturally annotated dataset with a multi-tool generative pipeline specifically for cloisonne design. This study proposes a multi-tool AIGC workflow integrating a culturally annotated dataset, a sequential generative pipeline, and a three-stage design process. We constructed the Cloisonne Design Dataset (CDD) comprising 1000 images with a three-level annotation schema (metadata, visual description, and cultural semantics). The pipeline chains Midjourney for initial generation, Seedream 5.0 for detail refinement, and Tencent Hunyuan 3D for three-dimensional visualization. Over 500 design schemes were generated across three categories: high-fidelity reproduction, modernist translation, and cross-disciplinary fusion. Quantitative evaluation and a 20-participant user survey confirm that the pipeline effectively balances cultural fidelity with creative innovation. We discuss the transformation of designer roles, risks of cultural homogenization, and ethical implications, offering a preliminary framework for AIGC-assisted intangible cultural heritage design.