HarmonyCut: Supporting Creative Chinese Paper-cutting Design with Form and Connotation Harmony

TL;DR

HarmonyCut tackles the challenge of aligning form and cultural connotation in Chinese paper-cutting by introducing a factor-guided ideation pipeline and a reference-based composition workflow powered by GenAI. The system combines an ideation module with domain-annotated knowledge and a composition module that retrieves, segments, and arranges references on a mood board, all grounded in a taxonomic design space for paper-cutting patterns. Evaluation with a user study (n=16) and expert interviews (n=3) shows improved ideation guidance, richer reference exploration, and greater editing controllability, while preserving cultural meaning and boosting engagement with ICH. The work demonstrates a practical path for using structured cultural knowledge to augment creative design with GenAI, offering a scalable approach to preserve and disseminate traditional crafts through digital tools.

Abstract

Chinese paper-cutting, an Intangible Cultural Heritage (ICH), faces challenges from the erosion of traditional culture due to the prevalence of realism alongside limited public access to cultural elements. While generative AI can enhance paper-cutting design with its extensive knowledge base and efficient production capabilities, it often struggles to align content with cultural meaning due to users' and models' lack of comprehensive paper-cutting knowledge. To address these issues, we conducted a formative study (N=7) to identify the workflow and design space, including four core factors (Function, Subject Matter, Style, and Method of Expression) and a key element (Pattern). We then developed HarmonyCut, a generative AI-based tool that translates abstract intentions into creative and structured ideas. This tool facilitates the exploration of suggested related content (knowledge, works, and patterns), enabling users to select, combine, and adjust elements for creative paper-cutting design. A user study (N=16) and an expert evaluation (N=3) demonstrated that HarmonyCut effectively provided relevant knowledge, aiding the ideation of diverse paper-cutting designs and maintaining design quality within the design space to ensure alignment between form and cultural connotation.

Figures (12)

• Figure 1: The pipeline consists of Ideation and Composition components, structured around the summarized workflow and design space. The Ideation component offers knowledge-based guidance, allowing designers to explore and select content that aligns with their intent to form ideas. These ideas are then fed into multi-modal models within the Composition component, which retrieves and generates related content as references. This exploration of references empowers users to arrange reference and plan cut-out areas, facilitating the composition of their own paper-cutting designs.
• Figure 2: The interface of HarmonyCut supports user creative paper-cutting design through several panels with guidance and exploration. (a) In the Idea Prompting panel, the system guides users through idea navigation using four factors: function, subject matter, style, and method of expression. It also provides related patterns with interpretations. Users can select suggested content that aligns with their intent or manually edit ideas. (b) The Interactive Mood Board panel allows users to arrange all references, including contours and patterns, for composition. (c) In the Reference Exploration panel, users explore system-suggested content, both retrieved and generated, to select contours and plan pattern layouts, thereby gaining ideas for their paper-cutting designs.
• Figure 3: The detailed process of design with each view and result in HarmonyCut. (a) The idea description from the former ideation. (b.1) The reference generation view based on the prompt; (b.2 and b.3) the contour segmentation view for the selected references and their contours; (b.4) cut-outs from original or other paper cuttings used in the design. (c) The final design is displayed on the mood board.
• Figure 4: The 6 paper-cutting design examples were created by the 6 participants in the user study. All solid references are retrieved, while dashed references are generated.
• Figure 5: Sixteen participants ratings the design goals questionnaire across different expertise levels on paper-cutting and GenAI.