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

Generative AI (Text, Image, Music, Video)Museum & Cultural Heritage DigitizationVisual Artists & DesignersCraft Artisans (Textiles, Ceramics, etc.)

Research Background and Issues

  • Identified Problems or Challenges:

    • The Chinese paper-cutting art, an intangible cultural heritage, faces a crisis in creativity and inheritance under the impact of modernization and urbanization.
    • The traditional themes and symbolic meanings of paper-cutting are gradually fading, and the scarcity of related paper-based literature limits the opportunities for the public and professional practitioners to access this cultural knowledge.
    • Homogeneity in form and content during creation, along with a tendency toward realism, has affected the creativity and heritage value of paper-cutting.
    • Current design methods based on generative artificial intelligence (GenAI) still lack capabilities in handling abstract themes and cultural contexts.
  • Significance:

    • Paper-cutting is not only an art form but also an important medium for cultural exchange. Preserving and innovating this art can promote cross-cultural understanding and integration.
  • Research Motivation and Related Work:

    • Although existing research has utilized computer technology for pattern generation and interactive design, most methods are limited to predefined forms and styles, lacking comprehensive handling of complex cultural knowledge and personalized expression.
    • While GenAI provides efficient generative capabilities and data knowledge, it lacks deep understanding of domain-specific knowledge, such as the integration of paper-cutting forms and cultural connotations.

Solution

Methodology and Core Innovations

  1. Proposed the HarmonyCut tool:
    • By combining generative artificial intelligence with domain knowledge of paper-cutting, it constructs a more creative and structured design process to address the limitations of existing generative technologies.
  2. Introduced the following innovative designs:
    • Factor-based Guidance: Decomposes creative paper-cutting design into four core factors—“Function,” “Content Theme,” “Style,” and “Expression Method”—to clarify the key dimensions to consider during the design process.
    • Reference-based Exploration: Retrieves and generates related paper-cutting works or elements to provide users with design references.
    • Editability and Controllability: Enables users to adjust generated results, enhancing their control over the creative process.

Implementation Steps and Key Technologies

  1. Formative Research:

    • Conducted interviews and content analysis (140 paper-cutting works) to summarize the core processes of paper-cutting design (creative generation and visualized creation) and its challenges.
    • Refined the classification of creative factors and pattern types in paper-cutting.
  2. Tool Development:

    • Built a design workflow component consisting of two modules:
      1. Creative Module: Utilizes the GLM-4 model to guide users in selecting and constructing design ideas.
      2. Composition Module: Supports users in exploring reference materials and completing paper-cut compositions through model extraction and pattern recommendations.
    • Implemented contour extraction and pattern generation using multimodal models (e.g., DALL-E-3) and cutting algorithms (Segment Anything Model).
  3. User Interaction and Editing Support:

    • Developed an interactive Mood Board supporting SVG format, allowing users to arrange, adjust, and cut selected elements based on their design goals.

Research Outcomes

Specific Results

  1. Systematically defined the design space for paper-cutting, clarifying the two-step workflow of creative generation and composition.
  2. Developed the HarmonyCut prototype tool, offering design support based on factor guidance and reference exploration.
  3. Validated the annotation method for categorical datasets (e.g., classification of unit patterns and composite patterns, providing 140 works and 1,279 pattern annotations).

Advantages and Experimental Results

  • Advantages Over Existing Solutions:

    • HarmonyCut integrates cultural knowledge and multimodal generative technology, successfully enhancing the creative process of paper-cutting design, addressing deficiencies in cultural expression and form control in existing systems.
    • The system’s guidance reduces cognitive load for novice users during the creative phase while allowing experts to flexibly explore more references, avoiding creative stagnation.
  • Experimental and Evaluation Results:

    • User experiments showed that HarmonyCut significantly improved user experiences in creative generation and reference exploration.
    • NASA-TLX results indicated that HarmonyCut significantly reduced users’ “mental load” during design, though it slightly increased physical operational burden.
    • Expert feedback highlighted the system’s outstanding performance in understanding and expressing cultural connotations, particularly in guiding users to comprehend traditional paper-based culture.

Limitations and Future Directions

  1. Limited Dataset Scale: The current dataset (140 paper-cutting works) cannot fully cover the broad cultural domain of paper-cutting. Expanding annotated materials with diverse backgrounds and types is recommended.
  2. User Load Balance: While the system enhances controllability, it requires users to invest more time and effort in paper-cut composition. Future work should aim to balance controllability and operational complexity.
  3. Distinction Between Imitation and Innovation: Users may overly rely on system-recommended references. Future improvements could focus on guiding users to integrate new themes and meanings to foster innovation.
  4. Collaborative Design Expansion: The system currently does not support multi-user collaborative design. Future exploration could include features for collective creativity to further enhance the design experience.

Conclusion

HarmonyCut integrates cultural knowledge with generative artificial intelligence, providing a novel methodological framework and tool platform for creative paper-cutting design. Through standardized definitions of the design process and user research validation, this tool not only aids individual designers in their creative practices but also opens new pathways for the preservation and promotion of intangible cultural heritage.

Quick Actions

Share

Share this page

ios_share

https://hci.top/en/papers/chi/188649/2025

AdRecommended

Learn AI Coding at CodeNow

open_in_newOpen DOI Link
DOI: https://dl.acm.org/doi/10.1145/3706598.3714159
At a Glance

Paper Snapshot

fact_check
dataset
Source
CHI
calendar_month
Year
2025
emoji_events
Award
No award tagged
group
Authors
5 authors
sell
Subtopics
Generative AI (Text, Image, Music, Video), Museum & Cultural Heritage Digitization
work
Professions
Visual Artists & Designers, Craft Artisans (Textiles, Ceramics, etc.)
article
Content Status
Full text indexed
hub
Related Papers
3 related papers