Pattern generation and development integration method for paper-cut cultural and creative products

Abstract

The invention discloses a pattern generation and development integration method for paper-cut cultural and creative products, which comprises the following steps of: performing input acquisition and semantic prompt generation, and constructing a semantic vector by using an NLP (Network Length Polymorphism) and a paper-cut style template; generating a paper-cut style pattern based on the transformed model; extracting a contour and optimizing a structure through algorithms such as edge detection and binarization, and outputting a vector format; matching a text and creation product template, and recommending materials and colors; a user is supported to feed back local modification; the method can solve the problems that traditional paper-cut creation is low in efficiency, difficult to industrialize and the like, and has the technical advantages that creation efficiency and style diversity are improved, non-abandoned paper-cut cultural creation industrialization adaptability is enhanced, cultural inheritance and innovation are promoted, and the method is suitable for digital development and industrial conversion of paper-cut cultural creation products.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence generated content (AIGC), image processing, computer vision, digitalization of intangible cultural heritage, and design of cultural and creative products, in particular to a method for generating paper-cut patterns and developing cultural and creative products based on AIGC, which is used to promote the intelligent and digital innovative design of paper-cut art and its application and transformation in the cultural and creative industry. BACKGROUND

[0002] As an important representative of intangible cultural heritage, Chinese paper-cutting has unique folk symbolic and decorative values. However, traditional paper-cutting relies on hand design and carving, has a long creation cycle, and its style is limited to the individual creator, making it difficult to meet the demand for aesthetic diversity and individualization of modern consumption, and facing efficiency and adaptability bottlenecks in industrial transformation. In recent years, AIGC (AIGenerated Content) technology has developed rapidly, and the image generation capabilities based on diffusion models (such as Stable Diffusion) and conditional generative adversarial networks (such as StyleGAN, ControlNet) have shown potential in the fields of art design, animation, fashion, etc. However, existing AIGC systems lack understanding of the symbolic language (such as folk symbolic elements) and composition rules (such as symmetry and hollowing) of Chinese paper-cut patterns, and the generated content is difficult to be directly used for paper-cut production and cultural and creative product development, lacking a complete adaptation mechanism from pattern generation to product landing. Therefore, the traditional paper-cut creation mode cannot meet the needs of modern industry, and although AIGC technology has potential, it has not been deeply integrated with paper-cut cultural and creative development, and there is an urgent need for an integrated method that combines AIGC technology, image structure optimization algorithm, and cultural and creative product adaptation mechanism to realize automatic generation of paper-cut patterns, structure optimization, vectorization processing, and productization landing, and to promote the expression and industrial transformation of intangible cultural heritage. SUMMARY

[0003] The present application aims to provide a method for developing paper-cut cultural and creative products based on AIGC, which solves the problems of low efficiency, single style, and difficulty in industrialization of traditional paper-cut creation, and realizes automatic generation of paper-cut patterns, structure extraction, carrier adaptation, and product output, helping the development of intangible paper-cut culture through digital technology innovation and expanding the industrial application scenarios.

[0004] To achieve the above-mentioned purpose, the present application is realized by the following technical solutions:

[0005] The present application is a method for generating patterns and developing paper-cut cultural and creative products, which comprises the following steps:

[0006] Step (1) Input collection and paper-cut semantic prompt generation: users input ideas through text, images, and hand-drawn sketches, use NLP and visual understanding modules, and combine paper-cut composition rules to build semantic vectors P=f NLP(T) + a · S style , and the prompt information is completed, wherein P represents a final Prompt vector; f NLP (T) represents an encoding of the text T by the NLP model; S style represents a semantic embedding template of the paper-cut style; a represents a style weight factor and 0 < a ≤ 1;

[0007] Step (2) Paper-cut style pattern generation (AIGC): based on the Stable Diffusion model of the modified fusion ControlNet structure control mechanism, a preliminary image meeting the requirements is generated according to the semantic prompt of step (1);

[0008] Step (3) Pattern contour extraction and structure optimization: through edge detection, binarization, and morphological operation to refine the boundary, the cuttable structure is retained, the broken floating area is removed, the pattern connection logic is automatically optimized, and the SVG / DXF vector format is output through Bernstein polynomial fitting to adapt to laser cutting or traditional cutting;

[0009] Step (4) Design adaptation of cultural and creative products and material mapping: embedding the optimized pattern into the product template, recommending the material and color according to the characteristics of the carrier, and matching the pattern and the carrier to meet:

[0010] T match (P) = s · R (θ) · P + t, generating a product rendering, wherein P is the original pattern coordinate, s is the scaling factor, R (θ) is the rotation matrix, and t is the translation vector;

[0011] Step (5) User feedback and regeneration module: the user specifies the local modification area, and calls the Inpainting or resampling algorithm to realize style replacement and element redrawing;

[0012] Step (6) Output and manufacturing: exporting the vector graph SVG / DXF, and connecting the laser engraving machine and 3D printing system to output physical products, or uploading the e-commerce platform to order production.

[0013] Preferably, the NLP model in step (1) adopts the BERT model, and the style description of the paper-cut style semantic embedding template includes “black and white contrast” and “symmetry”.

[0014] Preferably, in step (3), the Canny operator is used for edge detection, and the gradient amplitude and direction are calculated by , wherein S x , S y respectively represent Sobel horizontal and vertical convolution kernels; I represents an input image; G represents a gradient amplitude; and θ represents a gradient direction.

[0015] Preferably, the binarization in step (3) uses an adaptive threshold T(x, y) = μ(x, y) + k·σ(x, y), wherein T(x, y) represents a local threshold; μ(x, y) represents a local window mean; σ(x, y) represents a local window standard deviation; and k represents a sensitivity coefficient (default 0.5).

[0016] Preferably, the morphological closing operation formula in step (3) is A represents a binary image, B represents a structural element (circular, radius 3 pixels), represents an expansion operation, is a corrosion operation.

[0017] Preferably, the vectorization conversion in step (3) is generated by fitting an SVG path through a Bernstein polynomial , wherein P i is a control point coordinate, and n is the order of the Bezier curve.

[0018] Preferably, the text creation product template in step (4) is a carrier type, which includes a lampshade, a bookmark, a clothing pattern, and a paper-cut lamp, and the material thereof includes paper, metal, and cloth.

[0019] Preferably, the laser engraving machine in step (6) has a precision of ≥0.1 mm, and can be connected to a 3D printing system (FDM or light curing type) to cooperatively output a physical product.

[0020] Preferably, the preliminary image in step (2) satisfies the conditions of “single color, high black and white contrast, and paper-cut style”, contains obvious visual contours, composition centers, and traditional and / or modern creative elements.

[0021] Preferably, the automatic optimization pattern connection logic includes a “bridge” structure in paper-cut, and the width of the “bridge” structure is ≥0.5 mm.

[0022] Beneficial effects: 1) Improved creative efficiency: Compared to traditional handmade creation, this method automatically generates paper-cutting patterns using AIGC technology, significantly shortening the creation cycle. From creative input to initial pattern generation, it can be completed in a short time. Combined with an automated structure optimization process, it allows designers to focus on creative refinement and product adaptation, improving overall development efficiency several times over; 2) Style and creative expansion: Integrating NLP semantic guidance and paper-cutting style templates, it retains the traditional paper-cutting composition rules while incorporating modern creative elements, generating diverse patterns that meet personalized needs. This breaks through the limitations of handmade creation styles and provides the market with paper-cutting cultural and creative materials with rich visual experiences; 3) Industrial adaptation. Enhanced functionality: Through structural optimization, the manufacturability of patterns is ensured. Vectorized output is adapted to modern manufacturing processes such as laser cutting and 3D printing. Product adaptation modules are precisely matched to different carriers, connecting the entire process from design to production. This helps paper-cutting cultural and creative products transform from intangible cultural heritage techniques to large-scale industries, expanding commercial application scenarios (such as fashion apparel, home decoration, cultural gifts, etc.); 4) Intangible cultural heritage inheritance and innovation: It can realize the inheritance of paper-cutting cultural symbols and technical rules through digital technology. By integrating modern creative elements and productizing them, intangible cultural heritage paper-cutting can be closer to contemporary life, attract the attention of young people, realize the living inheritance and innovative development of intangible cultural heritage, and enhance cultural dissemination and industrial value. Attached Figure Description

[0023] Figure 1 This is a system flowchart of the present invention.

[0024] Figure 2 A comparison chart of the binarization and edge detection effects of this invention.

[0025] Figure 3 Schematic diagram of the bridging structure before and after connectivity restoration according to the present invention.

[0026] in, Figure 1 It includes input → AIGC generation → structure optimization → product adaptation. Detailed Implementation

[0027] The following will refer to the appendices in the embodiments of the present invention. Figures 1-3 The technical solutions in the embodiments of the present invention are clearly and completely described herein. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0028] Prior art description: Stable Diffusion is a text-to-image generation model based on latent diffusion model, its core consists of three parts: U-Net: gradually denoising in latent space to generate images. CLIP text encoder: convert text prompts into feature vectors; Variational Autoencoder (VAE): responsible for mapping image space and latent space. ControlNet is a conditional control architecture that guides the Stable Diffusion generation process by introducing an additional "control model" to make it follow specific structural constraints (such as edge maps, key points, etc.).

[0029] Invention solution / principle:

[0030] 1. Input collection and paper-cut semantic prompt generation

[0031] Users input their creative ideas through text, images, hand-drawn sketches, etc. The system uses NLP (Natural Language Processing) and visual understanding modules to construct semantic vectors (Prompt Embedding), and combines paper-cut composition rules (such as symmetry, negative and positive carving, etc.) to automatically complete the prompt information. Semantic vector construction formula: P = f NLP (T) + a · S style , where P represents the final Prompt vector; P represents the encoding of text T by NLP model (such as BERT); S style represents the semantic embedding template of paper-cut style (such as "black and white contrast", "symmetry", etc.); a represents the style weight factor, 0 < a < 1, which precisely controls the creative and paper-cut style guidance of the generated pattern.

[0032] 2. Paper-cut style pattern generation (AIGC) based on the modified Stable Diffusion model (fusing ControlNet's structural control mechanism), according to the semantic prompt generated in step 1, to generate a preliminary image that meets the requirements of "single color, high contrast between black and white, paper-cut style"; the generated image has obvious visual contours, composition center, and integrates traditional totems (such as zodiac, folk symbols) or modern creative elements (such as cartoon characters, trendy symbols), providing basic pattern materials for subsequent processing.

[0033] 3. Pattern contour extraction and structure optimization

[0034] 3.1 Edge detection and binarization: through edge detection (using Canny operator, using gradient amplitude and direction calculation, formula: where S x , S ySobel horizontal / vertical convolution kernel; I represents the input image; G represents the gradient magnitude; θ represents the gradient direction), binarization (adaptive threshold: T(x, y) = μ(x, y) + k · σ(x, y), where T(x,y) represents the local threshold; μ(x,y) represents the local window mean; σ(x,y) represents the local window standard deviation; k represents the sensitivity coefficient (generally 0.5), the closing operation formula in morphological operation: where A represents a binary image, B represents a structural element (circular, radius 3 pixels), represents the expansion operation, represents the erosion operation) and other algorithms refine the pattern boundary.

[0035] 3.2 Structure optimization: retain the shearable structure, eliminate the broken and floating area, and automatically optimize the pattern connection logic (such as the "bridge" structure in paper cutting, ensure that the bridge structure width ≥ 0.5 mm, meet the laser engraving process requirements); output in vector format (SVG / DXF), adapt to laser cutting or traditional cutting process, and ensure the production of patterns.

[0036] 3.3 Vector conversion: SVG path is generated by Bernstein polynomial fitting, formula: where P i represents the control point coordinates; n represents the order of the Bezier curve, which realizes accurate vector storage and reuse of the pattern.

[0037] 4. The optimized paper-cut pattern is automatically embedded into the product template (such as lampshade, bookmark, clothing pattern, paper-cut lamp, etc.), and the recommended suitable material (such as paper, metal, cloth) and color style are generated according to the product carrier characteristics (material texture, size specification), and the final product rendering is generated. The pattern area and the carrier meet the transformation formula:

[0038] T match (P) = s · R(θ) · P + t, where P represents the original pattern coordinates; s represents the scaling factor; R(θ) represents the rotation matrix; t represents the translation vector, which ensures the adaptation of the pattern on different carriers.

[0039] 5. User feedback and regeneration module

[0040] Users can specify local modification areas and call local editing algorithms (such as Inpainting or resampling) to realize style replacement and element redrawing. The system supports multi-language prompt input, enhances user participation and pattern customization, and makes the generated results more consistent with user creative needs.

[0041] Output with manufacturing system export vector graph (SVG / DXF), connect laser engraving machine, 3D printing system to realize physical product output, also can be uploaded to e-commerce platform to order production, break through the conversion channel from digital design to physical product, and promote the industrialization of paper-cut cultural and creative industry.

[0042] A specific embodiment:

[0043] (I) Implementation environment building

[0044] Hardware configuration: such as using workstations with high-performance GPU (such as NVIDIA RTX40 series) to ensure the efficiency of AIGC model running and image processing; equipped with high-definition digitizer to facilitate user hand-drawing input; connected with laser engraving machine (such as CO2 laser engraving machine, precision 0.1mm), 3D printer (FDM or light curing machine type) for entity output.

[0045] Software environment: based on Python language development, such as using PyTorch framework to deploy StableDiffusion+ControlNet model, integrating NLP library (such as HuggingFace's Transformers, calling BERT model) to process semantic prompts, using OpenCV library to realize image processing (edge detection, binaryzation, etc.), using SVG.js library for vector graph generation and editing, building product template library (storing lampshade, bookmark template parameters, etc.).

[0046] (II) Specific operation process

[0047] Input collection: the user submits the system with "Chinese zodiac dragon + modern minimalist style + paper-cut bookmark" as the creative idea, through text input combined with hand-drawing (sketching the general shape of the dragon).

[0048] Semantic prompt generation: the system NLP module encodes the text, combined with the "symmetry, black and white contrast" paper-cut style template (α=0.6), constructs the semantic vector P, and completes the prompt information (such as "highlight the symmetry of the dragon's outline, high contrast between black and white, and adapt to the size of the bookmark").

[0049] Pattern generation: the modified StableDiffusion model generates a preliminary paper-cut pattern (including the outline of the dragon, with black and white contrast and symmetrical composition) according to the semantic vector.

[0050] Contour extraction and optimization: through Canny operator edge detection, adaptive binaryzation, and morphological closing operation, the pattern boundary is refined; the bridge structure is optimized (to ensure the width of 0.6mm), the broken area is removed, and the SVG format vector graph is output.

[0051] Product adaptation: The system calls the bookmark product template, embeds the vector graphics, recommends paper materials (fine texture, suitable for laser engraving), light gold decorative colors, and generates product renderings.

[0052] User feedback and adjustment: The user requests to strengthen the scale details of the dragon, calls the Inpainting algorithm for local redrawing, and the system updates the pattern and re-adapts the template. Output manufacturing: Export the SVG file, connect the laser cutting machine to cut the paper material, and make the paper-cut bookmark product; it can also be uploaded to the cultural and creative e-commerce platform to synchronize the opening of custom ordering.

[0053] (Three) Efficiency comparison of verification effect: Traditional handcrafting of similar bookmark patterns takes 2-3 days, while this method takes about 4 hours from idea to product output, significantly improving efficiency.

[0054] Style and quality: The generated pattern combines traditional zodiac culture and modern minimalist style, and after structure optimization, it can be produced with strong productivity, clear details (such as scales and bridge structures), and visual effects and process quality meeting market cultural and creative product standards.

[0055] Industrialization test: Applying this method to series of paper-cut cultural and creative development (including lampshades and clothing patterns), after going online on e-commerce platforms, custom order volume increases by 30% compared to traditional handcrafting, verifying the adaptability and market acceptance of industrialization.

[0056] The present application generates semantic vectors by input collection and semantic prompt generation, uses NLP and paper-cut style templates to construct semantic vectors, generates paper-cut style patterns based on the modified Stable Diffusion+ControlNet model, extracts contours and optimizes structures through edge detection and binarization algorithms, outputs vector formats, adapts cultural and creative product templates, recommends materials and colors, supports user feedback for local modification, and connects manufacturing equipment or e-commerce platforms for output. Solving the problems of low efficiency and industrialization difficulty in traditional paper-cut creation, improving the efficiency and style diversity of creation, enhancing the adaptability of non-heritage paper-cut cultural and creative industrialization, promoting cultural inheritance and innovation, and being applicable to digital development and industrial transformation of paper-cut cultural and creative products.

[0057] Finally, it should be noted that the present application is not limited to the above embodiments, but can have many variations. All variations that can be directly derived or inferred from the disclosed content by those of ordinary skill in the art should be considered within the scope of the present application.

Claims

1. A method for generating and developing integrated patterns for paper-cutting cultural and creative products, characterized in that, Includes the following steps: Step (1) Input Collection and Semantic Hint Generation for Paper Cutting: Users input their ideas through text, images, and hand-drawn drafts. Using NLP and visual understanding modules, combined with paper cutting composition rules, a semantic vector P = f is constructed. NLP (T)+α·S style Complete the prompt information, where P represents the final Prompt vector; f NLP (T) represents the NLP model's encoding of text T; S style A semantic embedding template representing the paper-cutting style; α represents the style weighting factor and 0 < α ≤ 1; Step (2) Paper-cut style pattern generation (AIGC): Based on the modified StableDiffusion model with integrated ControlNet structure control mechanism, a preliminary image that meets the requirements is generated according to the semantic prompts in step (1); Step (3) Pattern contour extraction and structure optimization: refine the boundary through edge detection, binarization, and morphological operations, retain the shearable structure, eliminate broken floating areas, automatically optimize the pattern connection logic, and output SVG / DXF vector format after Bernstein polynomial fitting to adapt to laser cutting or traditional cutting. Step (4) Cultural and Creative Product Design Adaptation and Material Mapping: Embed the optimized pattern into the product template, recommend materials and colors based on the characteristics of the carrier, and ensure that the pattern and carrier match to meet the requirements of T. match (P) = s·R(θ)·P+t, generating the product rendering, where P is the original pattern coordinates, s is the scaling factor, R(θ) is the rotation matrix, and t is the translation vector; Step (5) User feedback and regeneration module: The user specifies a local modification area, and the Inpainting or resampling algorithm is called to realize style replacement and element redrawing; Step (6) Output and Manufacturing: Export vector graphics SVG / DXF and connect to a laser engraving machine or 3D printing system to output physical products, or upload to an e-commerce platform to place an order for production.

2. The method for pattern generation and development integration of paper-cutting cultural and creative products according to claim 1, characterized in that, In step (1), the NLP model used is the BERT model, and the style description of the paper-cut style semantic embedding template includes "black and white contrast" and "symmetry".

3. The method for pattern generation and development integration of paper-cutting cultural and creative products according to claim 1, characterized in that, In step (3), edge detection uses the Canny operator, through... Calculate the gradient magnitude and direction, where S x S y θ represents the horizontal and vertical convolution kernels of Sobel, respectively; I represents the input image; G represents the gradient magnitude; and θ represents the gradient direction.

4. A method for pattern generation and development integration of paper-cutting cultural and creative products according to claim 1 or 3, characterized in that, In step (3), binarization uses an adaptive threshold: T(x,y)=μ(x,y)+k·σ(x,y), where T(x,y) represents the local threshold; μ(x,y) represents the local window mean; σ(x,y) represents the local window standard deviation; and k represents the sensitivity coefficient.

5. The method for pattern generation and development integration of paper-cutting cultural and creative products according to claim 4, characterized in that, The formula for the morphological closing operation in step (3) is: A represents a binary image, and B represents a structuring element. This is represented as an expansion operation. This is a corrosion operation.

6. The method for pattern generation and development integration of paper-cutting cultural and creative products according to claim 5, characterized in that, In step (3), the vectorization transformation is performed using Bernstein polynomials. Fitting and generating SVG paths, where P i Here are the coordinates of the control points, and n is the order of the Bézier curve.

7. A method for pattern generation and development integration of paper-cutting cultural and creative products according to claim 1, 2, or 3, characterized in that, Step (4) The Chinese creative product template is a carrier type, which includes lampshade, bookmark, clothing pattern, paper sculpture lamp, and its material includes paper, metal and cloth.

8. A method for pattern generation and development integration of paper-cutting cultural and creative products according to claim 1, 2, or 3, characterized in that, In step (6), the laser engraving machine has an accuracy of ≥0.1mm and can be connected to the existing 3D printing system to collaboratively output physical products.

9. The method for pattern generation and development integration of paper-cutting cultural and creative products according to claim 1, characterized in that, The preliminary image in step (2) meets the requirements of "monochrome, high contrast black and white, paper-cut style", and contains obvious visual outline, composition center and traditional and / or modern creative elements.

10. The method for pattern generation and development integration of paper-cutting cultural and creative products according to claim 1, characterized in that, The automatic optimization pattern connection logic includes the "bridge" structure in paper cutting, and the width of the "bridge" structure is ≥0.5mm.

Images