Research on the Innovative Application of Song Dynasty Boundary Painting in Interior Soft Decoration Design Based on AIGC

Abstract

An analysis of the practice path and methodology system of Artificial Intelligence Generated Content (AIGC) technology has been conducted in the field of inheritance and innovation of boundary paintings from Song Dynasty. This paper aims to provide valuable reference and guidance for the application of AI technology in Song Dynasty boundary painting (Song painting) in the interior decoration design, so as to promote the effective integration of traditional aesthetics and modern design concepts. Firstly, the natural processing language model is used to generate the index layer suitable for the indoor soft decoration style of Song paintings, and the Analytic Hierarchy Process weight classification is used to select the cue words of the generated image. Secondly, Midjourney is used to generate Song Dynasty style images for keywords. Finally, Stable Diffusion control model is used to transfer the style of Song painting elements to interior decoration design. AIGC technology can effectively generate images with the style of Song painting elements, and play a unique role in style transfer and pattern design. It provides an innovative path for the integration of traditional art and modern design, and provides a wealth of possibilities for the modern application of Song painting. AIGC technology has significant potential in the inheritance and innovation of Song painting, which can bring new ideas and methods for interior decoration design, and contribute to the wide application and innovative development of Song painting art in the field of modern design.

1. Introduction

With the development of AI technology, generative Artificial intelligence (AIGC) is increasingly used in the field of art and design, especially in the inheritance and innovation of cultural heritage and traditional art forms, which shows great potential [1,2]. Study of Song painting, a traditional Chinese painting form with a profound historical background and unique artistic style, and its application to contemporary interior soft decoration. Song Dynasty boundary paintings are famous for their fine lines and rigorous structure, It is a very characteristic category of Chinese painting. In painting the use of boundary ruler lead, so the name of boundary painting, can also be referred to as Song painting, which not only reflect the social style of Song Dynasty, but also show the aesthetic taste at that time. However, in modern society, how to combine this traditional art form with modern design concepts, so that it can glow new vitality in contemporary design, has become a problem worth discussing [3]. Although the existing research has made some attempts in the combination of traditional art and modern design, most of them focus on the direct application of pattern, color and other elements, and there are few studies on the transfer and innovative application of the overall style. In addition, existing research mostly adopts traditional design methods and lacks in-depth exploration of emerging technologies such as AIGC, which limits the application scope and innovation of traditional art in modern design to a certain extent [4].

In the field of interior design, soft decoration design as an important means to create a space atmosphere and improve the quality of living, its innovation has been widely concerned by the industry inside and outside. The integration of traditional art elements can enrich the design language, but also bring a unique cultural flavor to the modern space [5,6]. At present, Song painting style has a wide range of demand in interior soft decoration, furniture design, product design, etc. [7,8]. The main consumer groups include high-end consumers who pursue cultural heritage, office buildings, high-end residences, B&Bs, hotels and other commercial Spaces. For example, the Wenyi Song Hotel successfully integrates the aesthetics of the Song Dynasty into the modern hotel design by extracting the design form from the Song painting and Constructing the French Style, showing the “level sense”, “line sense” and “exquisite sense” of the Song style architecture. In the research of AIGC design, some achievements have been made in the fields of product, cultural and creative, clothing design and facade design. For example, the application of AIGC technology in the design of cultural heritage cultural and creative products is able to generate high-quality image content and support efficient style transfer and image classification functions [9,10], through deep learning and feature extraction. In addition, AIGC technology has also demonstrated its application potential in traditional costume design, providing designers with valuable creative support, cultural exploration and design improvement tools, thus promoting the innovation and progress of traditional costume design [11]. In the field of architectural design, the application of AI image generation is more aimed at the stylized design and analysis of facades, which realizes various possibilities of building facade generation [12]. In general, AIGC technology has achieved certain results in the application and level, including image generation, text generation, audio generation, video generation and multi-modal generation, but the application in interior decoration design is still relatively few, and the support in the theoretical and method level is still not sufficient.

This study aims to systematically comb the historical background, artistic characteristics and style characteristics of Song paintings, analyze their modern value, explore the application of AIGC technology in the inheritance and innovation of Song paintings, and provide a new perspective for interior decoration design. The innovative contributions of this study are as follows: (1) a systematic workflow is proposed, which includes data collection and processing, prompt word generation, image generation and optimization, style transfer and detail control, and image application and optimization, ensuring the transformation from traditional art to modern design; (2) a cue word set is established, and AHP method is used to weight and grade the cue words to ensure that the generated image not only conforms to the artistic style of Song Dynasty painting, but also meets the needs of modern interior decoration design. (3) It provides a method for fine adjustment of Song style modern interior style images by using ControlNet technology to ensure the detail fidelity and artistic integrity of style transfer. These innovations not only enriched the creativity and methods of interior decoration design, but also provided an empirical research basis for the wide application and innovative development of the Song painting art in contemporary design. To help interior designers solve the problem of how to use AI tools to quickly and efficiently generate high-quality Song painting style interior soft decoration design. The research results show that AIGC technology can effectively integrate traditional art and modern design, open up a new path for the modern application of Song painting, enrich the creativity and methods of interior decoration design, and promote the wide application and innovative development of Song painting art in contemporary design.

2. Background

This chapter focuses on the cultural background of the Song Dynasty boundary painting, the application of AIGC technology in art design and its technical principles. This paper first sorts out the artistic characteristics and cultural value of Song Dynasty boundary paintings, and discusses how artificial intelligence technology breaks through the bottleneck of traditional research. Then the application of AIGC technology in painting, interior design and architectural design is analyzed, and its important role in improving design efficiency and promoting artistic innovation is pointed out. Finally, it briefly introduces the core principle of AIGC technology and reveals its innovative potential in artistic creation.

2.1. Cultural Background of Boundary Paintings in Song Dynasty

Song painting is an important part of Song Dynasty painting art. It depicts buildings and their environment, inherits the Tang and five Dynasties traditions, and forms a unique style [13]. The Song Dynasty lasted from 960 to 1279, divided into the Northern Song (960–1127) and the Southern Song (1127–1279), totaling 319 years. In the development process of Song Dynasty culture, the Northern Song Dynasty was dominated by academic culture, Neo-Confucianism germinated and gradually formed, and Confucianism had a new development in this period. At the same time, boundary painting was favored by the royal family and became a part of court painting [14]. The painting art of the Northern Song Dynasty also achieved high achievements, such as “Along the River During the Qingming Festival” and other works show the social style at that time. During the Southern Song Dynasty, boundary painting reached its peak, with mature techniques and a wide range of subjects, including not only palaces and temples, but also folk buildings and daily life scenes. Cultural focus gradually shifted to lyric painting, literator painting became the mainstream, emphasizing artistic conception and emotional expression. This change reflected the change of social environment in the Southern Song Dynasty and the literati paid more attention to inner emotions. Influenced by scientific and technological progress and urban planning, the development of boundary paintings in Song Dynasty provided rich expressive materials. At the same time, the development of painting theory also promoted the deepening of the artistic expression of boundary paintings [15,16]. The artistic characteristics of Song paintings are reflected in lines, themes and colors. The lines are precise and expressive, which not only depict the architectural structure, but also give the picture a sense of three-dimensional and vivid. The echo and foil between the lines create a sense of rhythm and enhance the artistic effect [17,18,19]. The lines of boundary paintings go beyond structural expression, assume a decorative role, and symbolize skyline and power. The painters innovated in the inheritance, adopted texturing method, dyeing method and other techniques, made the lines rich and diverse, and integrated into the new theme, reflecting the characteristics of the times [20,21]. In terms of subject matter, the boundary paintings of Song Dynasty not only included traditional buildings, but also covered life scenes such as markets, rural areas and transportation, reflecting the economic prosperity and social life of Song Dynasty [22,23]. Painters capture emerging urban landscapes, such as Hongqiao and commercial streets, showing a sensitivity to changing times. The fine depiction of ship design and bridge structure in the boundary paintings highlights the engineering technology of Song Dynasty, which makes the works have a distinctive mark of the times and historical value [24]. In the use of color, Song paintings preferred elegant colors, such as light ink, light purple, and flower green, which expressed architectural texture and quiet atmosphere. Through color shade and light and shade contrast, the performance space sense and light and shadow effect cleverly. Near the bright color, distant color fade, enhance the depth of the picture and three-dimensional sense [25]. While inheriting the tradition, Song paintings also innovatively introduced gold and red, which made the pictures more abundant and reflected the characteristics of the times. Color not only reproduces nature, but also carries specific cultural symbols and aesthetic feelings [26]. See Figure 1 for the common color scheme of Song paintings.

In terms of research methods, the traditional painting discipline mostly relies on literature research and physical inspection, lacking interdisciplinary research methods, resulting in a relatively single research perspective. The processing ability of large-scale art data is insufficient, and it is difficult to support the systematic mining of style rules. In terms of research content, traditional painting disciplines pay more attention to painting techniques and styles, and lack of research on the social and cultural background and historical value of painting. These limitations have given rise to the need to explore new techniques. At present, artificial intelligence technology can automatically parse artistic style through deep learning models, breaking through the boundaries of artificial experience. Able to model style composition in artistic creation and generate multiple interpretation possibilities. This provides new ideas and methods for artistic style analysis and helps to break through the limitations of traditional artificial experience [27]. For example, Generative Adversarial networks (GANs) can learn the line distribution and color level of traditional paintings to realize style transfer [28]. At the same time, AIGC technology can generate diversified design schemes in batches, providing an efficient path for the modern transformation of traditional art. Therefore, the integration of AI technology is not only an inevitable choice to deal with the bottleneck of traditional research, but also a key breakthrough to promote the rejuvenation of traditional painting in contemporary design.

2.2. Application of AIGC Technology in Artistic Design Creation

The development and influence of AIGC technology in contemporary painting are manifested in many aspects. First, advances in AIGC technology, including deep learning, generative adversarial networks (GANs) [29], Style transfer, adaptive algorithms, and augmented reality, among others, provide artists with new tools that allow them to incorporate traditional elements into their creations while exploring new forms of artistic expression [30,31,32]. The development of these technologies not only improves the productivity of professional designers, but also opens new creative doors for the general public, especially AI image generation technology, so that everyone has the opportunity to participate in the creation of art [33,34]. Although AIGC technology may mimic human creativity in some respects, human “sense-making” capabilities remain a unique strength of human creativity [35]. In the field of interior design, AIGC technology is applied by integrating Stable Diffusion and Dreambooth models [36], And popular deep learning algorithms [37], We demonstrate the potential of AI in room interiors generation, offering a significantly different approach from traditional labor-intensive techniques [38]. The current methods of AIGC not only improve the design efficiency [39], but also realizes the personalization of the design, and meets the needs of the design industry for efficiency and personalization [40,41]. The Stable Diffusion model is well known for its capabilities in the area of image generation and editing, capable of generating diverse and realistic images from conditional data [42,43]. The application of AIGC technology in the field of architectural design provides architects with new ways of expression and creative tools by imitating and going beyond traditional artistic styles [44,45,46]. The application of these technologies not only enhances the interactivity and engagement of art works, but also opens up new possibilities for the interaction between art and the audience [47,48]. The development of AIGC technology heralds a new trend of the integration of art and technology and promotes innovation in the field of art and design [49].

In summary, the application of AIGC technology in the fields of art design, interior design and architectural design not only improves the efficiency and quality of design, but also provides designers with new creation tools and methods [50,51]. The development of these technologies provides valuable experience and inspiration for the application of AI in a wider range of fields [52,53], It indicates the application and development of artificial intelligence technology in a wider field in the future [54].

2.3. Principle of AIGC Technology

At present, among the mainstream AIGC image generation tools, Midjourney and Stable Diffusion are two highly concerned high-performance platforms [55]. Midjourney is a text-based image generation tool that leverages advanced deep learning models and is able to quickly transform user input descriptive text into unique visual images. Midjourney is characterized by its strong ability to capture details and understand complex scenes, which makes the generated images not only conform to the text description, but also have a strong sense of art and creativity [56,57,58]. Stable Diffusion, on the other hand, is an open source text-to-image generation model that is popular for its efficient generation speed and high-quality image output. Stable Diffusion leverages diffusion models from deep learning, which generate images by gradually adding noise and then reversing the process to produce clear images. The advantage of Stable Diffusion is its stability and ease of use, where users can easily adjust parameters to control the style, resolution, and level of detail of images [43].

The core principle of AIGC technology is based on deep learning algorithms, especially technologies such as GANs, Variational Autoencoders (VAEs), Transformers and Diffusion Models [59,60,61]. These techniques learn the distribution characteristics of the data by training on large amounts of data, thus being able to generate new content without direct input. In GANs, there are a generator and a discriminator. The task of the generator is to create new data that is as close to the real data as possible, and the task of the discriminator is to distinguish between the generated data and the real data. Both compete with each other, improving their own performance until the discriminator cannot distinguish between generated and real data [62]. VAEs encodes the input data into a vector in a latent space through an encoder, and then samples from this latent space and reconstructs new data through a decoder. Transformer models, which incorporate multiple attention mechanisms, are used to automatically learn contextual relationships and syntactic structures between input text sequences. Diffusion Models define a Markov chain of diffusion steps that gradually add random noise to the real data (forward process) and then learn a backward diffusion process (inverse diffusion process) to build the desired data samples from the noise [63,64]. See Figure 2 for a schematic of the overall AIGC.

3. Methodology

This chapter describes the overall research steps and methods of AIGC used in Song painting, and explains in detail the principles, questionnaires and steps of AHP, the core method in the research process.

3.1. Steps and Methods to Construct the Innovative Application of AIGC in Song Painting

(1) This study makes full use of the rich resources of authoritative institutions such as the Palace Museum, the Zhejiang Provincial Bureau of Cultural Heritage, and the Large Collections of Chinese Paintings in the past Dynasties. Using high-resolution scanning technology and advanced image enhancement algorithms, these institutions have carried out comprehensive digital information storage for precious artistic works such as Song paintings, ensuring accurate restoration of color and complete retention of details. On this basis, the research team extensively collected digital information through on-site filming, official website downloading and network crawling, and carried out detailed classification according to the author and age, and a structured database of Song paintings is established.

(2) Cue word generation: Combined with the artistic characteristics and cultural semiotics of Song paintings (

Table 1. Comparison table of artistic characteristics and cultural symbols of Song paintings.

Artistic Features	Connotation of Cultural Symbols	Bibliography
Realism	The Confucian Spirit of Rationality in “Realization of Things”	[65]
Whitespace technique	Zen “Emptiness and Silence” and Literati Seclusion	[66]
Literators draw poems and books	The Independent Personality of the Scholars and Political Seclusion	[67]
Landscape artistic conception	The Taoist View of Nature and the Ideal of Seclusion	[68]
Flowers and birds are auspicious	Palace Auspiciousness and Folk Prayer Culture
The customs of the city	Commodity Economy and the Rise of Civic Culture	[69]

), the massive database of natural processing language model is used to generate cue words suitable for indoor soft decoration effect. In order to ensure the scientificity and representativeness of the cue words, the AHP is used to adjust and screen the weight of the generated cue words. The first is to build A hierarchy: the decision objective is decomposed into a goal level A, a criterion level B and a scheme level C. The target layer is “Song painting style hint words suitable for indoor soft decoration”. The criterion layer includes artistic style, cultural connotation, visual effects and other dimensions. The scheme layer is the specific cue word. Secondly, the pairwise comparison matrix is constructed: experts are invited to make pairwise comparisons of the factors, and the judgment matrix is constructed to determine the relative importance of each factor. Then calculate the weight and consistency test: the weight of each factor is calculated by the geometric mean method, and the consistency test is carried out (CR < 0.1 is passed). Finally, the hint phrases highly related to the style of Song paintings were selected. Through the AHP method, this study provides a scientific quantitative basis for the generation of prompt words to ensure their effectiveness and pertinence in art and design applications.

(3) Image generation and optimization: A new AI Song painting image is constructed using Midjourney and the cued phrases based on AHP weights. Specifically, based on Song painting database constructed in the previous stage, Midjourney is used to generate preliminary images according to the optimal cue phrases. Through the Generator and Discriminator Adversarial Training (GANs), multi-scale feature extraction and detail enhancement are performed on the images, and the image quality is gradually improved. The generated image is processed by super-resolution reconstruction and denoising to further optimize the visual effect and make it closer to the original artistic style of Song painting. Midjourney’s strong detail capture ability and artistic creativity have advantages in the initial stage of image generation. Finally, designers choose the weight of Song painting elements according to their needs, so as to realize the fine control of image style and content.

(4) Style transfer and detail control: The Stable Diffusion model and ControlNet technology are used to accurately control the style and content of the image, The appropriate control network type and weight are selected through different units, and the generated Song painting pictures are accurately transferred to the interior decoration design Focus on brushstroke texture reproduction, color level transition and composition proportion control to ensure the fidelity of details and artistic integrity of style transfer.

(5) Image application and optimization: The final generated style transfer image is optimized and adjusted, so that each design element can accurately convey the essence of Song Dynasty aesthetic art. Designers will constantly iterate and optimize the design scheme according to the actual application effect and user feedback. Firstly, the weight of different units is adjusted and optimized, followed by the modification of prompt words. Finally, the color, texture and composition of the generated image are processed in the later stage to enhance the fineness and cultural depth of the work, so that the design effect is in line with traditional culture, and can meet the practical and aesthetic needs of modern living. Figure 3 shows the overall research framework and process.

3.2. Weight Calculation of Cue Words in Song Paintings

AHP proposed by American operations research expert T.L. Saaty in the 1970s, is a decision-making tool that integrates qualitative and quantitative analysis, and is specially used to deal with complex decision-making problems containing multiple objectives [70]. This method decomposes the decision problem into different hierarchical structures, and quantifies the subjective judgment by mathematical model, so that the decision maker can evaluate and rank the various factors in each level, so as to make reasonable and logical decisions [71]. In this study, the role of AHP is to transform the complex and abstract artistic elements of Song paintings into instructions that can be directly and quickly accessed by designers. This process requires experienced professionals to convert artistic elements from images to textual descriptions, and provide accurate prompt words for subsequent image generation and design applications through quantitative evaluation and ranking. The AHP matrix construction process mainly includes the following steps:

(1) Construct the hierarchical structure of the content elements of Song paintings: AHP decomposes the decision-making problem into three levels: the goal level, the criterion level and the scheme level. The goal layer refers to the final decision purpose, while the criteria layer contains those factors that are essential to achieve these goals, and the scheme layer covers the specific implementation plans available for alternative.

(2) Construction of judgment matrix: through pairwise comparison of various factors at the same level, multiple experts are invited to determine the relative importance of the content elements of Song paintings. Saaty proposed a scale of 1 to 9 to measure the difference in importance of two factors, which specifically indicates the relative importance between two factors.

Table 2. Judgment matrix elements comparative scale table.

Serial Number	Meaning of Scale	Ratio
1	Equal importance	1
2	Slight importance	3
3	Fundamental importance	5
4	Very important	7
5	Absolute importance	9
6	The median of two adjacent judgments	2, 4, 6, 8
7	Indicates the importance of comparing the exchange order of the corresponding two factors	reciprocal

shows the comparison scale of the elements of the judgment matrix:

(3) Merge the expert matrix. Firstly, 25 experts are invited to fill in the questionnaire, and finally 21 valid questionnaires are collected. The evaluation standard is “which Song painting adjectives can represent the description words of interior decoration design elements in your mind”. The geometric mean method (square root method) was used to multiply the scoring matrix formed by 21 experts bitwise, and the m th root of the obtained product result was taken (m is the total number of experts, that is, 21). The unique integrated matrix $\overline{\mathrm{A}}$ is obtained, and the elements in $\overline{\mathrm{A}}$ are the geometric mean of m judgment matrices. Formula (1) is as follows:

$$\overline{\mathrm{A}} = {({\prod }_{k=1}^m{a}_{ij}^k)}^{{\displaystyle \frac{1}{m}}}$$

(1)

(4) When calculating the relative weight of the judgment matrix, the geometric mean method (that is, the square root method) is used to process the integrated matrix to determine the weight of each element. As shown in Equation (2):

$${\mathrm{W}}_i={\displaystyle \frac{{({{\displaystyle \prod }}_{j=1}^n{a}_{ij})}^{\frac{1}{n}}}{{{\displaystyle \sum }}_{i=1}^n{({{\displaystyle \prod }}_{j=1}^n{a}_{ij})}^{\frac{1}{n}}}},i=1,2,3\dots ,n$$

(2)

(5) Judge the consistency of the matrix

In the process of comparative analysis of indicators by experts, inconsistent results may be encountered, which requires consistency checking of the current judgment matrix to ensure the rationality and accuracy of index weight allocation. The Consistency ratio (CR), which is obtained by dividing the consistency index (CI) by the random consistency index (RI), is widely recognized in the academic community as a metric for assessing the consistency of a judgment matrix. When the CR value is lower than 0.1, the judgment matrix is considered reasonable and does not need to be adjusted. However, when the CR value reaches or exceeds 0.1, the expert must be asked to re-evaluate and modify the judgment matrix until the CR value decreases below 0.1. The formula for calculating CR is shown in Equation (3).

$$CR={\displaystyle \frac{CI}{RI}}={\displaystyle \frac{{\lambda }_{max}-n}{\left(n-1\right)RI}}<0.1$$

(3)

The formula for calculating CI is shown in Equation (4):

$$\mathit{CI}={\displaystyle \frac{{\lambda }_{max}-n}{\left(n-1\right)}}$$

(4)

λ_max is the largest eigenvalue of the judgment matrix, and the calculation formula is shown in Equation (5). The λ_max is the largest eigenvalue of the matrix, where $\overline{\mathrm{A}}$ is the judgment ensemble matrix, W is the weight vector, and [$\overline{\mathrm{A}}$W]_i is the i th component of the matrix [$\overline{\mathrm{A}}$W].

$${\lambda }_{max}={{\displaystyle \sum }}_{i=1}^n{\displaystyle \frac{{\left[\overline{\mathrm{A}}W\right]}_i}{n{W}_i}}$$

(5)

4. Result

The research results are described in the following parts. The first part is the establishment of Song Dynasty painting style cue word set, the second part is the artificial intelligence-assisted generation of Song painting interior soft decoration design elements, and the experimental results of style transfer. The results show that AIGC technology can effectively integrate the artistic style of Song Dynasty painting into the modern interior design, and show a variety of design effects. Finally, through comparative experiments, it verifies the advantages of using AHP method to generate prompt words in design, which can better meet customer needs and improve design efficiency and quality.

4.1. The Establishment of Song Painting Style Cue Word Set

4.1.1. Natural Language Generation Metrics

In the natural processing language model, ChatGPT has become a production efficiency tool with excellent learning mechanism, and the networking function can collect a large amount of relevant data. This time, ChatGPT 4.0 tool is used as a prompt word index production tool (

Table 3. ChatGPT dialog list of question instructions.

Dialogue	Answer Questions with ChatGPT Messages
Dialogue 1	Prompt: Suppose you are an expert on Song paintings ChatGPT: Well, if you have any questions about Song paintings, I’d be happy to help you from my knowledge. Whether it is about the artistic style of Song paintings, technical characteristics, representative painters, or appreciation of works and other aspects of the question, you can ask me at any time.
Dialogue 2	Prompt: I will use AIGC technology to generate Song paintings for interior soft decoration design. Please give me some relevant keywords, including but not limited to idioms. ChatGPT: To generate the image of Song painting style, it can be described from the following keywords: landscape of Song dynasty, fine brushwork, vivid ink rhyme, ink has five colors, smooth lines, outline in white, literary artistic conception, flowers and birds for spring, harmony of nature, light colour, delicate silk, brilliant stone, far and near high and low, profound artistic conception, flat and distant composition, emptiness and reality coexist……
Dialogue 3	Prompt: Divide these keywords into four broad categories. ChatGPT: Ok, I’ve divided these keywords into four main categories: 1. Style and technique: fine brushwork, ink has five colors, smooth lines, outline in white. 2. Topic and content: landscape of Song dynasty, literary artistic conception, flowers and birds for spring, natural harmony. 3. Color and texture: vivid ink rhyme, elegant color, delicate silk, bright stone color. 4. Composition and artistic conception: far and near high and low, profound artistic conception, flat distance composition, emptiness and reality coexist fine brushwork: emphasis on the fine depiction of Song painting details……

). In dialog 1, ask ChatGPT to assume an expert in Song painting, so that it can answer about Song painting in the following. Dialog 2 provides a series of keywords about generating AI Song painting, and dialog 3 summarizes these keywords and becomes an important indicator in the process of AHP (Figure 4).

4.1.2. Expert Questionnaire Interview

This questionnaire presents the responses of one of the experts about filling in, using the AHP to determine the importance of different design elements. The questionnaire divided the design elements into five main categories: style and technique, theme and content, color and texture, composition and artistic conception. Each category is subdivided into a number of specific design elements. By means of pairwise comparison, experts scored the importance of these elements, and the results were presented in the form of a judgment matrix. In order to show the importance of each element more intuitively, the questionnaire results are visualized as a heat map, where the degree of cold or warm of the color represents the level of weight. The cooler the color (e.g., Green): the lower the weight, i.e., the less important the feature is in the design. The warmer the color (e.g., red, yellow): the higher the weight, i.e., the relatively high importance of the feature in the design. (Figure 5).

4.1.3. Calculation Results

After the above indicators and formula calculation, the judgment matrix of the interior decoration design elements of Song painting and the consistency test results are as follows (

Table 4. The criterion layer judgment matrix and consistency test for soft decoration design of boundary painting in Song Dynasty.

	Style and Technique	Topic and Content	Color and Texture	Composition and Artistic Conception	λmax	CI	CR
Style and technique	1	0.897	1.680	0.758	4.061	0.020	0.023 < 0.1 The consistency check passes
Topic and Content	1.114	1	4.227	0.964
Color and texture	0.595	0.237	1	0.272
Composition and artistic conception	1.319	1.037	3.680	1

,

Table 5. Style and technique judgment matrix and consistency test.

	Fine Brushwork	Ink Has Five Colors	Smooth Lines	Outline in White	λmax	CI	CR
Fine brushwork	1	0.182	0.580	0.821	4.038	0.013	0.014 < 0.1 The consistency check passes
Ink has five colors	5.491	1	3.139	2.374
Smooth lines	1.723	0.319	1	1.026
Outline in white	1.219	0.421	0.975	1

,

Table 6. Subject and content judgment matrix and consistency test.

	Landscape of Song Dynasty	Literati Artistic Conception	Flowers and Birds Fight for Spring	Harmony of Nature	λmax	CI	CR
Landscape of Song dynasty	1	4.596	1.566	4.795	4.017	0.006	0.007 < 0.1 The consistency check passes
Literati artistic conception	0.218	1	0.278	1.384
Flowers and birds fight for spring	0.639	3.601	1	4.484
Harmony of nature	0.209	0.723	0.223	1

,

Table 7. Color and texture judgment matrix and consistency test.

	The Consistency Check Passes	Light Colour	Delicate Silk	Brilliant Stone	λmax	CI	CR
The consistency check passes	1	2.181	3.223	1.096	4.056	0.019	0.021 < 0.1 The consistency check passes
Light colour	0.458	1	1.137	0.508
Delicate silk	0.310	0.880	1	0.727
Brilliant stone	0.912	1.970	1.376	1

and

Table 8. Composition and artistic conception judgment matrix and consistency test.

	Far and Near, High and Low	Profound Artistic Conception	Flat Distance Composition	Emptiness and Reality Coexist	λmax	CI	CR
far and near, high and low	1	2.106	0.933	0.697	4.050	0.017	0.019 < 0.1 The consistency check passes
Profound artistic conception	0.475	1	0.521	0.436
Flat distance composition	1.072	1.920	1	1.423
Emptiness and reality coexist	1.434	2.293	0.703	1

):

According to the results in the consistency test of the judgment matrix, all the results meet the requirement of CR < 0.1, indicating that the consistency test has passed. According to the above content, it can be known that the weight of each level index is corresponding to the same level. In the sub-criterion level, each criterion level is regarded as an independent overall unit, and each sub-criterion level is regarded as the calculation unit in the index level, so that the weight distribution of each level index can be further determined. Finally, according to the calculation process of the above example, the weight calculation results of the first and second level indicators of the evaluation system established in this paper (

Table 9. Comprehensive weight value and ranking of interior soft decoration design elements of Song painting.

Target Layer A	Criterion Layer B	Weights	Index Layer C	Weights	Comprehensive Weight	Sorting
Song painting interior soft decoration design elements	Style and technique	0.233	Fine brushwork	0.114	0.027	14
			Ink has five colors	0.529	0.123	2
			Smooth lines	0.181	0.042	8
			Outline in white	0.176	0.041	9
	Topic and Content	0.329	Landscape of Song dynasty	0.468	0.154	1
			Literati artistic conception	0.104	0.034	11
			Flowers and birds fight for spring	0.346	0.114	3
			Harmony of nature	0.083	0.027	13
	Color and texture	0.100	Vivid ink rhyme	0.387	0.039	10
			Light colour	0.167	0.017	15
			Delicate silk	0.155	0.016	16
			Brilliant stone	0.291	0.029	12
	Composition and artistic conception	0.338	far and near, high and low	0.258	0.087	6
			Profound artistic conception	0.137	0.046	7
			Flat distance composition	0.312	0.105	4
			Emptiness and reality coexist	0.294	0.099	5

) can be obtained as follows.

As shown in Table 9, the results calculated by the AHP show that in Criterion layer B, topic and content as well as composition and artistic conception are the most critical in the design. In the index layer C, style and technique and the ink has five colors occupy the highest weight (0.529), followed by outline in white (0.176), which indicates that in the style and technique of Song painting, the ink has five colors and outline in white are the most representative elements, which are very important for inherits the artistic style of Song painting. In terms of topic and content, landscape of Song dynasty (0.468) and flowers and birds fight for spring (0.346) are the two most important themes, accounting for most of the weight of the criterion layer. This shows that in Song paintings, landscape and flowers and birds are the most popular themes, which has guiding significance for integrating Song painting elements into interior soft decoration design. In terms of color and texture, vivid ink rhyme (0.387) and light colour (0.167) are the two most important elements, which occupy a higher weight in the criterion layer of color and texture. This emphasizes that in the use of color, the vividness of ink rhyme and the elegance of color are the most distinctive elements in Song paintings. In terms of composition and artistic conception, flat distance composition (0.312) and Emptiness and reality coexist (0.294) are the two most important elements, which occupy a high weight in the criterion layer of composition and artistic conception. This reveals the artistic characteristics of Song paintings in the pursuit of profound artistic conception and the contrast between virtual and real in composition.

In summary, these data provide a scientific basis and guidance for designers to apply Song painting elements in interior soft decoration design, and help to better inherit and innovate Song painting art. According to these quantitative indicators, designers can choose and apply the elements of Song painting more scientifically and systematically, so as to realize the harmonious integration of tradition and modernity in interior decoration design.

4.2. Artificial Intelligence Assisted Generation of Song Painting Interior Soft Decoration Design Elements

Through the above calculation, four hint words suitable for Song painting in the interior soft decoration design elements are screened out according to the weight. Firstly, the weight is sorted from high to low in the criterion layer, and then the hint words of the first and second weight in the index layer are selected according to the weight order of the criterion layer. Finally, the first weight hint words are “flat distance composition, landscape of Song dynasty, ink has five colors and vivid ink rhyme”. The second weight of the cue word is “far and near, high and low, flowers and birds fight for spring, smooth lines and brilliant stone”. Import its keywords into the Midjourney text prompt and add proprietary instructions for quality and proportion. The visual quality of an image is affected by many factors, including the resolution and clarity of the resulting effect (e.g., high quality, 4 k, 8 k, masterpiece, super realistic, highly detailed). When constructing the cue word sequence, considering that the position of the cue word in the sequence has a significant impact on its weight, the higher the position, the greater the weight. Therefore, the cue words that can reflect the quality should be placed at the beginning of the sequence in the design to suit the output effect of the screen. The visual balance and overall effect of the screen is also affected by the size ratio, which refers to the aspect ratio of the screen (for example, typing “--ar16:9” in the description prompt will produce a 16:9 image). This time, the square image is generated and placed at the end of the prompt word, and “--ar1:1” is input to finally generate the corresponding image of the interior soft decoration design elements of Song painting (Figure 6).

4.3. Results of Style Transfer

In the Stable Diffusion model, due to its high computer configuration, we use Ali Cloud PAI-ArtLab platform to generate graphs, and use graph generation graph mode for style transfer of specific functions. Graph generation graph mode can restore the similarity of the original graph more. Firstly, the original image that needs style transfer is uploaded in the graph generation interface, and the style transfer of the preliminary generated AI image is carried out by using ControlNet technology. ControlNet allows the user to adjust the style of the image by setting the control grid, for example, to specify a specific stroke weight, color depth, or composition ratio. Then Upload the original image that needs style transfer in ControlNet Unit0, check Enable, Perfect Pixel mode, Upload and Allow preview, select Lineart as the control type, and lineart_realistic as the preprocessor. The Control weight Control Weighth and the Ending Control Step are both 1, and the control mode is balanced. Then, put the Song painting image generated in ControlNet Unit1, choose shuffle as the control type, and keep the other variables unchanged, and the control weight is about 0.7. Too high may over-transfer elements from the reference image, too low may make the effect less noticeable. The number of iteration steps is 30, and the realistic large model is selected for the large model. Finally, a series of rendings of Song Dynasty style indoor soft decoration are generated. During the migration process, ControlNet fine-tunes every detail of the image according to the parameters set by the user [72,73]. Among them, Shuffle plays an obvious role in style transfer in architectural design and interior design. The principle of Shuffle technology is that it randomly shuffles the pixels or blocks of the image through the algorithm, and then rearranges the style elements in the picture to the interior decoration design, so that the final output image not only retains the style of Song painting, but also meets the requirements of the interior decoration design. Through this technical means, the seamless transformation from traditional art to modern design can be realized, bringing infinite innovation possibilities to the field of interior design (Figure 7 and Figure 8).

According to the two groups of style transfer effects shown in Figure 6 and Figure 7, the application of AIGC technology in the interior decoration design can indeed realize the effective integration of the artistic style of Song paintings, and present diversified design effects. Specifically, in the first set of images, due to the higher weight assigned, the generated images are closer to the traditional style of Song paintings. It shows that by adjusting the weight parameters of AIGC technology, the presentation degree of traditional artistic elements in modern design can be accurately controlled, so as to meet the needs of different designers for the fusion of traditional and modern styles [74]. In the second set of images, the effect of Song paintings shows a more modern style due to the slightly lower weight. This shows that the sequence of prompt words affects the final presentation result of the image, and the user’s personal choice provides a variety of design styles for the interior decoration design, which enhances the controllability and innovation of the design. The experimental results show that the Shuffle control type can transfer the pattern elements in the original image, and all the elements in the original image are transferred to the final effect. The application of Shuffle technology plays a key role in style transfer. This technology integrates the mountain, tree, and flower and bird patterns in the Song Dynasty boundary paintings into the interior soft decoration design by randomly rearranging the image elements, rather than transferring the color blocks separately. Through the adjustment of weight, the larger the weight, the more obvious the elements retained in the Song paintings, and the more traditional style. The lower the weight, most of the original tone and texture are retained, and the more modern effect is obtained. It provides a rich style transfer effect scheme for designers. Through the weight adjustment of different units, the design works have modern aesthetics while maintaining traditional aesthetics. The application of AIGC technology not only improves the efficiency of design, but also broadens the creative space of designers [75]. Through technical means, designers can quickly try and compare the design effects of different styles, so as to more accurately grasp customer needs, improve design quality and customer satisfaction.

The above research results further verify the practical value of AIGC technology in the field of art design. This technology opens up a new path for academic research, provides technical support and innovation tools for interior design practice, and promotes the deep integration of traditional art and modern design. AIGC technology shows great application potential in the style transfer of boundary paintings in Song Dynasty. AIGC technology provides designers with a variety of effect schemes, promotes the organic combination of traditional art elements and modern design, and infuses new innovative vitality into the field of art design.

4.4. Comparative Experiment

In this experiment, we will set up a specific interior design soft decoration design scene, that is, design a Song Dynasty style modern style living room, in order to visually compare the differences in practical application between the use of AHP (experimental group) and Song painting style images generated without AHP method (control group). In the experiment, we will set up a real scenario of interior design project iteration: in the first stage, the owner puts forward the vague requirements of Song Dynasty style modern interior style; In the second stage, the owners continued to propose more detailed style preferences. Two sets of images will be generated based on cue words weighted by AHP screening and cue words selected randomly or empirically without AHP analysis, respectively. These images will be applied to the soft decoration design of the living room to show the artistic style of Song paintings. The experimental steps include prompt word generation, image generation, style transfer, and final visual comparison.

By displaying the images side by side and intuitively comparing the differences between the two, we can show the difference in the effect of the two groups of images in the soft decoration design (Figure 9).

In the first stage, the experimental group selected a group of cue words that could strongly represent the core artistic characteristics of Song paintings according to the AHP analysis results, that is, “Flat distance composition, Song Dynasty landscape, ink divided into five colors and vivid ink rhyme” with the highest weight in the cue words set. To generate a preliminary image scheme. These images can not only reflect the basic style of Song paintings, but also have high consistency and accuracy in style due to the systematic analysis of AHP. The control group inputted the prompt words “Song painting, Song dynasty, landscape painting” according to conventional experience, and generated multiple sets of ai Song painting images with different styles. The results of style transfer in the first stage show that diversified Song painting style indoor effects can also be generated only by random or empirically selected cue words in the stage of fuzzy demand of the owner.

In the second stage, when customers continue to propose more specific personal preferences based on the first stage scheme, the experimental group can take advantage of the flexibility and adjustability of AHP cue phrases to respond quickly. For example, customers prefer more style. The experimental group realized the “natural agile” style by weight adjustment, that is, Song Dynasty landscape (0.468), natural harmony (0.083), vivid ink rhyme (0.387) and smooth lines (0.181) were added. While maintaining the white drawing and bone method of Song painting, the spatial soft decoration elements showed dynamic rhythm, and the matching degree between the dexterity of Song painting style and soft decoration scheme was improved. For the control group, the process was more difficult. In the scheme of using empirical prompts such as “natural + agile + Song Dynasty”, conflicts between modern abstract art and Song painting techniques appeared (such as geometric lines destroying the structure of white lines), the matching degree of core styles decreased, and the generated images were more random and variable. The consistency of design iterations is harder to ensure (Figure 10).

The experimental results show that the experimental group using AHP has shown significant advantages in interior decoration design. In the first stage, although the needs of the owner are relatively vague, the images generated by the experimental group have high consistency and accuracy in style, which can better reflect the core artistic characteristics of Song paintings and lay a solid foundation for the subsequent design. Although the control group can also generate diversified Song painting style interior effects, the accuracy and consistency of style are relatively weak. In the second stage, when the owners proposed more specific personal preferences, the advantages of the experimental group became more prominent. Due to the flexibility and adjustability of AHP prompt phrases, the experimental group can quickly respond to customer needs, easily adjust and iterate between different elements by adjusting the prompt words and their weights, and generate new image schemes, and the coherence of design iterations is effectively guaranteed.

5. Discussion and Conclusions

Through a deeper exploration of the application of AIGC technology in Song paintings, this study provides a new perspective and method for interior decoration design. The results show that AIGC technology has significant advantages in the modern application of Song paintings. It can effectively integrate traditional art and modern design, and provide rich ideas and methods for interior decoration design. It mainly presents three aspects of value:

• Innovative design of traditional cultural elements. Through the big data of natural language processing model to generate target images describing design features, designers can use AHP to weight and rank the generated prompt words, and filter out the most representative keywords. These keywords further guide Midjourney and Stable Diffusion to generate images in the style of Song paintings, and achieve style transfer through ControlNet technology, and finally integrate traditional elements into modern interior soft decoration design. Compared with traditional design methods, this method can quickly capture the artistic characteristics and style characteristics of Song paintings, and integrate them with modern design techniques to create design works that inherit the traditional essence and integrate modern elements. Such a design not only retains the essence of traditional culture, but also infuses it with the meaning of the new era.
• Break traditional workflows. By using AHP to determine the representative keywords of Song Dynasty paintings, the designer’s design thinking has been expanded and breakthrough, implement the workflow of “natural language model + AHP + AI image generation + style transfer”, and the artistic essence of Song paintings can be more deeply understood, and the creativity can be more freely displayed, which provides more possibilities for the extension design of traditional cultural elements, This workflow innovation enables designers to deal with traditional art elements more systematically, breaking through the simple stacking of elements in traditional design, and realizing the systematic transformation from traditional to modern.
• Promotion and expansion of commercial value. With Midjourney for generating Song painting images and Stable Diffusion for style transfer, the design end can generate many designs quickly to meet market demand for individualized and customized designs.
• In terms of theoretical value, through a systematic workflow, this study provides a methodological basis for the integration of traditional art and modern design. The application of AHP method provides a scientific basis for the quantitative evaluation of traditional art elements, and helps to promote the systematic research of traditional art in modern design. In addition, this study provides theoretical support for the application of AIGC technology in cultural heritage protection and innovative design, and provides a reference for future related research.

However, there are some deficiencies and limitations in this study. On the technical level, despite the outstanding performance of AIGC technology in generating images, more emerging technologies are needed to assist designers and enable more development and personalization of design workflows. For example, the generalization ability of the model is a concern, and although Midjourney and Stable Diffusion perform well in generating images, their generalization ability in different styles and scenarios still needs to be further verified. In addition, on the cultural level, it is necessary to dig more deeply into traditional culture, so as to better integrate art and new quality productivity. For example, when using AIGC technology to generate images, it is necessary to ensure that the generated images do not infringe copyright, while respecting the inheritance and protection of traditional culture.

In the future research, we will focus on exploring the construction of traditional art innovation design methodology system supported by AIGC technology, focusing on the following research directions: (1) Reconstruct and optimize the design process, deepen the research on the parameter regulation mechanism of AIGC technology in the extraction and translation process of Song painting elements, develop the special prompt word database and style transfer template database for interior designers, and make the generated results more in line with the functional requirements and aesthetic orientation of modern living space. (2) The systematic application of cultural symbols to establish the mapping relationship model between the artistic elements of the Song painting and the context of contemporary interior design. By constructing a cultural gene map containing architectural components, decorative patterns, spatial artistic conception and other dimensions, a dynamically updated digital material library is formed, which provides an underlying logical framework for AI generation with both cultural depth and design feasibility.

In a word, AIGC technology provides new possibilities for the inheritance and innovation of Song paintings. In the future, it is necessary to further explore the creation paradigm of human-computer collaboration to promote the sustainable development of traditional art in modern design.