Sustainable Application and Evaluation of the Novel Stingray Model in Non-Heritage Packaging: The Case of Clay Sculptures in Joon County

Featured Application

This study is based on the innovative Stingray model, combining perception engineering, the Analytic Hierarchy Process (AHP), and the TOPSIS method to establish a systematic and scientific design process for intangible cultural heritage (ICH) product packaging. Taking the packaging design of Henan county-level clay sculptures as an example, the study achieved innovation and sustainable development in packaging design, enhancing design efficiency and the rationality of design solutions. This method promotes the deep integration of generative tools with traditional design, facilitates the application of eco-friendly materials (such as wheat straw), effectively reduces the carbon footprint, and enhances product competitiveness and cultural value. The model demonstrates good versatility and is applicable to green packaging design for other intangible cultural heritage and handicrafts, supporting the modernization and sustainable development of intangible cultural heritage.

Abstract

Generative tools often lack the guidance of scientific design methods in the design of non-heritage products. This study proposes a new Stingray model, which collects perceptual vocabularies of modeling and other aspects by integrating the perceptual engineering method to clarify the design direction and establishes the design objectives by ranking the importance of the vocabularies using the Analytic Hierarchy Process (AHP) hierarchical analysis method. Taking the Joon County clay sculpture as an example, this study uses generative tools to achieve the innovation of packaging patterns, selects sustainable materials such as straw to complete the sustainable non-heritage packaging design, and verifies its feasibility using the TOPSIS method. The results show that the new Stingray model effectively integrates multiple design methods and solves the subjectivity and feasibility deficiencies of a single model. Meanwhile, the system-guided generative tool significantly improved design efficiency and simplified program adjustment. This study provides theoretical support for generative tools and opens a new path for the sustainable development of non-heritage packaging.

1. Introduction

In the context of the rapid development of digital technology and artificial intelligence, the rise of generative tools has provided unprecedented opportunities for modernization and innovation in Intangible Cultural Heritage (ICH) product design [1,2]. However, most existing research focuses on the technical implementation level and lacks guidance on systematic design methodology, resulting in two major bottlenecks in the generation of results [3]. First, the design direction relies on designers’ subjective experience, and it is not easy to quantitatively align emotional needs with rational goals, such as the imprecise communication of cultural semantics [4]. Second, the scientific assessment of design solutions is insufficient, and environmental benefits and economic feasibility are often neglected [5]. This “technology first, methodology lagging” paradigm restricts the synergistic optimization of packaging in terms of eco-friendliness and cultural inheritance [6]. It is evident that the choice of design method is crucial to product design. In the traditional design process, the double-diamond model is the most commonly used and proven design method. The new Stingray model constructed in this study draws on the divergent thinking model of the double-diamond model and incorporates multiple concepts to enhance the technical and commercial feasibility of solutions, helping product design become more flexible and efficient in the new era.

As a national intangible cultural heritage of Henan Province, clay sculptures in Joon County have carried rich folk culture and local characteristics for thousands of years [7]. This traditional clay sculpture not only demonstrates the living customs and aesthetic concepts of the Central Plains but also plays an important role in important occasions, such as festivals and rituals [8,9]. With their diverse shapes and gorgeous colors, they are a very distinctive type of handicraft and have the potential to become distinctive cultural and creative products. However, during field research, it was found that the circulation of Joon County clay figurines in other regions is not high, and one of the important reasons is that the current packaging of Joon County clay figurines is too simple and homogenized to stimulate consumers’ purchasing desire. Therefore, to make Junxian clay figurines more competitive in the cultural and creative market, optimizing the design process of the packaging of this traditional craft has become an urgent task.

Starting from the typical non-heritage product of clay sculpture in Joon County, this study innovatively proposes a new Stingray model, aiming to guide non-heritage packaging design scientifically. The model is based on the Stingray framework, combined with perceptual engineering methods to obtain users’ perceptual-cognitive information, and uses AHP hierarchical analysis to rank the importance of perceptual vocabulary to establish a reasonable design weighting system [10]. On this basis, the non-heritage packaging design of clay sculpture in Joon County is implemented by combining generative tools, and the design scheme is objectively evaluated and feasibility verified by the TOPSIS approximation of the ideal solution ranking method [11].

The new Stingray model overcomes the subjective limitations of traditional generative tools and provides a systematic and scientific design method, which significantly improves the efficiency of the design process and the reasonableness and appropriateness of the design results. The innovations of this study are reflected in two aspects. First, by introducing perceptual engineering and AHP hierarchical analysis, user perception information is quantified to improve the accuracy and relevance of the generative design. Second, the TOPSIS method is used to evaluate the design scheme and ensure the feasibility of the design results, which provides a new idea to solve the problem of the lack of effective verification means in traditional generative design. The objectives of this study were as follows:

• To explore the applicability of the new Stingray model to the sustainable design process of packaging.
• Facilitating the integration of design methods and generative tools is essential.
• Conduct a feasibility assessment of the packaging.

In summary, this study enriches the theoretical system of non-heritage packaging design and provides scientific design guidance for the application of generative tools, which improves the rationality and user suitability of design solutions [12]. In addition, the promotion of this research method will provide an important reference for the design of other non-heritage derivatives, help non-heritage culture achieve modern development and innovative design, and open up a new path for the inheritance of traditional culture and digital innovation [13]. The subsequent chapters are arranged as follows: the second part is the literature review, the third part is the research methodology, the fourth part is the research findings, and the fifth part is the conclusion.

2. Literature Review

2.1. Theoretical Framework

With the rapid development of generative AI technology, its application in ICH product design is becoming increasingly popular [14,15]. Studies have shown that generative tools can rapidly generate diverse design solutions through algorithms, significantly shorten the design cycle, and stimulate the creative expression of cultural symbols [16]. For example, some scholars have adopted the pattern generation technique of the Generative Adversarial Network (GAN), a deep learning framework consisting of a generator and a discriminator in competition, for the development of non-heritage derivatives such as Miao embroidery and cloisonné. The experimental results show that the use of generative tools greatly saves design time and achieves good design results [17,18].

In terms of generative tools, this study chose the design strategy of using Midjourney as the main tool, supplemented by other tools such as ChatGPT (https://openai.com/index/chatgpt/). Among existing studies, Midjourney is favored for its user-friendly interface and artistic style [19]. Compared with another mainstream tool, Stable Diffusion, Midjourney provides a stronger artistic sense and visual impact, which is suitable for creating diverse illustrations and conceptual art [20,21]. In practice, Park contributed to the development of apparel design by using Midjourney to create color combinations appearing in apparel images in the AI image creation tool for fashion product advertisements, bespoke fashion style suggestions, and design development [22]. Sukkar et al. generate descriptive text through Midjourney to generate creative images of Islamic architectural heritage, widening the preservation path for architectural heritage preservation [23]. Kim et al. Exploring the possibilities of using AI-generated texts and images such as ChatGPT and Midjourney in creating interactive fairy tales, the results demonstrate that the use of AI technology provides new insights into new media art and contributes to children’s interactive educational material development [24]. In summary, the emergence of Midjourney provides designers and creators with new ways of creating and thinking, promoting the development and diversification of artistic creation [25,26]. In addition, Midjourney’s personalization options, such as built-in stylization parameters, allow users to adjust the design effect more flexibly, particularly for the needs of clay packaging patterns [27].

At the level of research methodology, design can be divided into pre-, mid-, and post-design processes [28]. In the pre-design stage, perceptual engineering was introduced into the field of Non-Heritage Resource Management (NRM), which extracts users’ perceptual needs for shape, color, and other dimensions through the semantic difference method to provide objective data support for design [29]. Perceptual engineering is an approach for studying how human perception, emotion, and experience influence the design of products and systems [30]. The field involves the incorporation of perceptual and emotional factors into the product development and design processes to create products that are more responsive to user needs and expectations [31]. At the center of this is the application of specific engineering techniques to transform the perceptual imagery of “people” into the design characteristics of “things” [32]. In perceptual engineering, researchers and designers often use various methods, including user surveys, field studies, prototype testing, and psychological experiments, to better understand user experiences and expectations [33]. By integrating perceptual factors into the design process, more popular and successful products can be produced [34]. For example, Wang et al. used perceptual engineering to obtain consumers’ perceptions of tea packaging and ultimately designed excellent tea NRL packaging designs based on user needs, enhancing branding [35]. Greek scholars used perceptual engineering to investigate the emotional relationship between various design elements and saffron Non-heritage Related Labeling (NRL) packaging [36]. The most important design elements in the existing saffron packaging in the Greek market were explored, laying the foundation for the subsequent entry of saffron packaging into the international market. Nasution et al.’s use of perceptual ergonomics to explore visual design elements and structural design in the design of packaging for aloe vera bottles was used for the development of a product based on the consumer’s perceptions by relating personal experience to the product’s attributes [37]. It can be seen that perceptual engineering can create products and experiences that have a deeper interpersonal connection beyond technical functionality and can find and define the type of solution that is suitable for setting goals and gathering intelligence upfront, a step that will determine whether the final goal can be achieved.

AHP hierarchical analysis is often used to select elements from a large number of elements in the middle of the design process [38,39]. The method is based on integrating factors related to the qualitative problem and the decision-making process and constructing a recursive hierarchical model based on the dominance relationships, expanding the objective level to the criterion, sub-criterion, and program levels, and solving the weights of factors at each level in turn by building a judgment matrix [40]. For example, Ali et al. used multi-criteria decision analysis using AHP hierarchical analysis to select the natural fiber enhancer with the highest level of user satisfaction from various types of commonly used plant fibers for food packaging [41]. AHP hierarchical analysis was used to understand the criteria important for customers to choose a beauty brand, which is beneficial for beauty brand companies to develop strategic objectives [42]. Porto De Lima et al. used AHP Hierarchical Analysis to obtain different experts’ opinions from the obtained elements of mechanical packaging design; the decision maker’s hesitation in designing was solved [43]. It can be seen that AHP hierarchical analysis can inform the selection of design metrics and help capture the characteristics, feasibility, and ultimate deliverability of the solution by exploring the problem and solution.

The TOPSIS method plays a crucial role in the later stages of design to verify the feasibility and scientific validity of the solution [44]. TOPSIS is a multi-attribute decision analysis method for solving problems involving multiple conflicting decision criteria [45]. The core idea is to select the optimal solution by finding an ideal solution and a negative ideal solution and comparing the advantages and disadvantages of each [46]. In recent years, a large number of scholars have introduced TOPSIS into design studies for solution evaluation, for example, Rahim et al. [47]. The TOPSIS methodology is used in the selection of materials for automotive body panels, which can consider not only uncertainties associated with qualitative judgments but also possible uncertainties in the measurements of quantitative or qualitative parameters present in the assessment of safety, health, and environmental risks. Bekesiene et al., on the other hand, proposed the use of the TOPSIS method for assessing the quality of distance learning surveys, validating the scientific validity of the decisions [48]. TOPSIS provides a quantitative and structured decision-making tool for decision-makers, which is particularly suitable for solving problems with multiple evaluation criteria [49]. In conjunction with this study, the perceptual engineering method, AHP hierarchical analysis method, and generative tools only work on the design and cannot understand whether the final design solution meets the user’s needs; therefore, the introduction of the TOPSIS method is helpful to complement the integrity of the design process.

2.2. Conceptual Framework

The model adopted in this study is inspired by the Stingray model, a conceptual framework proposed by the Board of Innovation [50]. This model is an evolution of the widely recognized Double Diamond Model, which was originally developed by the UK Design Council to represent the process of design thinking through four stages: Discover, Define, Develop, and Deliver. The Double Diamond emphasizes divergent and convergent thinking, encouraging designers to first explore the problem space broadly before narrowing down to specific solutions. However, it has been criticized for being time-consuming and susceptible to cognitive biases during decision-making [51].

In response to these limitations, the Board of Innovation proposed the Stingray model, which reconfigures the design process to address the challenges of speed, objectivity, and complexity—especially relevant in an era dominated by generative AI. Although the Stingray model has not been extensively discussed in academic literature (Figure 1), it introduces a tri-phasic structure—Training; Development; and Iteration—that mirrors the learning and refinement processes found in AI systems. The “Training” phase involves collecting data and identifying core insights, the “Development” phase focuses on generating and selecting ideas, and the “Iteration” phase emphasizes rapid testing and refinement [52,53].

This study refines and extends the Stingray model, proposing a modified framework tailored specifically for packaging design in the context of generative AI. The model is adopted here because packaging design increasingly demands adaptive creativity, accelerated iteration, and data-informed decision-making—all of which align with the strengths of the Stingray approach. By integrating AI capabilities, the improved model aims to facilitate faster identification of design problems and more efficient generation of optimized solutions while minimizing human bias and cognitive load.

The new Stingray model, which integrates perceptual engineering, AHP hierarchical analysis, and generative tools, is not only a way of thinking but also a set of design processes (Figure 2). The innovation lies in the inclusion of perceptual engineering in the “training” phase, which transforms users’ perceptual preferences into executable design parameters and measures the perceptual evaluation of the packaging through questionnaire collection, clarifying the direction of improvement and thus guiding the design of packaging designs that are more in line with users’ psychological needs and preferences [54]. In the “development” phase, AHP hierarchical analysis was used to rank the perceptual vocabulary collected in the previous phase and to judge the importance of each indicator based on the scores provided by experts [55]. In the “iterative” stage, ChatGPT (GPT-4, OpenAI) and Midjourney (V5.2) were used to input the design instructions for the AI-generative design. Artificial Intelligence enhances the iterative potential of the product at a later stage, and it is possible to change the design instructions to satisfy the customer’s needs at different stages of the process, which is much more cost-effective and time-saving than traditional design methods [56].

3. Research Methodology

3.1. Sample Size

3.1.1. Sensory Vocabulary Collection

To ensure the comprehensiveness of the selected samples, 50 samples of non-heritage products were initially obtained from various provinces at the beginning of the study by searching for non-heritage-related information and reviewing non-heritage packaging literature [57]. Excluding the interference of fonts and materials, we used the image classification method to classify the morphology, delete similar samples, and select more representative sample images [58]. In the end, a total of 15 sample images were selected, and the selection of sample images is shown in Figure 3.

The sample size used for the perceptual vocabulary collection was 100 people. To improve the quality and representativeness of the sample to more accurately reflect the real feelings and needs of the target users, the specific sampling method was as follows.

• Stratified sampling was used to collect perceptual vocabulary. Participants were stratified by gender (male and female), age group (e.g., 18–25 years old, 26–35 years old, 36 years old and above), and geographic region (e.g., urban versus rural) to reflect a wide range of user characteristics and reduce the impact of potential cultural differences on the results.
• The sample was selected considering the cultural background and consumption habits of different regions. By increasing the number of participants from each province, we ensured that we could represent the understanding and preferences of NCS in different regions, which in turn enhanced the external validity of the findings.
• In the process of participant selection, the study explicitly excluded non-target users (e.g., people who lack interest in products or relevant practitioners in professional fields) to obtain a more precise and perceptive vocabulary for the questionnaire. Before collecting information, the participants were provided with a brief background on products to ensure that they had a certain level of knowledge in their responses, which will help to enhance the credibility of the data.

These measures will provide a reliable basis for subsequent design decisions and broaden the impact of the research results in practical applications.

3.1.2. Perceptual Vocabulary Extraction

The collection of emotional vocabulary was conducted through an online survey. Non-targeted users were first excluded, and the final selection of 100 participants was approximately evenly divided between males and females, accounting for about 50 percent each. In terms of age distribution, 30 percent of participants were in the 18–25 age group, 40 percent in the 26–35 age group, and 30 percent in the 36-and-over age group. The study intentionally focused on the 18–35 demographic, as this cohort represents the most active user base in the context of digital product consumption and generative AI interaction, especially in areas such as e-commerce, packaging design preferences, and AI-enabled interfaces. While the 36-and-over group was not excluded, their smaller representation reflects the study’s emphasis on user groups most likely to engage with AI-augmented design processes. Future studies may further explore intergenerational differences in perception and engagement to provide a more comprehensive view.

In terms of geographical distribution, this study ensured that the proportion of urban and rural participants was relatively balanced, with a certain proportion of representative participants from each province to reflect diverse cultural backgrounds. Next, the testers were arranged to fill in the evaluation vocabulary on the 15 NRH product packaging test samples screened with regard to the stylistic, functional, material, and color aspects and to elaborate on their intuitive feelings towards the products, with no less than five perceptual vocabularies for describing the NRH product packaging required for each aspect [59]. The collected words were then transformed into adjectives to describe NRH product packaging, and focus group discussions were organized to eliminate words that were not suitable for describing NRH product packaging, such as worldly, wasteful, and useful, or those with repetitive semantics and ambiguous meanings. Then, 10 words for each of the aspects of shape, function, material, and color were obtained, making a total of 40 suitable perceptual vocabularies [60]. The focus group continued to be organized to score each of the 40 perceptual vocabularies using a 1–7 scale, with higher scores meaning that the perceptual vocabulary is more suitable. Finally, the 12 perceptual vocabularies with the highest mean scores were selected for the study [61]. The finalized 12 words, in order, are “Innovative”, “Straight”, “Stable”, “Practical”, “Versatile”, “Safe”, “Sustainable”, “Lightweight”, “Soft”, “Personalized”, “Cutting-edge”, “Interesting”.

3.2. Construction of the AHP Model

3.2.1. Hierarchical Modelling

According to the hierarchical division based on the interrelationship and affiliation between the elements, the optimized design of non-heritage packaging design of clay figurines in Joon County is divided into three levels: target, guideline, and indicator. The target layer was the optimal design of non-heritage mud sculpture packaging, the criterion layer was composed of shape, material, function, and color, and the indicator layer consisted of 12 perceptual vocabularies (Figure 4).

3.2.2. Constructing Judgment Matrices

A judgment matrix was constructed based on a two-by-two comparison of scores by experts. The judgment matrix can evaluate the relationship between elements at the same level and is constructed as follows:

$$A=\left[\begin{array}{ccc}{a}_{11}& \dots & a_{1n}\\ ⋮& \ddots & ⋮\\ a_{n1}& \dots & a_{nn}\end{array}\right]$$

The above equation indicates the importance of element I relative to element j at the same level, and vice versa, is 1/$a_{ij}$. The 1–9 scale method was used for labeling, and the specific scoring scale is shown in

Table 1. Scale and description.

Scale	Definition	Meaning
1	equally important	Indicates that both factors are of equal importance to the upper factor
3	slightly important	Indicates that element i is slightly more important than element j for the factor of the previous level
5	importance	Indicates that element i is more important than element j for upper-level factors
7	obvious importance	Indicates that element i is significantly more important than element j for factors of the upper level
9	very important	Indicates that element i is more important than element j for factors of the upper level
2, 4, 6, 8	median	Represents the middle value of two adjacent grades

.

3.2.3. Hierarchical Single Ordering

A hierarchical single sort is used to find the product Mi of each row of the two-by-two judgment matrix, then to open m times, normalize the data after the square, and finally to find the eigenvalues and take the maximum value. This time, we used the characteristic root calculation, which is as follows:

(1)

Compute the product M_i of the elements of the judgement matrix:

$$M_i={\displaystyle \prod _{j=1}^m{B}_{ij}}i=1,2,\dots ,m$$

(2)

Calculate the geometric mean W of each row M_i separately.

$$\overline{W_i}=\sqrt[m]{M_i}$$

(3)

Normalization:

$$W_i={\displaystyle \frac{\overline{W_i}}{{\displaystyle \sum _{j=1}^m\overline{W_j}}}}$$

(4)

Calculate the largest characteristic root of the judgment matrix.

$${\lambda }_{\max }={\displaystyle \sum _{i=1}^m\left({(AW)}_l/W_i\right)}$$

Based on the results of the hierarchical single sorting, the total weight of the decision objectives was calculated, the indicators were sorted by weight, and the decision was made based on the results of the total hierarchical sorting.

3.2.4. Consistency Check

(1)

Calculate the Consistency Index (CI):

$$CI={\displaystyle \frac{{\lambda }_{\max }-n}{n-1}}$$

where ${\lambda }_{max}$ represents the maximum eigenvalue of the judgment matrix, and there is a positive correlation between n and the CI value; as n increases, the CI value also increases.

(2)

Look up the corresponding Random Consistency Index (RI) from a table (

Table 2. Average random consistency index (RI).

n	1	2	3	4	5	6	7	8	9
RI	0	0	0.58	0.90	1.12	1.24	1.32	1.41	1.45

).

(3)

Calculate the Consistency Ratio (CR):

$$CR={\displaystyle \frac{CI}{RI}}$$

If the CR is less than 0.10, the result is considered reasonable; otherwise, adjustments must be made.

3.2.5. Determination of Weights

This study aimed to determine the importance of indicators at various levels of intangible cultural heritage packaging design. Five experts with extensive experience and professional knowledge in related fields were invited to participate in offline scoring to ensure the scientific and reliable nature of the assessment results. The test subjects included two professional designers, two visual communication professors, and one enterprise manufacturer. These test subjects have rich experience and professional knowledge in their respective fields and are able to provide reliable opinions for the study. Among them, the two professional designers have over five years of industry experience, specializing in industrial design and packaging design. The two visual communication professors are from well-known universities in China, with over ten years of teaching and research experience, focusing on visual communication and brand design. The manufacturing company has over ten years of practical experience in the production of intangible cultural heritage products, with a deep understanding of market demands and consumer preferences.

In the scoring process, the hierarchical analysis method was used to conduct the 1–9 degree scale commonly used for evaluation. The experts scored each indicator based on their full understanding of its background. After the summary analysis, a two-by-two judgment matrix was obtained (Table 3).

(1)

Criterion-level judgment matrix and weight.

Table 3. Judgment matrix and weights for the optimal design of clay figurine packaging.

Intangible Cultural Heritage Packaging Design	Stylistic	Functional	Material	Color	wi
Stylistic	1	2.449490	4.820571	8.050305	0.537995
Functional	0.408228	1	3.722419	6.160141	0.301378
Material	0.207429	0.268636	1	3.223710	0.112171
Color	0.124234	0.162338	0.310178	1	0.048456
Consistency test	$\lambda max$ = 4.105810, $CI={\displaystyle \frac{{\lambda }_{\max }-n}{n-1}}$$={\displaystyle \frac{4.105810-4}{4-1}}$ = 0.035270
	When n = 4, the average random consistency index RI = 0.90
	$CR={\displaystyle \frac{CI}{RI}}$$={\displaystyle \frac{0.035270}{0.90}}$ = 0.039189 < 0.1, conformance check passed

Table 3. Judgment matrix and weights for the optimal design of clay figurine packaging.

Intangible Cultural Heritage Packaging Design	Stylistic	Functional	Material	Color	wi
Stylistic	1	2.449490	4.820571	8.050305	0.537995
Functional	0.408228	1	3.722419	6.160141	0.301378
Material	0.207429	0.268636	1	3.223710	0.112171
Color	0.124234	0.162338	0.310178	1	0.048456
Consistency test	$\lambda max$ = 4.105810, $CI={\displaystyle \frac{{\lambda }_{\max }-n}{n-1}}$$={\displaystyle \frac{4.105810-4}{4-1}}$ = 0.035270
	When n = 4, the average random consistency index RI = 0.90
	$CR={\displaystyle \frac{CI}{RI}}$$={\displaystyle \frac{0.035270}{0.90}}$ = 0.039189 < 0.1, conformance check passed

(2)

Indicator-level judgment matrix and weights

A comparison of the importance of each perceptual term under the indicator layer is presented in

Table 4. Modeling judgment matrix and weight.

Stylistic	Innovative	Straight	Stable	wi
Innovative	1	9.654894	5.241483	0.768368
Straight	0.103568	1	0.436776	0.074017
Stable	0.190792	2.289428	1	0.157615
Consistency test	$\lambda max$ = 3.005232, $CI={\displaystyle \frac{{\lambda }_{\max }-n}{n-1}}$ $={\displaystyle \frac{3.005232-3}{3-1}}$ = 0.002616
	When n = 3, the average random consistency index RI = 0.58
	$CR={\displaystyle \frac{CI}{RI}}$ $={\displaystyle \frac{0.002616}{0.58}}$ = 0.004510 < 0.1, conformance check passed

,

Table 5. Function judgment matrix and weight.

Functional	Practical	Versatile	Safe	wi
Practical	1	2.352158	0.334156	0.240195
Versatile	0.425124	1	0.195530	0.113589
Safe	2.992556	5.114623	1	0.646216
Consistency test	$\lambda max$ = 3.011348, $CI={\displaystyle \frac{{\lambda }_{\max }-n}{n-1}}$ $={\displaystyle \frac{3.011348-3}{3-1}}$ = 0.005674
	When n = 3, the average random consistency index RI = 0.58
	$CR={\displaystyle \frac{CI}{RI}}$ $={\displaystyle \frac{0.005674}{0.58}}$ = 0.009783 < 0.1, conformance check passed

,

Table 6. Material judgment matrix and weights.

Material	Sustainable	Lightweight	Soft	wi
Scheme 1.	1	4.373448296	2.352158045	0.610087
Lightweight	0.228652526	1	0.698799165	0.152220
Soft	0.425150001	1.430940461	1	0.237693
Consistency test	${\lambda }_{\max }$ = 3.007611, $CI={\displaystyle \frac{{\lambda }_{\max }-n}{n-1}}$ $={\displaystyle \frac{3.007611-3}{3-1}}$ = 0.003805
	When n = 3, the average random consistency index RI = 0.58
	$CR={\displaystyle \frac{CI}{RI}}$ $={\displaystyle \frac{0.003805}{0.58}}$ = 0.006561 < 0.1, conformance check passed

and

Table 7. Color judgment matrix and weight.

Color	Personalized	Cutting-Edge	Interesting	wi
Personalized	1	2.550849001	0.334155847	0.243698
Cutting-edge	0.392002819	1	0.185948311	0.107366
Interesting	2.992555739	5.378268785	1	0.648936
Consistency test	$\lambda max$ = 3.013647, $CI={\displaystyle \frac{{\lambda }_{\max }-n}{n-1}}$ $={\displaystyle \frac{3.013647-3}{3-1}}$ = 0.006824
	When n = 3, the average random consistency index RI = 0.58
	$CR={\displaystyle \frac{CI}{RI}}$ $={\displaystyle \frac{0.006824}{0.58}}$ = 0.011765 < 0.1, conformance check passed

.

The combined weights were obtained by multiplying the guideline layer weights by the indicator layer weights, and the weight results are presented in

Table 8. Combined weights of indicators.

Norm	Combined Weights
Innovative	0.4092
Safe	0.1942
Stable	0.0842
Practical	0.0725
Sustainable	0.0708
Straight	0.0396
Versatile	0.0343
Interesting	0.0322
Soft	0.0277
Lightweight	0.0177
Personalized	0.0122
Cutting-edge	0.0054

.

Based on the perceptual vocabulary and AHP model calculation results, the optimization direction and cue words are summarized, in which the optimization direction is shape > function > material > color. In terms of modeling, the first prompt word is “innovative”, which breaks the form of the original packaging using plastic bags or simple boxes; the second is “stable”, which highlights the sense of historical heaviness; and the last one is “straight”, which echoes the curve of the clay figure itself;. The flatness of the packaging can echo the curve of the clay sculpture itself, combining static and dynamic elements. Functionally, the first word to be considered is “safe”, as the characteristics of the clay sculptures themselves are more fragile, so safety is considered a prerequisite, followed by “practical”, always adhering to the purpose of practicality in design, and finally, “versatile”. The second is “practical”, which is always aimed at practicality in design. The last is “versatile”, in which various functions can increase the attractiveness of packaging. The prompt word pursued in material was “sustainable”, sustainable material is conducive to the harmony of the environment, followed by “soft”, and finally “light”. The prompt word paid attention to in color was “interesting”. The cue word is “interesting”, and the box needs to be highly recognizable and decorative, followed by “personalized” and “cutting-edge”. Therefore, in subsequent designs, more attention will be paid to “innovative”, “safe”, “sustainable”, and “interesting” designs, and other design requirements will be supplemented.

3.3. Generative Design

3.3.1. Establishment of API Virtual Search Engine

The Google material library is characterized by its diversity, and to use clay sculptures as reference comparisons for generating images in Midjourney, a specialized virtual search engine was created on the Google Cloud Platform using the API interface provided by Google. The core function of this virtual search engine is to interact directly with Google’s search service through a programming interface (API), allowing the designer to customize the search and access images and materials related to clay figurines [62]. The advantage of this technological tool lies in the flexibility of the API and the powerful search capability to efficiently retrieve specific styles of clay sculptures from Google’s vast open-source database and create a digital database in the Midjourney system. This allows designers to gain a comprehensive understanding of the cultural intersections and diversity of clay sculpture art nationwide [63].

3.3.2. Using ChatGPT to Transform the Cue Words Further

To transform content-based prompts into standardized prompts, ChatGPT was utilized as a key assistant for Midjourney prompt generation to further transform pre-summarized Chinese prompts [64]. The core task of ChatGPT is to provide comprehensive support for prompt generation within the Midjourney framework. By leveraging ChatGPTs language generation capabilities, the aim is to improve the quality and effectiveness of Chinese prompts, enabling the Midjourney platform to be more flexible and intelligent in assisting users with their creativity [27]. In this process, ChatGPT must be guided rationally and effectively to generate prompts that comply with the standards of the Midjourney framework to ensure that the generated prompts are not only grammatically and structurally correct but also relevant and informative [65].

The operation is based on the perceptual vocabulary for nudging, and the prompt words are entered in sequence to allow ChatGPT to further embellish and arrange them in order: sophisticated cover, bold illustration design, innovative design, flexible and versatile lightweight materials, elegant and soft, personalized color palette, engaging foreword, photo-realistic style, abstract graphics, and Thai colors (Thai colors are vibrant and match with clay sculptures’ usual black, red, green, white, pink, and yellow colors) and abstract shapes. ChatGPT was adjusted to produce the following cue words: an opened box with six small boxes inside, exquisite cover, bold illustration design, innovative design, flexible and versatile, lightweight material, elegant and soft, personalized color palette, interesting preface, photorealistic style, abstract graphics, Thailand color, and abstract shapes.

3.3.3. Midjourney Generates Images

The clay figurines called through the API interface were screened to identify six images that best represented the characteristics of clay figurines, processed into PNG format, and uploaded to the Describe command on the Midjourney platform [66]. Through this step, a link to the clay figurine images was successfully obtained, and based on the link combined with the English cue words obtained from the API call, innovative attempts were made. After six attempts, a set of brand-new images based on the shape of the clay figurines was preferred again through focus group voting. The initial trials resulted in a more satisfactory style, but because the initial instructions did not match the optimized box style, the instruction of an open box with six small boxes inside was chosen to be deleted, and the comparison before and after the adjustment was made (Figure 5).

After the adjustment of the four pictures, from the perspective of picture expressiveness and colorful interestingness, the third picture has more tension, both in the figurative expression of clay sculpture and abstract expression, and the coordination between the colors is also stronger; therefore, the third picture is chosen for the adjustment of the Vary (Strong) parameter (Figure 6). After repeated training, a complete set of decorative patterns that conform to the characteristics of clay sculptures and satisfy the importance ranking of the cue words was obtained (Figure 7).

3.3.4. Overall Package Presentation

(1)

Use of straw materials

Considering the principle that packaging materials should be adapted to local conditions, data related to the planting industry in recent years revealed that the actual planting area of wheat in Henan occupies the first place in the country [67]. Therefore, the use of wheat straw to make packaging for Henan’s non-heritage clay figurines has a unique advantage. In addition, the lightweight straw meets the demand for sustainable packaging. In this design, through the process of fibrillation, crushing, screening, flattening, hot pressing, drying, glue blistering, cutting, surface pre-treatment, disinfection, and sterilization of wheat straw raw materials, straw is turned into a treasure and used as the main material for the shell of the clay molding box, which, in addition to being a non-toxic and non-hazardous material, can also serve as a cushion in the event of a fall, which is in line with the need for safety [68].

(2)

Design of drawer structure

The illustration design of the clay figurine packaging was completed through the collaboration of ChatGPT and Midjourney. Through C4D modeling, the clay figurine packaging pattern was mapped, and the optimized clay figurine packaging was rendered (Figure 8). To realize the innovation and interesting purpose of the packaging box, a symmetrical double-door drawer gift box design was chosen, which is not only solid and firm but also enhances the grade of the product and the fun of using it. The inside of the box is divided into two layers; the first layer is a gift area, storing the clay used to make clay sculptures; consumers can buy the video teaching handmade to enhance the sense of experience of the product; the second layer is a clay sculpture storage area, surrounded by pearl cotton filling; a total of one large, two small, and three clay sculpture products can be placed (Figure 9).

The dimensions and structure of the box are shown in Figure 10.

3.4. TOPSIS Design Evaluation

Scenario 1 is the scenario of this study, Scenario 2 is the existing clay packaging in Joon County, and Scenario 3 is the clay non-heritage packaging in Huaiyang, as presented in

Table 9. The three packages.

Serial Number	Externally	Inside
Scenario 1
Scenario 2
Scenario 3

.

For the comparative evaluation of the three programs, 20 participants were invited to participate in the scoring, including 15 residents of Joon County and five experts in product packaging design. The evaluators were asked to use a seven-point Likert scale to score four important indicators and four sub-important indicators obtained from perceptual engineering for the three scenarios. Based on the scoring data, the average of all scores was calculated to form a preliminary decision matrix f (Table 10). Then, the positive and negative ideal solutions and their relative proximity to each scenario were obtained according to the TOPSIS operation process, and the scenarios were prioritized accordingly [69].

Step 1. The questionnaire results were homogenized to obtain an initial evaluation matrix denoted by f.

Table 10. Initial evaluation matrix.

Evaluation Indicators	Scenario 1	Scenario 2	Scenario 3
Innovative	5.4	5.2	3.2
Safe	5.2	4.8	4.9
Sustainable	5	4.9	4.2
Interesting	5.3	4.8	4.1
Stable	4.7	4.9	4.6
Practical	4.6	4.1	4.4
Soft	4.8	3.6	4.6
Personality	5.2	4.8	3.4

Table 10. Initial evaluation matrix.

Evaluation Indicators	Scenario 1	Scenario 2	Scenario 3
Innovative	5.4	5.2	3.2
Safe	5.2	4.8	4.9
Sustainable	5	4.9	4.2
Interesting	5.3	4.8	4.1
Stable	4.7	4.9	4.6
Practical	4.6	4.1	4.4
Soft	4.8	3.6	4.6
Personality	5.2	4.8	3.4

Step 2. The initial evaluation matrix is normalized to obtain the normalization matrix as follows:

$$R_{ij}={\displaystyle \frac{f_{ij}}{{\displaystyle {\sum }_{i=1}^m{f}_{ij}^2}}}(i=1,2,\dots ,m;j=1,2,\dots ,n)$$

Step 3. Calculate the weighted normalization matrix based on the target weights of the evaluation indicators:

$$u_{ij}=W_j{R}_{ij}(i=1,2,\dots ,m;j=1,2,\dots ,n)$$

where $W_j$ denotes the weights.

Step 4. Find the positive ideal solution and the negative ideal solution.

$$\begin{array}{l}{M}_j^{+}=\max \{u_{1j},u_{2j},\dots ,u_{n j}\}(j=1,2,\dots ,m)\\ M_j^{-}=\min \{u_{1j},u_{2j},\dots ,u_{n j}\}(j=1,2,\dots ,m)\end{array}$$

Then:

$$\begin{array}{l}{A}^{*}=(M_1^{+},M_2^{+},\dots ,M_m^{+})\\ A^{-}=(M_1^{-},M_2^{-},\dots ,M_m^{-})\end{array}$$

Step 5. The distance of each scenario to the ideal solution is calculated by Euclidean distance; the distance of each scenario to the positive ideal solution is S_i⁺, and the distance to the negative ideal solution is S_i⁻:

$$\begin{array}{l}{S}_i^{+}=\sqrt{{\displaystyle \sum _{j=1}^n{(u_{ij}-u_j^{+})}^2}} (i=1,2,\dots ,m)\\ S_i^{-}=\sqrt{{\displaystyle \sum _{j=1}^n{(u_{ij}-u_j^{-})}^2}} (i=1,2,\dots ,m)\end{array}$$

Step 6. The relative sticking point C_i of each scenario to the desired solution is calculated as follows:

$$C_i={\displaystyle \frac{S_i^{-}}{S_i^{+}+S_i^{-}}} ( i=1,2\dots , m)$$

Ultimately, the priority of the evaluation program is judged by the size of the Ci value, with larger values reflecting a higher program priority.

4. Research Findings

Based on the results of the calculation of the positive ideal solution, negative ideal solution, and relative closeness of each scheme (

Table 11. Calculation results.

Program	Positive Ideal Solution Distance (Si+)	Negative Ideal Solution Distance (Si−)	Composite Score Index (Ci)	Arranged in Order
Scenario 1	0.2555659	0.93240128	0.78487125	1
Scenario 2	0.65203719	0.65891042	0.50262148	2
Scenario 3	0.84767522	0.36236428	0.29946484	3

), Scheme 1 had the highest relative closeness ranking and was therefore rated as the best packaging design scheme for clay figurines. As an evaluation sample of the existing solutions in the market, they were all ranked lower than the non-heritage packaging design proposed in this study. This indicates that the novel Stingray model is highly feasible for non-heritage packaging design.

In this study, the main findings of sustainable design and assessment of non-heritage packaging of clay molding in Joon County based on the novel Stingray model are as follows:

(1)

Multi-method integrated modeling significantly improves design science and feasibility

The new Stingray model, constructed by integrating perceptual engineering, AHP hierarchical analysis, and TOPSIS, effectively solves the two bottlenecks of “insufficient quantification of perceptual needs” and “lack of sustainability assessment” in the application of generative tools. The perceptual engineering method extracts 12 core perceptual words from modeling, function, and other dimensions, such as “innovative” and “sustainable”, and after AHP weighting analysis (criterion weighting: modeling 0.538 > function 0.301 > material 0.112 > color 0.048), it is clear that the modeling is not enough to quantify the perceptual needs and the lack of sustainability assessment in the application of generative tools. (color 0.048), the design priorities of “innovative”, “safe”, “sustainable”, and “interesting” were clearly defined, which significantly reduced the number of design priorities, and the number of design priorities was reduced to “innovative”, “safe”, “sustainable”, and “interesting”. The TOPSIS evaluation results show that the relative closeness (Ci = 0.785) of this study’s scheme (Scheme 1) is significantly higher than that of the existing schemes in the market (Scheme 2, Ci = 0.503, and Scheme 3, Ci = 0.299), which verifies the validity of the model in balancing the goals of cultural expression and sustainability.

(2)

Generative tools and the scientific method synergistically drive efficiency optimization

Multiple control groups were set up, one using the traditional design method and the other using the generative design method, and the time to generate multiple design items was tracked separately to obtain the average generation time. It was calculated that the cycle time for generating NRL packaging patterns using generative tools saves at least 35% of the time compared to the traditional design time.

(3)

Enhanced Environmental Benefits from Sustainable Materials and Structural Innovations

Based on the advantage that Henan accounts for 28% of the country’s wheat cultivation area and is rich in straw resources, the use of a thermal compression molding process reduces the carbon footprint of packaging shells by 67% compared to traditional plastics and shortens the degradation cycle from 500 years to 6 months [70]. According to user surveys, the “natural texture” and “cultural affinity” of straw materials increased consumer satisfaction, confirming the market potential of bio-based materials in non-heritage packaging.

(4)

Methodological Innovation to Promote the Improvement of Design Theory System

The new Stingray model is the first to standardize the entire chain of “user perception analysis—target weighting decision—design generation—multi-dimensional verification”; providing a reusable design framework for generative tools. The method not only supports the efficient use of resources but can also be extended to other non-heritage items, such as ceramics and brocade, providing a systematic solution for the green transformation of traditional cultural products.

5. Conclusions

In this study, a packaging design method for intangible cultural heritage based on a novel Stingray model was constructed and verified by a practical application using clay sculptures in Joon County as an example. This study collected users’ perceptual vocabulary by introducing the perceptual engineering method and combined it with the hierarchical analysis method (AHP) for weight ranking to optimize the design decision-making process. Simultaneously, the innovative design of non-heritage packaging patterns was completed by combining the Midjourney generative tool and the TOPSIS method to evaluate different design scenarios objectively.

The results of the study show that compared with traditional design methods, the novel Stingray model provides scientific guidance in the application of generative tools to the design process of non-heritage packaging, which significantly improves the accuracy, rationality, and feasibility of designs. The specific contributions are summarized as follows.

• Theoretical Contribution: The first non-heritage design framework based on the Stingray model, integrating perceptual engineering, AHP hierarchical analysis, and the TOPSIS method, is proposed, which makes the process of non-heritage packaging design more scientific and systematic and improves the transparency and effectiveness of the design decision-making.
• Methodological contribution: The Stingray design model was optimized to better adapt to the characteristics of generative tools, thus improving the efficiency and controllability of the design process. This new design path provides feasible methodological support for the modernization and transformation of nonheritage products.
• Practical Contribution: Evaluating the generative design results using quantitative methods, we verified the feasibility of generative tools working in synergy with the scientific method and promoted the green and sustainable development of packaging for non-heritage products.

Based on this, the study further draws the following three important conclusions:

• The applicability of the new Stingray model in the sustainable design process has been verified: The study found that the new Stingray model supports the entire process from conceptual design to practical application of intangible cultural heritage packaging by integrating multiple methods. It demonstrates strong systematicity and scientific rigor, particularly in the early stages of user perception exploration and the later stages of solution validation, indicating its good applicability and promotional value in the sustainable design of intangible cultural heritage product packaging.
• Promoting the integration of design methods and generative tools is crucial: This study emphasizes that deeply integrating design science methods and generative tools is a key path to improving design efficiency and quality. Sensory engineering and the AHP method jointly optimized the accuracy of input data, while AI tools efficiently produced creative solutions, achieving a closed-loop system from “cognitive acquisition” to “solution output”.
• Feasibility assessment is an indispensable part of intangible cultural heritage packaging design: Using the TOPSIS method to analyze the relative proximity of the three packaging schemes, this study confirms that the proposed scheme has significant advantages in terms of innovation, safety, sustainability, and fun. This emphasizes the necessity of incorporating a scientific assessment system into intangible cultural heritage packaging practices, which is conducive to the implementation and market promotion of design results.

Given the successful application of this research methodology, this study concludes that the new Stingray has strong generalizability and transferability and can serve as a reference for the packaging design of other intangible cultural heritage or arts and crafts products worldwide. Especially when facing different cultural backgrounds and market demands, the idea of integrating perceptual engineering, hierarchical analysis, and generative tools can provide new perspectives for designing and promoting various types of ICH products. This cross-cultural design framework not only helps to enhance the market competitiveness of NRH products but also promotes the inheritance and innovation of traditional culture in a wider scope so that intangible cultural heritage can find a new space for survival and development in the process of globalization. Through the deepening and expansion of this research, we expect to promote the sustainable development and modernization transformation of non-heritage products and realize diversified cultural exchanges and integration.

Despite the results achieved in this study, there are still some limitations. First, the collection of user perceptual vocabulary relies on a specific sample group, and future research could consider expanding the data sources and incorporating more quantitative analysis methods to improve the adaptability and universality of the model’s results. Second, with the continuous development of generative AI, the enhancement of human-machine collaboration in the design process can be explored to improve the controllability and diversity of generative design and further promote intelligent design innovation in NRL product packaging. In addition, after a product is formally put on the market, the duration of the product’s extended life cycle can be assessed, adding empirical support to the superiority of a circular economy. These explorations will contribute to the sustainable development and modernization of ICH products.