AI-based cloud space resource management system for art teaching materials

Abstract

The present invention relates to the technical field of electronic digital data processing, and discloses an AI-based cloud space resource management system for art teaching materials, comprising: converting original art materials into parameterized seed data containing vector outlines, color primitives and grayscale guide maps through a material potential conversion module, and associating a lightweight rendering rule set; an instant response engine calls the rendering rule set based on interactive parameters to perform graphics operations on the seed data to generate visual images in real time; and a teaching interactive interface provides interactive parameter input controls. The present invention realizes a paradigm shift from static storage to dynamic response of art teaching materials, enabling a single basic material to derive diversified visual expressions in real time according to teaching needs. At the same time, a gradient compensation mechanism is used to ensure data stability under non-ideal acquisition conditions, providing flexible and reliable digital support for heuristic teaching.

Description

Technical Field

[0001] The present invention relates to an AI-based cloud space resource management system for art teaching materials, and belongs to the technical field of electronic digital data processing. Background Art

[0002] Current mainstream technologies use cloud-based databases to store static images or layered design files, and use a tagging system to retrieve materials. These solutions rely on pre-generated fixed material entities, forcing teachers to screen resources before class. If visual effects need to be dynamically adjusted during teaching, such as for real-time demonstrations of lighting changes or style transitions, they face systemic limitations: First, static materials cannot respond to the needs of real-time classroom interaction. For example, when students raise exploratory questions about the color changes of objects under backlighting, teachers can only interrupt teaching to retrieve new materials or rely on complex AI generation tools, resulting in a fragmented teaching process. Second, existing improvement solutions, such as refined tagging systems or cloud-based AI rendering, attempt to improve the flexibility of material retrieval. However, due to the need to frequently transfer large files or rely on remote computing, they still cannot meet the core demand for millisecond-level response in teaching.

[0003] Specifically, the existing technology has the following inherent defects: 1. The disconnect between materials and teaching time and space. There is a fundamental mismatch between static resource libraries and dynamic teaching scenarios, and materials cannot generate new visual expressions in real time as the class progresses; 2. Solutions that rely on background computing or network transmission are difficult to achieve delay-free interaction, which restricts the fluency of heuristic teaching; 3. In non-ideal acquisition environments (such as backlit shooting), traditional image processing can easily lead to distortion in material feature extraction, affecting the reliability of subsequent operations.

[0004] As the demand for interactive teaching grows, these shortcomings further expose the deep-seated contradictions in the data processing paradigm of existing technical architectures. Massive static storage and instant generation capabilities cannot be achieved simultaneously, and system scalability is limited. Therefore, how to build an art teaching material processing mechanism that combines real-time responsiveness, dynamic derivative potential, and environmental adaptability has become the technical problem to be solved by this invention. Summary of the Invention

[0005] The present invention provides an AI-based art teaching material cloud space resource management system, the main purpose of which is to solve the problem that static material management systems cannot generate dynamic visual effects in real time to meet the interactive needs of classroom teaching.

[0006] To achieve the above objectives, the present invention provides an AI-based art teaching material cloud space resource management system, comprising:

[0007] a material potential conversion module configured to receive raw art material, convert the raw art material into structured parameterized seed data in one go, and associate a set of lightweight rendering rule sets with the parameterized seed data; the parameterized seed data includes vector outline data defining the basic shape of the raw art material, color primitive data defining the basic hue of the raw art material, and a low-resolution grayscale guide map for suggesting basic texture and light and shadow relationships; the lightweight rendering rule set includes a shader program or drawing instructions for generating a visual image based on the parameterized seed data and interaction parameters;

[0008] An instant response engine, connected to the material potential conversion module, is configured to receive interactive parameters in a client environment and invoke a lightweight rendering rule set to perform graphics operations on the parameterized seed data, thereby generating and refreshing visual images in real time. Interactive parameters include light source position and color parameters for simulating lighting changes, as well as stylized parameters for simulating the texture of painting materials.

[0009] A teaching interaction interface is connected to the instant response engine, configured with input controls for providing interaction parameters, and sending the interaction parameters to the instant response engine; the vector outline data is layered, and the teaching interaction interface also provides step parameter input controls for controlling the number of rendering layers of the vector outline data.

[0010] Preferably, the material potential conversion module is further configured to: calculate the gridded brightness distribution variance of the original art material to identify uneven lighting; when the brightness distribution variance is greater than a set numerical threshold, the material potential conversion module is further configured to extract the characteristic change gradient of the light and dark boundary area in the original art material, and perform compensatory extraction on the global scale based on the characteristic change gradient to generate parameterized seed data.

[0011] Preferably, the teaching interaction interface is also configured to: monitor in real time the entropy value of the interactive information flow generated by the user through the touch input device, and judge whether the user is in an inspiration bottleneck state based on the entropy value; when it is judged that the user is in an inspiration bottleneck state, activate the set heuristic rendering rules in the instant response engine to superimpose a visual disturbance in the currently generated visual image.

[0012] Preferably, the material potential conversion module is specifically configured to: perform semi-automatic contour extraction on the original art material to generate an SVG vector path as vector contour data; and extract multiple key color values ​​of the main color blocks of the original art material as color primitive data.

[0013] Preferably, the lightweight rendering rule set is a set of shader programs or Canvas drawing instructions implemented based on the WebGL or WebGPU standard, configured to perform graphics operations based on set rules without relying on complex artificial intelligence calculations.

[0014] Preferably, the instant response engine is specifically configured to calculate and render the light and shadow changes within the vector outline data in real time through the Lambert lighting model when the light source position parameters and the light source color parameters change.

[0015] Preferably, the instant response engine is further configured to: when the stylization parameter is set to any one of watercolor style, crayon style or line drawing style, load the corresponding noise texture and adjust the edge smoothness of the rendering stroke to simulate the texture of the corresponding painting material.

[0016] Preferably, the teaching interaction interface is specifically configured to provide a draggable icon or slider as an input control for the interaction parameter when the teacher selects the material, and the dragging operation of the icon or slider is associated with the real-time change of the visual image.

[0017] Preferably, the entropy value of the interactive information flow is calculated by combining the change rate or irregularity of the following parameters: the change rate of the pressure value of the touch input device used by the user, the change rate of the pen speed, and the irregularity of the pen direction; wherein the entropy value of the interactive information flow is The calculation satisfies: ,in, Indicates the rate of change of pressure value, Indicates the rate of change of the stroke velocity. Indicates irregularities in the direction of the stroke, and To set the weight coefficient.

[0018] Preferably, the heuristic rendering rule is configured to automatically superimpose a recommended color in a set color scheme or a texture in a set artistic style on the visual image currently operated by the user when the user is in a state of inspiration bottleneck.

[0019] Compared with the prior art, the present invention has the following beneficial effects:

[0020] 1. By converting raw art materials into structured parametric seed data and associating it with a lightweight rendering rule set, the system fundamentally reconstructs the material calling logic. When teachers adjust light source parameters or stylization parameters through the interactive interface, the instant response engine directly performs graphics operations based on the seed data and rendering rules, allowing a single basic material to derive diverse visual expressions in real time. This mechanism avoids the temporal and spatial separation between traditional material libraries and teaching sites, enabling teachers to seamlessly generate targeted visual examples based on dynamic classroom needs, such as students' instant questions, transforming the teaching process from the presentation of preset resources to the co-creation of dynamic situations.

[0021] 2. For non-ideal lighting materials commonly found in real teaching environments, such as backlit images taken with mobile phones, the system diagnoses uneven lighting conditions through gridded brightness distribution variance and focuses on the characteristic change gradients in the light-dark boundary area for compensatory extraction. This mechanism forms a closed-loop collaboration with the seed data generation module. When conventional processing faces the risk of distortion, the gradient-guided algorithm automatically intervenes to infer global effective information from local physical optical features. This self-correction capability ensures the stable generation of vector contours and color primitive data, allowing the dynamic rendering function to maintain reliable output under diverse real-world acquisition conditions, avoiding the computing power burden brought by complex image enhancement technology.

[0022] 3. The teaching interactive interface continuously monitors changes in the entropy of touch input and builds a cognitive model of the user's creative state. When the entropy value continues to be higher than the dynamic baseline, the system automatically activates heuristic rendering rules, such as superimposing complementary color layers that conform to color harmony theory on the current visual image or injecting artistic stylized texture disturbances. This mechanism is deeply coupled with the real-time rendering engine. The state recognition module captures implicit bottleneck signals, and the rendering engine converts them into non-invasive visual cues through preset rules. This perception-response closed loop enables the system to evolve from a passive tool to an active creative partner, providing key inspiration touchpoints when the user has not yet formed clear instructions.

[0023] 4. The rendering engine only needs to perform preset graphics operations, such as the light and shadow calculations of the Lambert model, without relying on cloud-based AI models or complex file transfers. The hierarchical structure of vector outlines and the modular design of stylized parameters further expand the system's teaching adaptability. Teachers can control the outline rendering depth through step parameters to achieve step decomposition or switch brushstroke textures through style parameters to simulate different painting materials. This architecture enables the system to provide a delay-free interactive experience in low-bandwidth environments while maintaining flexible support for diverse teaching scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is the overall processing flow chart of the AI-based art teaching material cloud space resource management system of the present invention;

[0025] Figure 2 This is a graph showing the entropy change trend during the inspiration bottleneck identification process based on the entropy monitoring of the interactive information flow according to the present invention;

[0026] Figure 3 This is a flow chart of the teaching demonstration interaction based on parameterized seed data sharing and interaction parameter synchronization of the present invention.

[0027] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0028] In order to make the objectives, technical solutions and advantages of the present invention clearer, the specific embodiments of the present invention will be described in detail below. It should be understood that the specific embodiments described are only used to explain the present invention, not to limit the present invention.

[0029] The present invention discloses an AI-based cloud space resource management system for art teaching materials. Its overall architecture is mainly composed of three core components: a material potential conversion module, an instant response engine, and a teaching interactive interface. The three work together to achieve a fundamental transformation of art teaching materials from static storage to dynamic generation and interactive response. Among them, the material potential conversion module is responsible for in-depth analysis and structured processing of original art materials, generating parameterized seed data carrying core information and associating a set of lightweight rendering rule sets for it, and then passing this data unit to the instant response engine. The instant response engine calls the rendering rules to execute this on the seed data in the client environment according to the interactive parameters input in real time from the teaching interactive interface. The map is processed by graphical operations, and the generated visual images are finally presented on the teaching interactive interface. In teaching practice, teachers often face the challenge that static materials cannot meet the dynamic exploration needs of the classroom. For example, when students immediately raise questions about the changes in light and shadow, traditional image resources cannot provide real-time demonstrations. To meet this challenge, the material potential conversion module in this solution is configured to perform a one-time deep conversion on the received original art materials. This processing is not a simple format conversion, but through a semi-automatic contour extraction algorithm, a scalable SVG vector path is accurately generated as the vector contour data that defines the basic form of the material. At the same time, the system performs color quantization analysis on the original material, extracts multiple key color values ​​of its main color blocks, and constitutes The color primitive data that represents its core hue, and generates a low-resolution grayscale guide map to imply the basic texture and light and shadow relationship; these three together constitute structured parametric seed data. This data paradigm that converts bitmap information into a combination of parametric vectors and color primitives allows a single material to be transformed from a solidified visual entity into a digital seed with unlimited derivative potential, laying the data foundation for subsequent real-time dynamic rendering; in addition, the process of efficiently converting original art materials into parametric seed data in the material potential conversion module relies on a set of integrated client computing processes. The process first calls an image segmentation model based on a convolutional neural network, which is used on a general object image dataset. After pre-training, it can automatically extract pixel-level object masks from the input original art materials and generate initial vector outline data. Then, a k-means clustering algorithm is applied to the pixel color space within the mask to extract N key color values ​​as color primitive data in a deterministic manner; correspondingly, the calling mechanism of the lightweight rendering rule set is also further clarified. Each stylized parameter, such as watercolor style, is uniquely mapped to a rendering profile containing specific values. This file defines a series of underlying shader variables such as edge dissolution factor, pigment diffusion rate, and underlying noise texture path, ensuring complete determinism and instantaneous responsiveness of the path from user selection to visual presentation.

[0030] Furthermore, in order to ensure that the system can still work stably under diverse real-world acquisition conditions, especially in non-ideal lighting environments, the system has built-in a robustness guarantee mechanism. Given that teaching materials often come from images with uneven lighting, such as backlit shooting, this can easily lead to feature extraction distortion. For this reason, when the material potential conversion module processes the original art material, it first calculates its gridded brightness distribution variance and compares the variance value with a numerical threshold preset by statistical analysis of a large number of benchmark sample images. If the variance exceeds the numerical threshold, it is determined that there is uneven lighting. Accordingly, the module will activate a compensatory extraction algorithm that no longer relies on global information. Instead, it uses advanced edge detection operators to accurately locate the light and dark boundary areas in the image and calculate the feature change gradient in this area. This gradient is regarded as the core clue reflecting the real physical lighting characteristics. The system then makes compensatory corrections to the feature extraction process in the global scope based on this local gradient to generate more stable parameterized seed data. This closed-loop correction strategy that focuses on local physical optical features to infer global effective information ensures that even for materials derived from non-ideal acquisition conditions, the generated vector contours and color primitive data are still accurate and reliable, thereby providing a solid data source for the dynamic rendering function of the entire system that is not significantly affected by the front-end acquisition quality.

[0031] The fluency of teaching interaction is the key to heuristic teaching. Any delay caused by background calculation or network transmission will break the teaching process. To this end, this solution has built an instant response engine that runs completely in the client environment. The core of the engine is a set of lightweight rendering rules based on WebGL or WebGPU standards. Its essence is a series of highly optimized shader programs or Canvas drawing instructions, which are configured to perform deterministic graphics operations without relying on complex artificial intelligence calculations. When teachers adjust the light source position parameters or light source color parameters through the draggable icons or sliders provided in the teaching interaction interface, At the same time, the instant response engine receives the changing interaction parameters and immediately calls the Lambert lighting model in the rule set. Based on the vector contour data in the parameterized seed data, it calculates and refreshes the internal light and shadow changes in real time; similarly, when the teacher switches the stylization parameters, such as setting it to any of the watercolor style, crayon style or line drawing style, the engine loads the corresponding preset noise texture and adjusts the edge smoothness algorithm of the rendering strokes to simulate the texture of the corresponding painting material. By placing all graphics operations locally, the system achieves millisecond-level response to user interactions, avoids real-time bottlenecks, and ensures a seamless and immersive teaching exploration process.

[0032] In addition, in order to upgrade the teaching tool from a passive response to an active guiding creative partner, the system also integrates a set of heuristic intervention mechanisms based on user behavior analysis, which aims to solve the inspiration bottleneck that users may encounter during the creative process; specifically, the teaching interaction interface is configured to monitor the interactive information flow generated by the user through the touch input device in real time, and calculate an entropy value based on the information flow , the calculation of the entropy value satisfies: ,in is the pressure value change rate, is the rate of change of the stroke velocity, and represents the irregularity of the brush stroke direction. These three items are calculated by performing difference and statistical dispersion calculations on the input data in a continuous time window, respectively. It is a set weight coefficient calibrated based on a large amount of user test data; when the calculated entropy value When the value of the user's image quality is continuously higher than a dynamically adjusted behavioral baseline, the system determines that the user is in an inspiration bottleneck state and automatically activates a set of heuristic rendering rules in the instant response engine. This rule will automatically superimpose a recommended color from a set color scheme or a texture of a set artistic style on the visual image currently being operated by the user in a non-invasive manner. This precise visual prompt triggered by implicit interactive perception can provide key inspiration touchpoints when the user needs it, effectively helping him to break through creative obstacles. Specifically, the entropy value of the interactive information flow is The calculation procedure of the stroke direction irregularity is further determined. Quantified as a series of brush stroke direction angles collected within a preset time window The amount of change Standard deviation , and the weight coefficient It does not rely on empirical settings, but is the deterministic output of a set of offline calibration procedures. This procedure inputs a set of interactive data samples with smooth and bottleneck time intervals manually labeled by senior art teachers, and applies a logistic regression model with the goal of minimizing the cross entropy loss between the predicted labels and the manually labeled labels. Finally, after the model is trained, the three outputs correspond to the pressure value change rate. Stroke velocity change rate and directional irregularities The characteristic regression coefficient of The final value of .

[0033] Example 1: In the digital still life sketching session of an advanced art class, the core of the teaching task is to guide students to understand and express a copper vessel placed in a complex side-backlit environment. When a student uses a touch device to create, the original art material he refers to is an image taken by a mobile device under non-ideal lighting conditions, which makes it difficult to accurately reproduce the subtle light and dark transitions and environmental reflections of the vessel on the screen. As a result, the creation comes to a standstill, and the brushstrokes appear hesitant. This constitutes a typical challenging scenario where teaching interaction and creative bottlenecks coexist. To cope with this situation, the teacher imports the unevenly illuminated image taken by the student into the system as the original art material. At this time, the system's built-in synergistic efficiency mechanism is triggered, and the material potential conversion module first activates its robustness assurance procedure to calculate the gridded brightness distribution of the image. The system immediately identifies the uneven lighting condition and then calls a compensatory algorithm based on feature change gradient extraction to infer accurate physical lighting information from the light and dark boundary area of ​​the original image, thereby generating a high-quality parameterized seed data that is not contaminated by poor acquisition conditions. This stable and reliable seed data then provides a key high-quality input prerequisite for the dynamic rendering function of the instant response engine, allowing teachers to adjust the light source position parameters without delay through the slider controls on the teaching interactive interface. Based on this high-quality data and the Lambert lighting model, the instant response engine smoothly and physically demonstrates the continuous change of light angle from side light to backlight on the student screen, as well as the corresponding evolution of light and shadow on the surface of the vessel, transforming a static, defective material into a dynamic, explorable virtual teaching model.

[0034] In this process, the system further demonstrated its design wisdom in resolving deep contradictions, namely, unifying the inherent technical dilemma of massive static storage and instant generation capabilities within a single architecture. Traditional solutions may require preparing massive rendering products to cope with different lighting conditions or rely on expensive cloud computing. However, this solution fundamentally redefines the problem by pre-processing the materials into lightweight parameterized seed data. That is, instead of searching for matching materials, the desired effect is generated based on core parameters. In other words, when the system continuously monitors the entropy value of the interactive information flow, , which is calculated strictly according to When the formula determines that the student is in an inspiration bottleneck due to persistent abnormalities in the rate of change of brushstroke speed and pressure, the set heuristic rendering rules in the instant response engine are automatically activated, and a layer of recommended colors based on a preset color scheme is superimposed on the dark part of the current visual image. This non-invasive visual disturbance breaks the student's creative deadlock and enables them to gain a new perspective. The complete process of this teaching example reveals the core design philosophy of the present invention's solution, namely, through a data-driven paradigm shift, the processing of art materials is transformed from the management of isolated, static entities to the control of a dynamic system with inherent potential and derivative logic. The system does not simply add functions, but deeply couples robust data processing at the front end, lightweight real-time rendering at the mid-end, and implicit interactive perception at the back end to construct a complete closed loop from data purification to context generation to inspiration. This gives digital teaching materials unprecedented flexibility and inspiration while ensuring the core teaching requirements of low computing power and high responsiveness.

[0035] Example 2: This example aims to conduct quantitative experiments. The construction of the test platform is based on a set of standardized hardware and software environments. The hardware part uses a computer equipped with a high-performance graphics processing unit, connected to a professional-grade touch input device that can accurately collect pressure values, pen stroke speed and coordinate positions; the software part is a test version of the system of the present invention. In addition to all the aforementioned functions, this version also integrates a high-precision data log module for recording time-synchronized interaction parameter sequences and system status determination results. In order to ensure the authenticity of the test environment, the core parameters in the test, especially the dynamic baseline threshold of the entropy value used to trigger the inspiration bottleneck state determination, are set according to a rigorous decision-making logic chain. The setting of this threshold is fundamentally based on the following principles: The technical consideration for this approach is to achieve an optimal balance between the sensitivity of intervention and the intrusiveness to the user. A threshold that is too low will cause the system to be frequently falsely triggered when the user is thinking normally, while a threshold that is too high will cause the system to miss the opportunity to intervene. Based on this, the system is configured to collect the user's basic interaction data through a guided free creation task at the beginning of each session, and calculate the user's average entropy value and standard deviation in this specific environment based on this data. The dynamic baseline threshold is set to the average entropy value plus twice the standard deviation. This calibration procedure enables the determination of the threshold to adapt to the user's individual habits and the basic complexity of the current task, thereby providing a non-arbitrary and statistically significant reference benchmark for subsequent judgments.

[0036] Ten participants with diverse digital art backgrounds were recruited for the experiment. Each participant was asked to complete two tasks: Task A involved copying a clearly structured geometric figure composition, aiming to induce a fluent and confident creative state; Task B involved free-form creation based on an abstract vocabulary, aiming to simulate and induce a state of inspirational bottleneck, where stagnation, revision, and hesitation may occur. Throughout the entire process, the data log module continuously recorded pressure values, pen speed, and direction data streams at a 50-millisecond interval, and calculated entropy values ​​in real time. At the same time, an observer, unaware of the entropy data, independently marked the time intervals when the test subjects showed obvious creative hesitation based on the video records and behavioral performance. The experiment observed a clear data trend. When performing Task A, the entropy values ​​of all test subjects were All remained below their respective dynamic baseline thresholds, showing stable low-amplitude fluctuations. When performing Task B, the data log showed that the entropy values ​​of all testers were Multiple spike pulses with duration exceeding the preset minimum trigger duration appeared in all cases, and the occurrence time of these pulses was highly consistent with the hesitation intervals independently marked by the observers.

[0037] Table 1: Comparison of key performance indicators extracted from the data logs of three representative testers.

[0038]

[0039] As shown in Table 1, the sample data clearly confirms the above trend. There is a high degree of consistency between the judgment results of and those of human observers; the internal mechanism of this phenomenon is that the formula The model we constructed essentially quantifies the physical irregularities of creative behavior. Flowing brushstrokes typically have stable and predictable pressure and velocity curves, as well as smooth directional changes, resulting in low entropy. However, hesitation, repeated revisions, and aimless exploration in bottleneck states are directly reflected in frequent fluctuations in pressure. Increase, speed of emergency stop and start, The chaos of increase and direction, Increase, eventually leading to entropy of the increase.

[0040] Example 3: This example combines Figures 1 to 3 , describes the AI-based cloud space resource management system for art teaching materials, such as Figure 1As shown, first, the original art material is input into the material potential conversion module, which is used to analyze the input original art material and determine whether the lighting is uniform. If the judgment result is no, the system starts the gradient compensation extraction mechanism to extract the feature change gradient of the material to cope with non-ideal acquisition conditions; if the lighting is uniform or the compensation is completed, a structured parameterized seed data is generated. The parameterized seed data is combined with the lightweight rendering rule set to form the basic input for subsequent graphics operations. The parameterized seed data is then passed to the instant response engine client for operation. The engine uses the client to execute the interactive parameters input based on the teaching interactive interface. The local rendering logic performs graphics rendering operations, and the teaching interaction interface is used to interact with users. The interface integrates an interactive information flow entropy monitoring module to continuously monitor the behavioral signals generated by users through touch input devices. When the interactive information flow entropy monitoring module determines whether the user is in an inspiration bottleneck state, if the result is yes, the system will automatically call the heuristic rendering rule visual perturbation to assist the user's creation, and inject the rule into the instant response engine for processing; if the user is not in an inspiration bottleneck state, the normal rendering process will continue to be maintained. Finally, the instant response engine will feed back the processing results to the teaching interaction interface in real time, presenting it as a real-time generated visual image.

[0041] like Figure 2 As shown, the horizontal axis is time in minutes and the vertical axis is entropy value , where two different types of tasks are compared: Task A is represented by a solid line, and its corresponding entropy value is The overall stability is maintained, with a small fluctuation range; Task B creation is represented by a dotted line, and the entropy value There are spike fluctuations in multiple time periods. The horizontal dotted line in the figure indicates the dynamic baseline, which is used as a reference value to determine whether it is in a bottleneck state. When the entropy value of Task B is When the dynamic baseline is exceeded, the system determines that the user has entered the inspiration bottleneck state; the figure uses the gray background area to mark three typical bottleneck periods, marked as bottleneck period 1, bottleneck period 2 and bottleneck period 3, which correspond to the entropy value in task B. The time period is significantly higher than the dynamic baseline.

[0042] like Figure 3As shown, the teacher uploads parameterized seed data to the instant response engine through the teaching interaction interface, and then the instant response engine returns a shared link. The teacher distributes the shared link to student A and student B. After student A and student B each select a shared material, they load the shared material. The teaching interaction interface obtains the seed data and transmits the data to student A and student B respectively. Next, the teacher demonstrates the light source adjustment on the teaching interaction interface. The teaching interaction interface uploads the changed parameters to the cloud synchronization service to realize the broadcast of interactive parameters. Student A and student B respectively receive the synchronization parameter changes and trigger local rendering updates, and finally display the synchronization effect on the student end. At the same time, the teaching interaction interface also synchronously executes parameter changes and triggers local rendering updates and displays synchronization effects to achieve consistent presentation of interactive images in multi-terminal teaching.

[0043] Example 4: In a specific application scenario, the system receives an original art material, which is a photo of a plaster sphere under the side illumination of a single-point strong light source. The material is characterized by a wide and clear light-dark boundary area, as well as large-area shadows and local highlights caused by the sharp attenuation of light intensity. This type of material poses a technical challenge to conventional automated contour extraction and color analysis algorithms. Direct processing will result in the interruption of contour lines in dark areas and distortion of color primitive extraction.

[0044] To address this situation, the material potential conversion module is configured to execute a four-stage deterministic guarantee procedure; first, in the gridded brightness distribution variance calculation stage, the system divides the pixel matrix of the input original art material into a 16x16 macroblock grid and calculates the average brightness value of all pixels in each macroblock one by one. Subsequently, the system calculates their overall variance based on these 256 average brightness values. This variance value is used as a quantitative indicator to measure the uniformity of global illumination; second, in the light-dark boundary area identification stage, the system compares the variance value with a preset numerical threshold. If the variance value is greater than the numerical threshold, the system determines that the illumination is uneven and starts the identification program. The program first performs a Sobel operator convolution on the grayscale image of the original material to generate an edge intensity map containing all brightness change gradients. Then, the program only retains those pixels whose edge intensity is higher than a specific local contrast threshold. Finally, through a region connectivity algorithm, these high-intensity pixels are connected into one or more continuous regions, where The largest continuous area is identified as the core light-dark boundary area; thirdly, in the compensatory extraction stage, its core is to reconstruct global and balanced lighting information through the identified local physical optical features; at this stage, the system first calculates the average brightness change gradient vector of all pixels in the light-dark boundary area. This vector contains information about the lighting direction and intensity change rate. Based on this vector, the system constructs a global illumination compensation map with the same size as the original material. This map is essentially a two-dimensional function. Its value in the light-dark boundary area is strongly correlated with the original gradient vector, and it extends smoothly and nonlinearly to the highlight and shadow areas to simulate a virtual, uniform diffuse light environment; finally, the system applies this global illumination compensation map to the brightness channel of the original art material to generate a lighting-balanced intermediate image. Subsequent vector contour data extraction and color primitive data extraction operations are all performed on this compensatory corrected intermediate image to ensure the accuracy and stability of the final generated parameterized seed data.

[0045] It should be noted that the aforementioned numerical threshold is set based on an offline, reproducible experimental calibration process, which uses a benchmark dataset containing hundreds of images, half of which are uniformly illuminated samples taken under ideal diffuse light conditions, and the other half are non-uniformly illuminated samples taken under various single-point strong light conditions; the system performs the aforementioned gridded brightness distribution variance calculation on all samples to obtain two sets of variance value distributions. The final value of the numerical threshold is determined to be the variance value that can minimize the sum of the classification error rates of the two groups of samples. This procedure ensures that the threshold setting has a clear statistical basis and classification efficiency, thereby making the startup judgment link of the entire guarantee mechanism objective.

[0046] Example 5: To further ensure the adaptability of the technical solution of the present invention in different deployment environments and the consistency of the output results, the system is configured to execute a set of standardized offline calibration and data filling procedures before being put into use. The purpose of this procedure is to pre-build a set of verified structured rules and knowledge bases for the logical modules involved in stylized rendering and heuristic intervention in the system to avoid any unfounded and random decisions in real-time operation; specifically, in the process of constructing a lightweight rendering rule set, the technicians first define a set of representative basic material templates, covering high-gloss metal, matte fabric, semi-transparent The typical textures of watercolor and rough chalk, and a specific rendering rule set containing corresponding shader logic and texture resources is manually written for each template; when the material potential conversion module processes a new original art material, it will not only generate parameterized seed data, but also classify it into the most similar basic material template mentioned above by analyzing the texture frequency distribution and color saturation statistical characteristics of the material, and automatically associate the rendering rule set corresponding to the template with the newly generated seed data. This procedure ensures that any input material can be deterministically matched to a pre-optimized rendering pipeline that best suits its visual characteristics.

[0047] Similarly, in the process of filling in the content for heuristic rendering rules, the system adopts a construction method based on the knowledge graph of art theory; this method analyzes the color composition of a large number of classic works of art and combines it with color harmony theory to establish a database that associates basic hue combinations with harmonious color schemes, where harmonious color schemes are defined as preset models including complementary colors, approximate colors and triadic colors; when the system determines that the user is in an inspiration bottleneck state in the teaching interaction interface and activates the heuristic rendering rules, it will extract the color primitive data in the user's current work as query input, retrieve one or more recommended colors that form a harmonious relationship with it from the database, and superimpose them on the picture as visual disturbances. This recommendation mechanism based on a pre-built knowledge base ensures that all heuristic interventions are not random behaviors, but guidance with clear art theory basis to provide effective creative clues.

[0048] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0049] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An AI-based cloud space resource management system for art teaching materials, characterized by: include: A material potential conversion module is configured to receive raw art material, convert the raw art material into structured parametric seed data in one go, and associate a set of lightweight rendering rules with the parametric seed data; the parametric seed data includes vector outline data defining the basic shape of the raw art material, color primitive data defining the basic hue of the raw art material, and a low-resolution grayscale guide map for suggesting basic texture and light and shadow relationships; The lightweight rendering rule set includes shader programs or drawing instructions for generating visual images based on parameterized seed data and interaction parameters; An instant response engine, connected to the material potential conversion module, is configured to receive interactive parameters in a client environment and invoke a lightweight rendering rule set to perform graphics operations on the parameterized seed data, thereby generating and refreshing visual images in real time. Interactive parameters include light source position and color parameters for simulating lighting changes, as well as stylized parameters for simulating the texture of painting materials. a teaching interaction interface, connected to the instant response engine, configured with input controls for providing interaction parameters, and sending the interaction parameters to the instant response engine; Vector contour data is layered, and the teaching interactive interface also provides step parameter input controls for controlling the number of layers of vector contour data rendering; The teaching interaction interface is further configured to: monitor in real time the entropy value of the interactive information flow generated by the user through the touch input device, where the entropy value of the interactive information flow is calculated by combining the change rate or irregularity of the following parameters: the change rate of the pressure value of the touch input device used by the user, the change rate of the pen stroke speed, and the irregularity of the pen stroke direction; determine whether the user is in an inspiration bottleneck state based on the entropy value; and when it is determined that the user is in an inspiration bottleneck state, activate the set heuristic rendering rules in the instant response engine to superimpose a visual perturbation on the currently generated visual image; The specific configuration of the instant response engine is: when the light source position parameters and light source color parameters change, the Lambert lighting model is used to calculate and render the light and shadow changes within the vector contour data in real time.

2. The AI-based art teaching material cloud space resource management system according to claim 1 is characterized in that: The material potential conversion module is also configured to: calculate the gridded brightness distribution variance of the original art material to identify uneven lighting; when the brightness distribution variance is greater than a set numerical threshold, the material potential conversion module is further configured to extract the characteristic change gradient of the light and dark boundary area in the original art material, and perform compensatory extraction on the global scale based on the characteristic change gradient to generate parameterized seed data.

3. The AI-based art teaching material cloud space resource management system according to claim 1 is characterized in that: The material potential conversion module is specifically configured to: perform semi-automatic contour extraction on the original art material to generate SVG vector paths as vector contour data; and extract multiple color values ​​of the main color blocks of the original art material as color primitive data.

4. The AI-based art teaching material cloud space resource management system according to claim 1 is characterized in that: A lightweight rendering rule set is a set of shader programs or Canvas drawing instructions implemented based on the WebGL or WebGPU standards. It is configured to perform graphics operations based on set rules without relying on complex artificial intelligence calculations.

5. The AI-based art teaching material cloud space resource management system according to claim 1 is characterized in that: The instant response engine is further configured to: when the stylization parameter is set to any of the watercolor style, crayon style or line drawing style, load the corresponding noise texture and adjust the edge smoothness of the rendered stroke to simulate the texture of the corresponding painting material.

6. The AI-based art teaching material cloud space resource management system according to claim 1 is characterized in that: The teaching interaction interface is specifically configured to provide draggable icons or sliders as input controls for interactive parameters when the teacher selects materials. The dragging operation of the icons or sliders is associated with the real-time changes of the visual images.

7. An AI-based cloud space resource management system for art teaching materials according to claim 1, characterized in that: Interaction information flow entropy The calculation satisfies: ,in, Indicates the rate of change of pressure value, Indicates the rate of change of the stroke velocity. Indicates the direction of the brushstroke collected within the preset time window The standard deviation of the variation of , used to quantify the irregularity of the stroke direction, and To set the weight coefficient.

8. The AI-based art teaching material cloud space resource management system according to claim 1 is characterized in that: The heuristic rendering rule is configured as follows: when the user is in a state of inspiration bottleneck, a recommended color in a set color scheme or a texture in a set artistic style is automatically superimposed on the visual image currently operated by the user.

Images