Multi-angle artificial board cabinet product effect picture rendering method based on AI

Abstract

The invention discloses an AI-based multi-angle artificial board cabinet product effect picture rendering method, and relates to the technical field of model rendering, and the method comprises the steps: extracting the pixel data of a shadow part in a three-dimensional cabinet model, obtaining a shadow uniformity coefficient, extracting a halo region in the three-dimensional cabinet model, and obtaining a halo unreasonable coefficient; extracting three-dimensional coordinate data of the three-dimensional cabinet model to obtain a reflection position difference coefficient; and obtaining a rendering effect index according to the shadow uniformity coefficient, the halo unreasonable coefficient and the reflective position difference coefficient, comparing the rendering effect index with a preset rendering effect index threshold, and determining whether re-rendering is needed according to a comparison result. The AI-based multi-angle artificial board cabinet product effect picture rendering can accord with the actual effect, and the rendering accuracy is high. The unnatural or distortion phenomenon is avoided, and the authenticity and reliability of the final effect picture are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of model rendering, and in particular to an AI-based multi-angle artificial board cabinet product rendering method. Background Art

[0002] The AI-based multi-angle artificial board cabinet product rendering method refers to a method that combines deep learning, computer vision and graphics rendering technology to generate highly realistic three-dimensional cabinet model renderings from different perspectives. This method usually uses a large amount of training data to identify the material, color, texture of the cabinet and its interaction with the environment, thereby simulating the visual effects in the real world. In this way, designers and consumers can intuitively understand the appearance and details of the product during the decision-making process, improving the purchasing experience and satisfaction.

[0003] However, AI rendering methods may face some challenges under complex lighting conditions, especially when simulating natural light, AI may produce inaccurate shadows and light reflection effects. For example, when a cabinet is placed in a kitchen environment with multiple light sources, the reflection, refraction and layering of shadows are difficult to fully simulate, resulting in distortion of the rendering effect. In this case, the interaction between light and objects becomes complicated, and existing methods sometimes produce unnatural or distorted phenomena when dealing with the interaction of light and shadow in this case, thereby reducing the authenticity and reliability of the final rendering. Summary of the invention

[0004] The purpose of the present invention is to solve the problem of making the AI-based multi-angle artificial board cabinet product rendering consistent with the actual effect, with high rendering accuracy, without unnatural or distorted phenomena, and enhancing the authenticity and reliability of the final rendering. A multi-angle artificial board cabinet product rendering method based on AI is proposed.

[0005] The present invention proposes an AI-based multi-angle artificial board cabinet product rendering method, the method comprising:

[0006] For the 3D cabinet model that has been rendered, pixel data of the shadow part in the 3D cabinet model is extracted, and a shadow uniformity coefficient is obtained according to the pixel data of the shadow part;

[0007] Extracting the halo area in the three-dimensional cabinet model, and obtaining the halo unreasonable coefficient according to the halo area in the three-dimensional cabinet model;

[0008] Extracting three-dimensional coordinate data of the three-dimensional cabinet model, and obtaining a reflection position difference coefficient according to the three-dimensional coordinate data;

[0009] The rendering effect index is obtained according to the shadow uniformity coefficient, the halo unreasonable coefficient and the reflection position difference coefficient, and the rendering effect index is compared with the preset rendering effect index threshold, and whether re-rendering is needed is determined according to the comparison result.

[0010] Optionally, obtaining a shadow uniformity coefficient according to pixel data of the shadow portion includes:

[0011] Extract the pixel data of the shadow area from the rendered 3D cabinet model, denoted as set P, where the grayscale value of each pixel is I i ;

[0012] Calculate the mean grayscale value of all pixels in the shadow area The calculation formula is: N represents the total number of pixels in the shadow area, I i Represents the gray value of the i-th pixel;

[0013] Calculate the variance σ of the shadow grayscale 2 , the calculation formula is:

[0014] Calculate the shadow gradient G through the Sobel operator i , and calculate the variance σ of the gradient change 2 g , the calculation formula is: In the formula, G i is the gradient value of the i-th pixel, is the mean of all gradient values ​​in the shadow area;

[0015] Calculate the shadow uniformity coefficient, the calculation formula is:

[0016]

[0017] Where WER is the shadow uniformity coefficient, a1 and a2 are preset proportional coefficients, and both a1 and a2 are greater than 0, ∈ is a small constant used to avoid the denominator being zero, and its value is between 0 and 1.

[0018] Optionally, the halo unreasonable coefficient obtained according to the halo area in the three-dimensional cabinet model includes:

[0019] Extract the halo area from the rendered 3D cabinet model and record it as pixel set T, where the brightness value of each pixel is S j , and obtain the color information B of the halo area j And light source distance information Y j ;

[0020] Calculate the halo brightness attenuation factor: according to the distance Y of the light source j, calculate the brightness attenuation of the halo area, the calculation formula is: Where P j is the brightness attenuation factor of the jth pixel, S j is the brightness value of the jth pixel, Y j is the distance from the jth pixel to the light source; ε is a small constant used to avoid the denominator being zero, and its value is between 0 and 1;

[0021] Calculate the halo color deviation factor: extract the color information B of the halo area j , calculate the color uniformity deviation, the calculation formula is: Where DR is the color uniformity deviation, M is the total number of pixels in the halo area, and B j is the color value of the jth pixel, B s Is the color value of the light source.

[0022] Optionally, obtaining the halo unreasonable coefficient according to the halo area in the three-dimensional cabinet model also includes:

[0023] Calculate the diffusion coefficient E of the halo boundary using the following formula: Where R is the total number of pixels in the halo edge area, is the brightness gradient change of the j-th pixel in the x and y directions;

[0024] Calculate the halo unreasonable coefficient: The halo unreasonable coefficient is calculated by comprehensively calculating the brightness attenuation factor, color uniformity deviation and diffusion coefficient of the halo boundary. The calculation formula is:

[0025]

[0026] Where PKU is the halo unreasonable coefficient, b1, b2, and b3 are the preset proportional coefficients, and b1, b2, and b3 are all greater than 0.

[0027] Optionally, obtaining the reflection position difference coefficient according to the three-dimensional coordinate data includes:

[0028] Define the position of the light source in the rendered 3D cabinet model in 3D space, denoted as S(x s ,y s , z s ), where x s ,y s , z s Respectively represent the coordinates of the light source on the x, y, and z axes;

[0029] Extract the pixel set H of the reflective points from the rendered model, and the position of each reflective point is H q (x q ,y q , z q), where x q ,y q , z q is the coordinate of the qth reflection point in three-dimensional space;

[0030] Calculate the distance from the reflection point to the light source: For each reflection point H q , calculate its distance D from the light source S q , the calculation formula is:

[0031] The normal vector N based on the light source position and the object surface q Calculate the expected reflection position L q , the reflection position is usually in the middle between the light source and the reflection point: the calculation formula is: L q =H q +N q ;

[0032] Calculate the difference between the actual position of each reflection point and the expected reflection position In the formula, is the expected position coordinate of the qth reflection point;

[0033] Calculate the reflection position difference coefficient DTY, the calculation formula is: Where T is the total number of reflective points.

[0034] Optionally, the rendering effect index obtained according to the shadow uniformity coefficient, the halo unreasonable coefficient and the reflection position difference coefficient includes:

[0035]

[0036] Wherein, Dax is the rendering effect index, WER, PKU, and DTY are the shadow uniformity coefficient, halo unreasonable coefficient, and reflection position difference coefficient, respectively; f1, f2, and f3 are the preset proportional coefficients of the shadow uniformity coefficient, halo unreasonable coefficient, and reflection position difference coefficient, respectively; and f1, f2, and f3 are all greater than 0.

[0037] Optionally, comparing the rendering effect index with a preset rendering effect index threshold, and determining whether re-rendering is required according to the comparison result includes:

[0038] Compare the rendering effect index with the preset rendering effect index threshold. If the rendering effect index is not less than the preset rendering effect index threshold, no re-rendering is required.

[0039] If the rendering effect index is less than the preset rendering effect index threshold, re-rendering is required until the rendering effect index of the rendered model is not less than the preset rendering effect index threshold.

[0040] Beneficial effects of the present invention:

[0041] The present invention proposes a rendering method for a multi-angle artificial board cabinet product effect map based on AI, which obtains a shadow uniformity coefficient by extracting pixel data of the shadow part in the three-dimensional cabinet model, extracts the halo area in the three-dimensional cabinet model, obtains the halo unreasonable coefficient, extracts the three-dimensional coordinate data of the three-dimensional cabinet model, and obtains the reflection position difference coefficient; obtains a rendering effect index according to the shadow uniformity coefficient, the halo unreasonable coefficient and the reflection position difference coefficient, and compares the rendering effect index with a preset rendering effect index threshold, and determines whether re-rendering is required according to the comparison result, so that the rendering of the multi-angle artificial board cabinet product effect map based on AI can be consistent with the actual effect, and the rendering accuracy is high. There will be no unnatural or distorted phenomenon, and the authenticity and reliability of the final effect map will be enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The present invention will be further described below in conjunction with the accompanying drawings.

[0043] Figure 1 This is a flow chart of the AI-based multi-angle wood-based panel cabinet product rendering method. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0046] The embodiment of the present invention provides a method for rendering a multi-angle artificial board cabinet product effect picture based on AI. Figure 1 , Figure 1 A flowchart of a method for rendering a multi-angle artificial board cabinet product effect map based on AI provided in an embodiment of the present invention. The method comprises the following steps:

[0047] For the 3D cabinet model that has been rendered, pixel data of the shadow part in the 3D cabinet model is extracted, and a shadow uniformity coefficient is obtained according to the pixel data of the shadow part;

[0048] Extracting the halo area in the three-dimensional cabinet model, and obtaining the halo unreasonable coefficient according to the halo area in the three-dimensional cabinet model;

[0049] Extracting three-dimensional coordinate data of the three-dimensional cabinet model, and obtaining a reflection position difference coefficient according to the three-dimensional coordinate data;

[0050] The rendering effect index is obtained according to the shadow uniformity coefficient, the halo unreasonable coefficient and the reflection position difference coefficient, and the rendering effect index is compared with the preset rendering effect index threshold, and whether re-rendering is needed is determined according to the comparison result.

[0051] Based on the AI-based multi-angle artificial board cabinet product rendering method provided by the embodiment of the present invention, through the above method, the AI-based multi-angle artificial board cabinet product rendering is consistent with the actual effect, and the rendering accuracy is high. There will be no unnatural or distorted phenomena, and the authenticity and reliability of the final rendering are enhanced.

[0052] In one embodiment, obtaining the shadow uniformity coefficient according to the pixel data of the shadow part includes:

[0053] Extract the pixel data of the shadow area from the rendered 3D cabinet model, denoted as set P, where the grayscale value of each pixel is I i ;

[0054] Calculate the mean grayscale value of all pixels in the shadow area The calculation formula is: N represents the total number of pixels in the shadow area, I i Represents the gray value of the i-th pixel;

[0055] Calculate the variance σ of the shadow grayscale 2 , the calculation formula is:

[0056] Calculate the shadow gradient G using the Sobel operator or other edge detection methods i , and calculate the variance σ of the gradient change 2 g , the calculation formula is: In the formula, G i is the gradient value of the i-th pixel, is the mean of all gradient values ​​in the shadow area;

[0057] Calculate the shadow uniformity coefficient, the calculation formula is:

[0058]

[0059] Where WER is the shadow uniformity coefficient, a1 and a2 are preset proportional coefficients, and both a1 and a2 are greater than 0, ∈ is a small constant used to avoid the denominator being zero, and its value is between 0 and 1.

[0060] It should be noted that a1 and a2 are set by professionals according to actual conditions. Generally, the sum of a1 and a2 is 1. For example, a1 and a2 can be 0.5 and 0.5 respectively, or other numbers, without specific limitation.

[0061] It should be noted that first, the pixel data of the shadow area needs to be extracted from the rendered 3D cabinet model. This process usually relies on image processing software or image analysis tools in the graphics rendering engine. Specifically, the required data can be obtained through the following steps:

[0062] Image rendering and sampling: Render the 3D cabinet model into a high-resolution 2D image. At this point, make sure the rendering settings can capture accurate lighting and shadow effects to avoid data distortion caused by low image quality.

[0063] Extraction of shadow areas: Use threshold segmentation, region growing or image segmentation algorithms to identify and extract shadow areas. Determine which pixels belong to shadows by analyzing the grayscale value or color information in the image. This can be automated by writing algorithms or using image processing tools such as OpenCV and other libraries.

[0064] Grayscale value analysis: Once the shadow area is successfully extracted, the grayscale value data of each pixel can be obtained. The grayscale value of each pixel represents its brightness level, which is crucial for the subsequent calculation of mean and variance.

[0065] Edge detection: To calculate the shadow gradient, the Sobe I operator or other edge detection methods need to be applied to the pixels in the shadow area. This process generates a gradient image of the shadow area, which captures the intensity and direction of the shadow edge by analyzing the changes between pixels, thereby providing basic data for the calculation of the gradient value.

[0066] Data processing and statistical calculation: Finally, the collected grayscale and gradient value data are processed by writing scripts or using statistical analysis tools to calculate the shadow uniformity coefficient. This calculation involves solving statistical quantities such as mean and variance to evaluate the uniformity of the shadow.

[0067] Through the above steps, the shadow data in the rendered image can be comprehensively extracted and analyzed, providing a reliable basis for the calculation of the shadow uniformity coefficient, and then helping to evaluate the authenticity and quality of the rendering effect.

[0068] It should be noted that the shadow uniformity coefficient refers to the uniformity of the shadow in the rendering model. If the shadow uniformity is too large, it usually means that the shadow distribution is too consistent and lacks change, which is contrary to the shadow effect in the actual physical environment. In the real world, the formation of shadows is affected by many factors such as the position of the light source, the shape of the object, the surface texture and environmental factors, and usually presents the characteristics of gradual changes, blurring and local changes. If the shadows in the rendering are too uniform, consumers may mistakenly believe that there are problems with the material or light and shadow processing of the cabinet, thereby reducing the authenticity and credibility of the rendering. Therefore, reasonable shadow changes are a key factor in improving the quality and realism of renderings.

[0069] In one embodiment, obtaining the halo unreasonable coefficient according to the halo area in the three-dimensional cabinet model includes:

[0070] Extract the halo area from the rendered 3D cabinet model and record it as pixel set T, where the brightness value of each pixel is S j , and obtain the color information B of the halo area j And light source distance information Y j ;

[0071] Calculate the halo brightness attenuation factor: according to the distance Y of the light source j , calculate the brightness attenuation of the halo area. Ideally, the brightness should decrease with the distance Y j The ratio of the brightness of each pixel to its distance from the light source is calculated using the following formula: Where P j is the brightness attenuation factor of the jth pixel, S j is the brightness value of the jth pixel, Y j is the distance from the jth pixel to the light source; ε is a small constant used to avoid the denominator being zero, and its value is between 0 and 1;

[0072] Calculate the halo color deviation factor: extract the color information B of the halo area j , calculate the color uniformity deviation. Ideally, the halo color should be close to the light source color and there will be no obvious color jump. The color uniformity deviation is defined as: Where DR is the color uniformity deviation, M is the total number of pixels in the halo area, and B j is the color value of the jth pixel, B s is the color value of the light source;

[0073] By calculating the diffusion coefficient E of the halo boundary, the uniform diffusion degree of the halo is judged; the diffusion of the real halo should be smooth, rather than having obvious boundaries or abrupt jumps, and the halo boundary diffusivity is calculated using gradient changes: the calculation formula is: Where R is the total number of pixels in the halo edge area, is the brightness gradient change of the j-th pixel in the x and y directions;

[0074] Calculate the halo unreasonable coefficient: The halo unreasonable coefficient is calculated by comprehensively calculating the brightness attenuation factor, color uniformity deviation and diffusion coefficient of the halo boundary. The calculation formula is:

[0075]

[0076] Where PKU is the halo unreasonable coefficient, b1, b2, and b3 are the preset proportional coefficients, and b1, b2, and b3 are all greater than 0.

[0077] It should be noted that b1, b2, and b3 are set by professionals according to actual conditions. Generally, the sum of b1, b2, and b3 is 1. For example, b1, b2, and b3 can be 0.3, 0.3, and 0.4, respectively, or other numbers, without specific limitation.

[0078] It should be noted that the data involved in the above calculations can be extracted from the rendered three-dimensional cabinet model through a variety of image processing and analysis techniques. First, edge detection algorithms (such as Sobel operator, Canny edge detection) can be used to identify the boundaries of the halo area and separate it from the overall image of the cabinet model. Through color segmentation technology, the halo area and other parts can be accurately distinguished, so as to extract the brightness value and color information of each halo area pixel. In addition, in order to obtain the distance from each pixel to the light source, the geometric relationship between the three-dimensional coordinate data of the model and the position of the light source can be used to determine it by calculating the position vector of each pixel. These data combined with gradient analysis (such as calculating the gradient of brightness change) can comprehensively evaluate the brightness attenuation, color uniformity and boundary diffusion of the halo, thereby providing accurate data support for the subsequent calculation of the halo unreasonable coefficient.

[0079] The halo unreasonable coefficient refers to the unreasonable degree of halo in the rendered model. If the unreasonable degree of halo is greater, it means that the halo of the rendered model does not conform to the actual cabinet rendering, indicating that the rendering model is more likely to be distorted, because the greater the unreasonable degree of halo, it means that there are abnormalities in the brightness attenuation, color uniformity and boundary diffusion of the halo in the rendered model, which usually causes the halo to look unnatural. For example, if the brightness of the halo does not gradually decrease with distance, or the color does not match the color of the light source, the distribution of light will appear unrealistic; similarly, if the boundary of the halo is too sharp or there is an abrupt jump instead of smooth diffusion, this is also inconsistent with the soft halo effect in the actual scene. These phenomena will make people feel that the lighting effect of the rendered model lacks realism, thereby reducing the overall quality and authenticity of the cabinet rendering, indicating that the rendering model is more likely to be distorted.

[0080] In one embodiment, obtaining the reflection position difference coefficient according to the three-dimensional coordinate data includes:

[0081] Define the position of the light source in the rendered 3D cabinet model in 3D space, denoted as S(x s ,y s , z s ), where x s ,y s , z s Respectively represent the coordinates of the light source on the x, y, and z axes;

[0082] Extract the pixel set H of the reflective points from the rendered model, and the position of each reflective point is H q (x q ,y q , z q ), where x q ,y q , z q is the coordinate of the qth reflection point in three-dimensional space;

[0083] Calculate the distance from the reflection point to the light source: For each reflection point H q , calculate its distance D from the light source S q , the calculation formula is:

[0084] The normal vector N based on the light source position and the object surface q Calculate the expected reflection position L q , the reflection position is usually in the middle between the light source and the reflection point: the calculation formula is: L q =H q +N q ;

[0085] Calculate the difference between the actual position of each reflection point and the expected reflection position In the formula, is the expected position coordinate of the qth reflection point;

[0086] Calculate the reflective position difference coefficient DTY: Calculate the reflective position difference coefficient by taking statistics on the differences of all reflective points. Where T is the total number of reflective points.

[0087] It should be noted that the data involved in the above calculations can be obtained through image analysis and three-dimensional space coordinate extraction technology. First, through ray tracing or reflection modeling algorithms, all reflective points can be identified in the rendered three-dimensional cabinet model and their three-dimensional coordinates can be extracted. The position of the light source is also determined by the lighting configuration of the rendering scene and can be read directly from the light source settings of the model. In addition, the normal vector of the object surface can be obtained by calculating the normal information of each point on the model surface, usually using a surface normal extraction algorithm or directly extracted from the geometric structure of the model. These data allow the accurate calculation of the distance from each reflective point to the light source and the position where the reflective point should appear in an ideal situation, thereby further determining the difference between the actual reflective position and the expected reflective position. By counting the position differences of all reflective points, the reflective position difference coefficient can be obtained, providing a quantitative analysis basis for the light reflection accuracy of the rendered model.

[0088] It should be noted that the larger the reflection position difference coefficient is, the greater the deviation between the reflection position in the rendered model and the expected reflection position, which means that the reflection effect of light on the surface of the object is inaccurate. In real scenes, reflections usually appear in specific, physically correct positions, such as reasonable interaction positions between light sources, surface normal vectors, and observers. If the position of the reflection in the rendered model is inconsistent with these physical rules, the reflection will look unnatural, with misalignment or excessive reflection. This distortion will make the cabinet renderings appear unrealistic, and cause deviations in the texture and gloss of the material, thereby reducing the overall visual quality and authenticity of the renderings and increasing the possibility of inaccurate model rendering.

[0089] In one embodiment, the rendering effect index is obtained according to the shadow uniformity coefficient, the halo unreasonable coefficient and the reflection position difference coefficient, including:

[0090]

[0091] Wherein, Dax is the rendering effect index, WER, PKU, and DTY are the shadow uniformity coefficient, halo unreasonable coefficient, and reflection position difference coefficient, respectively; f1, f2, and f3 are the preset proportional coefficients of the shadow uniformity coefficient, halo unreasonable coefficient, and reflection position difference coefficient, respectively; and f1, f2, and f3 are all greater than 0.

[0092] It should be noted that f1, f2, and f3 are set by professionals according to actual conditions. Generally, the sum of f1, f2, and f3 is 1. For example, f1, f2, and f3 can be 0.4, 0.4, and 0.3, respectively, or other numbers, without specific limitation.

[0093] In one embodiment, comparing the rendering effect index with a preset rendering effect index threshold, and determining whether re-rendering is required according to the comparison result includes:

[0094] The rendering effect index is compared with the preset rendering effect index threshold. If the rendering effect index is not less than the preset rendering effect index threshold, it means that the rendering result of the AI-based multi-angle artificial board cabinet product rendering is accurate and will not be distorted, and no re-rendering is required;

[0095] If the rendering effect index is less than the preset rendering effect index threshold, it means that the rendering result of the AI-based multi-angle wood-based panel cabinet product rendering is inaccurate and may be distorted. It needs to be re-rendered and the rendering parameters need to be appropriately modified until the rendering effect index of the rendered model is not less than the preset rendering effect index threshold.

[0096] It should be noted that the preset rendering effect index threshold is set by professionals according to actual conditions, and no specific limitation or elaboration is made.

[0097] It should be noted that comparing the rendering effect index with the preset rendering effect index threshold is a key step in evaluating whether the AI-rendered multi-angle wood-based panel cabinet product renderings have reached the expected quality. Specifically, the rendering effect index is calculated based on multiple indicators (such as shadow uniformity coefficient, halo unreasonable coefficient, reflection position difference coefficient, etc.), reflecting the overall performance and realism of the image rendering.

[0098] During the comparison process, if the rendering effect index is not less than the preset rendering effect index threshold, it means that the rendered image performs well in light and shadow processing, material reflection and overall visual effects, is close to the actual physical scene, and can accurately show the texture, gloss and details of the cabinet. Therefore, the rendering result can be considered accurate, without obvious distortion, and no further modification or re-rendering is required.

[0099] However, if the rendering effect index is less than the preset threshold, it means that there is a certain degree of deviation in the rendering result of the image, such as unnatural light and shadow interaction, incorrect reflection position, overly uniform shadows, or halo distortion. In this case, it is necessary to adjust the parameters of the AI ​​rendering system appropriately, such as changing the brightness, position, and angle of the light source, or optimizing the light and shadow processing algorithm to more realistically simulate the lighting effect in the actual environment. After re-rendering, the rendering effect index is calculated again, and the process is repeated until the rendering effect index of the rendered model reaches or exceeds the preset threshold, thereby ensuring that the final cabinet rendering meets the requirements of realism and visual effects.

[0100] In one implementation, the above method makes the AI-based multi-angle artificial board cabinet product rendering consistent with the actual effect, with high rendering accuracy. No unnatural or distorted phenomena will occur, thus enhancing the authenticity and reliability of the final rendering.

[0101] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. An AI-based multi-angle artificial board cabinet product rendering method, characterized in that: The following steps are involved: For the 3D cabinet model that has been rendered, pixel data of the shadow part in the 3D cabinet model is extracted, and a shadow uniformity coefficient is obtained according to the pixel data of the shadow part; Extracting the halo area in the three-dimensional cabinet model, and obtaining the halo unreasonable coefficient according to the halo area in the three-dimensional cabinet model; Extracting three-dimensional coordinate data of the three-dimensional cabinet model, and obtaining a reflection position difference coefficient according to the three-dimensional coordinate data; The rendering effect index is obtained according to the shadow uniformity coefficient, the halo unreasonable coefficient and the reflection position difference coefficient, and the rendering effect index is compared with the preset rendering effect index threshold, and whether re-rendering is needed is determined according to the comparison result.

2. The AI-based multi-angle artificial board cabinet product rendering method according to claim 1 is characterized in that: The shadow uniformity coefficient obtained according to the pixel data of the shadow part includes: Extract the pixel data of the shadow area from the rendered 3D cabinet model, denoted as set P, where the grayscale value of each pixel is I i ; Calculate the mean grayscale value of all pixels in the shadow area The calculation formula is: N represents the total number of pixels in the shadow area, I i Represents the gray value of the i-th pixel; Calculate the variance σ of the shadow grayscale 2 , the calculation formula is: Calculate the shadow gradient G through the Sobel operator i , and calculate the variance σ of the gradient change 2 g , the calculation formula is: In the formula, G i is the gradient value of the i-th pixel, is the mean of all gradient values ​​in the shadow area; Calculate the shadow uniformity coefficient, the calculation formula is: Where WER is the shadow uniformity coefficient, a1 and a2 are preset proportional coefficients, and both a1 and a2 are greater than 0, ∈ is a small constant used to avoid the denominator being zero, and its value is between 0 and 1.

3. The AI-based multi-angle artificial board cabinet product rendering method according to claim 1 is characterized in that: According to the halo area in the three-dimensional cabinet model, the halo unreasonable coefficients include: Extract the halo area from the rendered 3D cabinet model and record it as pixel set T, where the brightness value of each pixel is S j , and obtain the color information B of the halo area j And light source distance information Y j ; Calculate the halo brightness attenuation factor: according to the distance Y of the light source j , calculate the brightness attenuation of the halo area, the calculation formula is: Where P j is the brightness attenuation factor of the jth pixel, S j is the brightness value of the jth pixel, Y j is the distance from the jth pixel to the light source; ε is a small constant used to avoid the denominator being zero, and its value is between 0 and 1; Calculate the halo color deviation factor: extract the color information B of the halo area j , calculate the color uniformity deviation, the calculation formula is: Where DR is the color uniformity deviation, M is the total number of pixels in the halo area, and B j is the color value of the jth pixel, B s Is the color value of the light source.

4. The AI-based multi-angle artificial board cabinet product rendering method according to claim 3 is characterized in that: According to the halo area in the three-dimensional cabinet model, the halo unreasonable coefficient also includes: Calculate the diffusion coefficient E of the halo boundary using the following formula: Where R is the total number of pixels in the halo edge area, is the brightness gradient change of the j-th pixel in the x and y directions; Calculate the halo unreasonable coefficient: The halo unreasonable coefficient is calculated by comprehensively calculating the brightness attenuation factor, color uniformity deviation and diffusion coefficient of the halo boundary. The calculation formula is: Where PKU is the halo unreasonable coefficient, b1, b2, and b3 are the preset proportional coefficients, and b1, b2, and b3 are all greater than 0.

5. The AI-based multi-angle artificial board cabinet product rendering method according to claim 1 is characterized in that: The reflection position difference coefficients obtained based on the three-dimensional coordinate data include: Define the position of the light source in the rendered 3D cabinet model in 3D space, denoted as S(x s ,y s , z s ), where x s ,y s , z s Respectively represent the coordinates of the light source on the x, y, and z axes; Extract the pixel set H of the reflective points from the rendered model, and the position of each reflective point is H q (x q ,y q , z q ), where x q ,y q , z q is the coordinate of the qth reflection point in three-dimensional space; Calculate the distance from the reflection point to the light source: For each reflection point H q , calculate its distance D from the light source S q , the calculation formula is: The normal vector N based on the light source position and the object surface q Calculate the expected reflection position L q , the reflection position is usually in the middle between the light source and the reflection point: the calculation formula is: L q =H q +N q ; Calculate the difference between the actual position of each reflection point and the expected reflection position In the formula, is the expected position coordinate of the qth reflection point; Calculate the reflection position difference coefficient DTY, the calculation formula is: Where T is the total number of reflective points.

6. The AI-based multi-angle artificial board cabinet product rendering method according to claim 1 is characterized in that: The rendering effect index obtained based on the shadow uniformity coefficient, halo unreasonable coefficient and reflection position difference coefficient includes: Wherein, Dax is the rendering effect index, WER, PKU, and DTY are the shadow uniformity coefficient, halo unreasonable coefficient, and reflection position difference coefficient, respectively; f1, f2, and f3 are the preset proportional coefficients of the shadow uniformity coefficient, halo unreasonable coefficient, and reflection position difference coefficient, respectively; and f1, f2, and f3 are all greater than 0.

7. The AI-based multi-angle artificial board cabinet product rendering method according to claim 1 is characterized in that: Comparing the rendering effect index with the preset rendering effect index threshold, and determining whether re-rendering is required based on the comparison result includes: Compare the rendering effect index with the preset rendering effect index threshold. If the rendering effect index is not less than the preset rendering effect index threshold, no re-rendering is required. If the rendering effect index is less than the preset rendering effect index threshold, re-rendering is required until the rendering effect index of the rendered model is not less than the preset rendering effect index threshold.

Images