A behavior feedback-based rendering parameter configuration intelligent recommendation system and method

Abstract

The application discloses a rendering parameter configuration intelligent recommendation system and method based on behavior feedback, relates to the technical field of rendering parameter configuration, and comprises the following steps: calling rendering parameter adjustment records, extracting a scene model corresponding to the adjustment records, obtaining first mark records in the adjustment records according to normal vector data of vertices, obtaining the fitting degree between the adjustment records and a to-be-rendered scene according to different illumination areas in the scene model, obtaining second mark records, obtaining target records, determining a target combination of rendering parameters in the to-be-rendered scene according to the number of vertices corresponding to the target records, parameter values of various rendering parameters and resource occupation values, and outputting a candidate parameter configuration scheme to a user. The application analyzes multi-source data such as rendering parameter adjustment records, vertices, illumination areas and resource occupation values, realizes intelligent recommendation of a rendering parameter configuration scheme, is helpful to generate a picture with rendering effects that are more fitted to users, and improves user satisfaction.

Description

Technical Field

[0001] This invention relates to the field of rendering parameter configuration technology, specifically to an intelligent recommendation system and method for rendering parameter configuration based on behavioral feedback. Background Technology

[0002] Differentiated rendering strategies for different game scenarios can rationally allocate rendering resources and significantly improve scene rendering efficiency. While ensuring smooth operation, they also take into account visual performance, providing users with a more stable and high-quality game visual experience. However, existing rendering parameter configurations mostly use fixed templates or general preset schemes, making it difficult to combine user operation habits and historical behavior data to achieve differentiated and refined adjustment of rendering parameters, and to intelligently output different candidate parameter configuration schemes to users. Ultimately, this leads to poor rendering effects and makes it difficult to meet the needs of personalized game experience. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent recommendation system and method for rendering parameter configuration based on behavioral feedback, so as to solve the problems raised in the prior art.

[0004] To solve the above-mentioned technical problems, the present invention provides the following technical solution: A method for intelligent recommendation of rendering parameter configuration based on behavioral feedback includes the following steps: Retrieve the user's historical rendering parameter adjustment records, extract the scene model corresponding to the adjustment records, and obtain the first marked record in the adjustment records based on the vertex normal vector data in the scene model before and after the user adjusted the rendering parameters. It should be noted that vertices are the most basic spatial coordinate points that constitute various geometric shapes in the scene model. The normal vector of a vertex is a unit vector that is perpendicular to the geometric surface to which the vertex belongs and represents the orientation of the surface at that vertex. It is a key geometric attribute of the vertex. The normal vector of a vertex is an inherent geometric attribute of the scene model itself, and its value does not change with the rendering parameters. Therefore, in this scheme, the normal vector of the vertex can be used as an evaluation index to obtain a reliable record in the rendering process, that is, the first marked record. Obtain the current scene to be rendered. Based on the different lighting areas in the scene model corresponding to the scene to be rendered and the adjustment record, obtain the fit between the adjustment record and the scene to be rendered. Based on the fit, obtain the second marker record in the adjustment record. The adjusted record that is simultaneously the first marked record and the second marked record will be used as the target record; Based on the number of vertices, the values ​​of various rendering parameters, and the resource consumption in the scene model corresponding to the target record, the target combination of rendering parameters in the scene to be rendered is determined, and several candidate parameter configuration schemes are output to the user.

[0005] Preferably, obtaining the first marker record in the adjustment record includes: extracting the user ID, parameter adjustment time and scene model corresponding to the adjustment record, obtaining several vertices in the scene model, and if the normal vectors of each vertex are the same before and after parameter adjustment, then the adjustment record is used as the first marker record, and thus all the first marker records are obtained.

[0006] Preferably, the second marker record obtained in the adjustment record includes: The lighting area of ​​a scene model includes the illuminated area, the occluded area, and the shadow area. The illuminated area is the area illuminated by the light source, the occluded area is the area away from the light source, and the shadow area is the area where the light source is blocked. Game scenes typically include different lighting areas to meet the needs of creating lighting atmosphere, visual layering, and realism in different areas of the scene, while also adapting to the visual presentation and interactive experience of different scenes and plots within the game. Extract the rendered image after adjusting the rendering parameters corresponding to the adjustment record R, and establish a first ratio set based on the display area ratio of different lighting areas in the rendered image; Obtain all object models in the scene to be rendered. Based on the display of object models on the screen during the user's historical gameplay, predict the display area ratio of different lighting areas when the user moves in the scene to be rendered, and establish a second ratio set accordingly. The cosine similarity between the first ratio set and the second ratio set is used as the fit of the adjusted record R. If the fit is greater than the preset threshold, the adjusted record R is used as the second marked record, and then all the second marked records are obtained.

[0007] It should be noted that the greater the cosine similarity, the more similar the first ratio set and the second ratio set are. In other words, the second marker record represents the record where the scene model corresponding to the adjusted record R is highly similar to the scene to be rendered. For such records, the rendering parameters corresponding to their tuning have certain reference value and adaptability for the scene to be rendered. Therefore, using the tuning parameter values ​​corresponding to such records can directly provide a reference for adjusting the rendering parameters of the scene to be rendered, effectively improving the tuning efficiency and adaptability of the scene to be rendered. Cosine similarity is an existing technology and will not be elaborated here.

[0008] Preferably, a second set of ratios is established, including: The system obtains the continuous display duration of each object model on the screen during the user's historical gameplay, and identifies object models whose continuous display duration exceeds a preset duration threshold as the user's models of interest; it then builds a neural network model and trains the neural network model based on the models of interest. Obtain all object models in the current scene to be rendered, and use the trained neural network model to obtain all models of interest in the scene to be rendered; Obtain the pixel area of ​​different lighting regions on each model of interest in the scene to be rendered, collect the pixel areas of the same lighting regions, obtain the display area ratio of different lighting regions, and establish a second ratio set accordingly.

[0009] Preferably, several candidate parameter configuration schemes are output to the user, including: Establish a neural network model, and train the neural network model by recording the number of vertices, the parameter values ​​of various rendering parameters, and the resource consumption values ​​in the corresponding scene model according to the target. The system captures the number of vertices in the scene model corresponding to the scene to be rendered, extracts the adjusted parameter values ​​of the rendering parameters from the target record, generates the target combination corresponding to the rendering parameters based on the trained neural network model, and outputs several candidate parameter configuration schemes to the user based on the target combination.

[0010] Preferably, the neural network model training includes: obtaining the number of vertices, the parameter values ​​of various rendering parameters, and the resource usage value corresponding to several target records; pre-setting a training set and a validation set; training the neural network model with the number of vertices and the parameter values ​​of each rendering parameter as input and the resource usage value as output; and using the trained neural network model to obtain the resource usage value based on the number of vertices and the parameter values ​​of each rendering parameter.

[0011] Preferably, the steps for generating candidate parameter configuration schemes include: Extract all adjusted parameter values ​​of a certain rendering parameter X from the target record. The adjusted parameter values ​​are the parameter values ​​set by the user after manually modifying the rendering parameter X. Since the adjusted parameter values ​​are set by the user based on actual needs and preferences, analyzing them can effectively reduce computational complexity and improve the matching degree of rendering parameters with user preferences and the accuracy of recommendations. Sort the adjusted parameter values ​​in descending order of usage duration to generate a parameter tuning set corresponding to the adjusted rendering parameter X; and then obtain the parameter tuning set corresponding to each adjusted rendering parameter. One adjusted parameter value is selected from each parameter tuning set to obtain a candidate combination of rendering parameters. Based on the number of vertices in the scene model corresponding to the scene to be rendered, and the parameter values ​​of each rendering parameter in the candidate combination, the values ​​are input into the trained neural network model to obtain the predicted resource usage value of the candidate combination. If the predicted resource usage value is less than the preset usage value threshold, the candidate combination of rendering parameters is taken as the target combination, and the sum of the indices of each rendering parameter in the candidate combination is taken as the target value of the candidate combination. Based on each parameter tuning set, extract M target combinations and output candidate parameter configuration recommendations to the user in ascending order of the target values ​​of each target combination.

[0012] A rendering parameter configuration intelligent recommendation system based on behavioral feedback includes a first labeled record extraction module, a second labeled record extraction module, a target record extraction module, and a parameter configuration scheme output module; First marker record extraction module: used to retrieve the user's historical rendering parameter adjustment records, extract the scene model corresponding to the adjustment record, and obtain the first marker record in the adjustment record based on the vertex normal vector data in the scene model before and after the user adjusted the rendering parameters; The second marker record extraction module is used to obtain the current scene to be rendered, and to obtain the fit between the adjustment record and the scene to be rendered based on the different lighting areas in the scene model corresponding to the scene to be rendered and the adjustment record. The second marker record in the adjustment record is obtained based on the fit. Target record extraction module: used to select the adjusted record that is simultaneously the first marked record and the second marked record as the target record; The parameter configuration scheme output module is used to determine the target combination of rendering parameters in the scene to be rendered based on the number of vertices, the parameter values ​​of various rendering parameters, and the resource usage values ​​in the scene model corresponding to the target record, and output several candidate parameter configuration schemes to the user.

[0013] Preferably, the first marker record extraction module includes a first marker record extraction unit; First Mark Record Extraction Unit: Used to extract the user ID, parameter adjustment time and scene model corresponding to the adjustment record, obtain several vertices in the scene model, and if the normal vectors of each vertex are the same before and after parameter adjustment, the adjustment record is used as the first mark record, and then all first mark records are obtained.

[0014] Preferably, the parameter configuration scheme output module includes a neural network model training unit and a parameter configuration scheme output unit; Neural network model training unit: used to build a neural network model. Based on the target, it records the number of vertices in the corresponding scene model, the parameter values ​​of various rendering parameters, and the resource consumption values ​​to realize the training of the neural network model. Parameter configuration scheme output unit: used to capture the number of vertices in the scene model corresponding to the scene to be rendered, extract the adjusted parameter values ​​of the rendering parameters from the target record, generate the target combination corresponding to the rendering parameters according to the trained neural network model, and output several candidate parameter configuration schemes to the user according to the target combination.

[0015] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention provides an intelligent recommendation system and method for rendering parameter configuration based on behavioral feedback, including: retrieving rendering parameter adjustment records, extracting the scene model corresponding to the adjustment records, obtaining a first marker record in the adjustment records based on the vertex normal vector data; obtaining the fit between the adjustment records and the scene to be rendered based on different lighting areas in the scene model, obtaining a second marker record, and obtaining a target record; determining the target combination of rendering parameters in the scene to be rendered based on the number of vertices corresponding to the target record, the parameter values ​​of various rendering parameters, and the resource usage value, and outputting candidate parameter configuration schemes to the user. This invention achieves intelligent recommendation of rendering parameter configuration schemes by analyzing multi-source data such as rendering parameter adjustment records, vertices, lighting areas, and resource usage values, which helps to generate rendering effects that are more suitable for users and improves user satisfaction. Attached Figure Description

[0016] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating an intelligent recommendation method for rendering parameter configuration based on behavioral feedback, according to the present invention. Detailed Implementation

[0018] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0019] Example: Figure 1As shown, this invention provides a technical solution for an intelligent recommendation method for rendering parameter configuration based on behavioral feedback, comprising the following steps: Retrieve the user's historical rendering parameter adjustment records, extract the scene model corresponding to the adjustment records, and obtain the first marker record in the adjustment records based on the vertex normal vector data in the scene model before and after the user's rendering parameter adjustment, including: To ensure the rendering effect meets user requirements, users need to manually adjust the rendering parameters. The adjustment record here refers to the record of the user's manual adjustment of the rendering parameters. In this embodiment, the rendering parameters include multiple types of parameters such as resolution parameters, light source parameters, and camera parameters. The user ID, parameter adjustment time, and scene model corresponding to the adjustment record are extracted. The scene model refers to the digital three-dimensional model built to achieve image rendering, which contains various scene elements required for rendering and is the basis for carrying out rendering parameter adjustment and image rendering.

[0020] Obtain several vertices in the scene model. If the normal vectors of each vertex are the same before and after parameter adjustment, then the adjusted record is used as the first marked record, and then all the first marked records are obtained.

[0021] Obtain the current scene to be rendered. Based on different lighting regions in the scene model corresponding to the scene to be rendered and the adjustment record, obtain the fit between the adjustment record and the scene to be rendered. Based on the fit, obtain the second marker record in the adjustment record, including: The lighting area of ​​a scene model includes the illuminated area, the occluded area, and the shadow area. The illuminated area is the area illuminated by the light source, the occluded area is the area away from the light source, and the shadow area is the area where the light source is blocked. In game scenes, different lighting areas are usually included to meet the needs of creating the light and shadow atmosphere, distinguishing visual layers, and restoring realism in different areas of the scene, while adapting to the visual performance and interactive experience of different scenes and plots in the game.

[0022] Extract the rendered image after adjusting the rendering parameters corresponding to the adjustment record R, and establish a first ratio set based on the display area ratio of different lighting areas in the rendered image; First, obtain multiple rendering frames, number N. Based on the display area of ​​each type of lighting region in the rendering frames, obtain the display area ratio of each type of lighting region. Taking the calculation of the display area ratio of the illuminated region as an example, the formula is:

[0023] Where r1 refers to the ratio of the display area of ​​the illuminated area, N refers to the number of rendered frames, and S refers to the display area of ​​one rendered frame (the display area can be obtained by multiplying the screen length and width of the display device; for example, a 27-inch monitor has a screen length of approximately 59.78cm and a screen width of approximately 33.62cm, so the display area is approximately 2010cm²). 2 ), s1 n This refers to the display area of ​​the illuminated region on the nth rendered frame. Then, the display area ratios r2 and r3 of the occluded and shadowed regions can be obtained, and the first set of ratios is established as {r1, r2, r3}.

[0024] Obtain all object models in the scene to be rendered. Based on the display of object models on the screen during the user's historical gameplay, predict the ratio of the displayed area of ​​different lighting regions when the user moves in the scene to be rendered, and establish a second ratio set accordingly. The system obtains the continuous display duration of each object model on the screen during the user's historical gameplay, and identifies object models whose continuous display duration exceeds a preset duration threshold as the user's models of interest; it then builds a neural network model and trains the neural network model based on the models of interest. Object models refer to the core entity models in the rendering scene, including vehicles, props, and interactive facilities. Since different object models have different levels of visual appeal to users, the content rendered on the screen will also differ when users move around in the scene to be rendered. In this embodiment, the preset time threshold is 5 seconds. If a user views an object model for more than 5 seconds, it means that the object model has a high degree of appeal to the user, so the object model is regarded as the model of interest.

[0025] The process of training a neural network model based on a model of interest is an existing technique and will not be elaborated here. The trained neural network model is used here to determine whether a given object model belongs to the model of interest.

[0026] Obtain all object models in the current scene to be rendered, and use the trained neural network model to obtain all models of interest in the scene to be rendered; Obtain the pixel area of ​​different lighting regions on each model of interest in the scene to be rendered, collect the pixel areas of the same lighting regions, obtain the display area ratio of different lighting regions, and establish a second ratio set accordingly.

[0027] First, obtain multiple models of interest, number H. Based on the pixel area of ​​each lighting region in the models of interest, obtain the display area ratio of each lighting region. Here, we take calculating the display area ratio of the illuminated region as an example, and the formula is:

[0028] Where t1 refers to the ratio of the display area of ​​the illuminated region, H refers to the number of models of interest, and w1 h W refers to the display area (specifically, pixel area) of the illuminated region on the h-th model of interest. h This refers to the total pixel area of ​​the h-th model of interest. Then, the ratios t2 and t3 of the displayed area of ​​the occluded and shadowed regions can be obtained, establishing a second set of ratios {t1, t2, t3}.

[0029] The cosine similarity between the first ratio set and the second ratio set is used as the fit of the adjusted record R. If the fit is greater than the preset threshold, the adjusted record R is used as the second marked record, and then all the second marked records are obtained.

[0030] It should be noted that the greater the cosine similarity, the more similar the first ratio set and the second ratio set are. In other words, the second marker record represents the record where the scene model corresponding to the adjusted record R is highly similar to the scene to be rendered. For such records, the rendering parameters corresponding to their tuning have certain reference value and adaptability for the scene to be rendered. Therefore, using the tuning parameter values ​​corresponding to such records can directly provide a reference for adjusting the rendering parameters of the scene to be rendered, effectively improving the tuning efficiency and adaptability of the scene to be rendered. Cosine similarity is an existing technology and will not be elaborated here.

[0031] Therefore, by combining the first and second marked records in this scheme, a reliable target record can be obtained, that is, the adjusted record that is simultaneously the first and second marked records is used as the target record.

[0032] Based on the number of vertices, the values ​​of various rendering parameters, and the resource consumption in the scene model corresponding to the target record, the target combination of rendering parameters in the scene to be rendered is determined, and several candidate parameter configuration schemes are output to the user, including: Establish a neural network model, and train the neural network model by recording the number of vertices, the parameter values ​​of various rendering parameters, and the resource consumption values ​​in the corresponding scene model according to the target.

[0033] It should be noted that the GPU, as a graphics processing unit, is the core hardware for image rendering. Its parallel computing capabilities support the geometric calculation of vertices and normals of scene models, as well as the lighting and shading calculations of rendering parameters such as light sources and materials. It directly determines the rendering efficiency, real-time performance, and image quality of the scene to be rendered. It is the key hardware foundation for achieving real-time updates and high-quality presentation of images in game rendering scenes. In this solution, the resource usage value refers to the GPU's hardware resource usage rate. The higher the resource usage value, the more hardware resources such as GPU computing power and video memory are consumed during the rendering calculation process, and the higher the GPU's operating load.

[0034] Obtain the number of vertices, the parameter values ​​of various rendering parameters, and the resource usage value corresponding to several target records. Preset training and validation sets, and set the ratio of training set to validation set to 7:3. Use the number of vertices and the parameter values ​​of each rendering parameter as input and the resource usage value as output to train the neural network model. The trained neural network model is used to obtain the resource usage value based on the number of vertices and the parameter values ​​of each rendering parameter.

[0035] The system captures the number of vertices in the scene model corresponding to the scene to be rendered, extracts the adjusted parameter values ​​from the target record, generates the target combination corresponding to the rendering parameters based on the trained neural network model, and outputs several candidate parameter configuration schemes to the user based on the target combination, including: Extract all adjusted parameter values ​​of a certain rendering parameter X from the target record. The adjusted parameter values ​​are the parameter values ​​set by the user after manually modifying the rendering parameter X. The adjusted parameter values ​​are sorted in descending order of usage duration, starting from 1, followed by 2, 3, 4, and so on. The smaller the number, the more the user prefers this rendering parameter value. This generates a set of tuning parameters corresponding to the adjusted rendering parameter X, and then obtains the tuning set corresponding to each adjusted rendering parameter.

[0036] One adjusted parameter value is selected from each parameter tuning set to obtain a candidate combination of rendering parameters. Based on the number of vertices in the scene model corresponding to the scene to be rendered, and the parameter values ​​of each rendering parameter in the candidate combination, the values ​​are input into the trained neural network model to obtain the predicted resource usage value of the candidate combination. If the predicted resource usage value is less than the preset usage value threshold, the candidate combination is taken as the target combination, and the sum of the indices of each rendering parameter is taken as the target value of the candidate combination.

[0037] The smaller the target value, the higher the user's adaptation to the parameter values ​​of each rendering parameter in the candidate combination of rendering parameters and the stronger the user's preference matching, under the premise of satisfying the GPU resource usage constraint. Therefore, in this solution, it is necessary to output the recommended candidate parameter configuration scheme to the user in the order of the target value from smallest to largest. In this solution, M=10, there are 10 recommended candidate parameter configuration schemes, which are recommended to the user in the order of the target value from smallest to largest. The user can choose one of them according to the actual needs.

[0038] Based on each parameter tuning set, extract M target combinations and output candidate parameter configuration recommendations to the user in ascending order of the target values ​​of each target combination.

[0039] This embodiment also provides an intelligent recommendation system for rendering parameter configuration based on behavioral feedback, including a first marker record extraction module, a second marker record extraction module, a target record extraction module, and a parameter configuration scheme output module; First marker record extraction module: used to retrieve the user's historical rendering parameter adjustment records, extract the scene model corresponding to the adjustment record, and obtain the first marker record in the adjustment record based on the vertex normal vector data in the scene model before and after the user adjusted the rendering parameters; The second marker record extraction module is used to obtain the current scene to be rendered, and to obtain the fit between the adjustment record and the scene to be rendered based on the different lighting areas in the scene model corresponding to the scene to be rendered and the adjustment record. The second marker record in the adjustment record is obtained based on the fit. Target record extraction module: used to select the adjusted record that is simultaneously the first marked record and the second marked record as the target record; The parameter configuration scheme output module is used to determine the target combination of rendering parameters in the scene to be rendered based on the number of vertices, the parameter values ​​of various rendering parameters, and the resource usage values ​​in the scene model corresponding to the target record, and output several candidate parameter configuration schemes to the user.

[0040] Finally, it should be noted that the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent recommendation of rendering parameter configuration based on behavioral feedback, characterized in that, Includes the following steps: Retrieve the user's historical rendering parameter adjustment records, extract the scene model corresponding to the adjustment records, and obtain the first marked record in the adjustment records based on the vertex normal vector data in the scene model before and after the user adjusted the rendering parameters. Obtain the current scene to be rendered. Based on the different lighting areas in the scene model corresponding to the scene to be rendered and the adjustment record, obtain the fit between the adjustment record and the scene to be rendered. Based on the fit, obtain the second marker record in the adjustment record. The adjusted record that is simultaneously the first marked record and the second marked record will be used as the target record; Based on the number of vertices, the values ​​of various rendering parameters, and the resource consumption in the scene model corresponding to the target record, the target combination of rendering parameters in the scene to be rendered is determined, and several candidate parameter configuration schemes are output to the user. Obtaining the first marker record in the adjustment record includes: extracting the user ID, parameter adjustment time and scene model corresponding to the adjustment record, obtaining several vertices in the scene model, and if the normal vectors of each vertex are the same before and after parameter adjustment, then the adjustment record is used as the first marker record, and then all first marker records are obtained. The second marked record in the adjusted record is obtained, including: The lighting area of ​​the scene model includes the illuminated area, the occluded area, and the shadow area. The illuminated area is the area illuminated by the light source, the occluded area is the area facing away from the light source, and the shadow area is the area where the light source is blocked. Extract the rendered image after adjusting the rendering parameters corresponding to the adjustment record R, and establish a first ratio set based on the display area ratio of different lighting areas in the rendered image; Obtain all object models in the scene to be rendered. Based on the display of object models on the screen during the user's historical gameplay, predict the display area ratio of different lighting areas when the user moves in the scene to be rendered, and establish a second ratio set accordingly. The cosine similarity between the first ratio set and the second ratio set is used as the fit of the adjusted record R. If the fit is greater than the preset threshold, the adjusted record R is used as the second marked record, and then all the second marked records are obtained.

2. The intelligent recommendation method for rendering parameter configuration based on behavioral feedback according to claim 1, characterized in that, Establish a second set of ratios, including: The system obtains the continuous display duration of each object model on the screen during the user's historical gameplay, and identifies object models whose continuous display duration exceeds a preset duration threshold as the user's models of interest; it then builds a neural network model and trains the neural network model based on the models of interest. Obtain all object models in the current scene to be rendered, and use the trained neural network model to obtain all models of interest in the scene to be rendered; Obtain the pixel area of ​​different lighting regions on each model of interest in the scene to be rendered, collect the pixel areas of the same lighting regions, obtain the display area ratio of different lighting regions, and establish a second ratio set accordingly.

3. The intelligent recommendation method for rendering parameter configuration based on behavioral feedback according to claim 1, characterized in that, Output several candidate parameter configuration schemes to the user, including: Establish a neural network model, and train the neural network model by recording the number of vertices, the parameter values ​​of various rendering parameters, and the resource consumption values ​​in the corresponding scene model according to the target. The system captures the number of vertices in the scene model corresponding to the scene to be rendered, extracts the adjusted parameter values ​​of the rendering parameters from the target record, generates the target combination corresponding to the rendering parameters based on the trained neural network model, and outputs several candidate parameter configuration schemes to the user based on the target combination.

4. The intelligent recommendation method for rendering parameter configuration based on behavioral feedback according to claim 3, characterized in that, The neural network model training includes: obtaining the number of vertices, the parameter values ​​of various rendering parameters, and the resource usage value corresponding to several target records; pre-setting training and validation sets; using the number of vertices and the parameter values ​​of each rendering parameter as input and the resource usage value as output to train the neural network model; and using the trained neural network model to obtain the resource usage value based on the number of vertices and the parameter values ​​of each rendering parameter.

5. The intelligent recommendation method for rendering parameter configuration based on behavioral feedback according to claim 3, characterized in that, The steps to generate candidate parameter configuration schemes include: Extract all adjusted parameter values ​​of a certain rendering parameter X from the target record. The adjusted parameter values ​​are the parameter values ​​set by the user after manually modifying the rendering parameter X. Sort the adjusted parameter values ​​in descending order of usage duration to generate a parameter tuning set corresponding to the adjusted rendering parameter X; and then obtain the parameter tuning set corresponding to each adjusted rendering parameter. One adjusted parameter value is selected from each parameter tuning set to obtain a candidate combination of rendering parameters. Based on the number of vertices in the scene model corresponding to the scene to be rendered and the parameter values ​​of each rendering parameter in the candidate combination, the values ​​are input into the trained neural network model to obtain the predicted resource usage value of the candidate combination. If the predicted resource usage value is less than the preset usage value threshold, the candidate combination of rendering parameters is taken as the target combination, and the sum of the indices of each rendering parameter is taken as the target value of the candidate combination. Based on each parameter tuning set, extract M target combinations and output candidate parameter configuration recommendations to the user in ascending order of the target values ​​of each target combination.

6. A rendering parameter configuration intelligent recommendation system based on behavioral feedback, used to execute the rendering parameter configuration intelligent recommendation method based on behavioral feedback as described in any one of claims 1-5, characterized in that, The system includes a first marker record extraction module, a second marker record extraction module, a target record extraction module, and a parameter configuration scheme output module; First marker record extraction module: used to retrieve the user's historical rendering parameter adjustment records, extract the scene model corresponding to the adjustment record, and obtain the first marker record in the adjustment record based on the vertex normal vector data in the scene model before and after the user adjusted the rendering parameters; The second marker record extraction module is used to obtain the current scene to be rendered, and to obtain the fit between the adjustment record and the scene to be rendered based on the different lighting areas in the scene model corresponding to the scene to be rendered and the adjustment record. The second marker record in the adjustment record is obtained based on the fit. Target record extraction module: used to select the adjusted record that is simultaneously the first marked record and the second marked record as the target record; The parameter configuration scheme output module is used to determine the target combination of rendering parameters in the scene to be rendered based on the number of vertices, the parameter values ​​of various rendering parameters, and the resource usage values ​​in the scene model corresponding to the target record, and output several candidate parameter configuration schemes to the user.

7. The intelligent recommendation system for rendering parameter configuration based on behavioral feedback according to claim 6, characterized in that, The first marker record extraction module includes a first marker record extraction unit; First marker record extraction unit: used to extract the user ID, parameter adjustment time and scene model corresponding to the adjustment record, obtain several vertices in the scene model, and if the normal vectors of each vertex are the same before and after parameter adjustment, the adjustment record is used as the first marker record, and then all first marker records are obtained.

8. The intelligent recommendation system for rendering parameter configuration based on behavioral feedback according to claim 6, characterized in that, The parameter configuration scheme output module includes a neural network model training unit and a parameter configuration scheme output unit; Neural network model training unit: used to build a neural network model. Based on the target, it records the number of vertices in the corresponding scene model, the parameter values ​​of various rendering parameters, and the resource consumption values ​​to realize the training of the neural network model. Parameter configuration scheme output unit: used to capture the number of vertices in the scene model corresponding to the scene to be rendered, extract the adjusted parameter values ​​of the rendering parameters from the target record, generate the target combination corresponding to the rendering parameters according to the trained neural network model, and output several candidate parameter configuration schemes to the user according to the target combination.

Images