Computer image generation method and system

Abstract

The invention relates to the technical field of image generation, in particular to a computer image generation method and system. The method comprises the following steps: acquiring current interactive behavior data of a user in a virtual environment and current rendering parameters of each image area in the virtual environment; according to the current interactive behavior data, determining the rendering fluency demand degree of a user for a virtual environment and the perception sensitivity of the user for picture details; performing image area judgment on the rendering fluency demand degree and the perception sensitivity to obtain an image adjustment area and an image adjustment coefficient; and adjusting the current rendering parameter of each image in the image adjustment area by using the image adjustment coefficient to obtain all images after rendering adjustment of the virtual environment. The method is used for solving the problems that when an existing computer image generation method is used for processing high-complexity virtual reality assets, a resource allocation mechanism is not accurate, computing resource utilization is unbalanced, the rendering frame rate is reduced, and the visual fidelity is insufficient.

Description

Technical Field

[0001] This invention relates to the field of image generation technology, and more specifically to a computer image generation method and system. Background Technology

[0002] In the digital creative industry, virtual reality content provides users with immersive experiences. As the pursuit of realism and immersion continues to increase, virtual reality projects are introducing significantly more complex and refined assets, such as high-resolution PBR textures and intricate geometric surfaces. Existing computer graphics generation methods typically employ resource allocation mechanisms based on preset rules and evaluation models for scenes of moderate complexity. When faced with highly complex assets, their scene complexity evaluation modules often struggle to accurately identify the true rendering overhead of the assets, leading to an imbalance in the allocation of computational resources such as graphics processing unit (GPU) cores and memory resources such as the read / write bandwidth of video memory chips.

[0003] Existing computer graphics generation methods suffer from inaccurate resource allocation schemes, directly leading to local imbalances in computational resource utilization. Simultaneously, this also causes severe local imbalances in computational resource utilization within the graphics processing unit (GPU). Furthermore, the concentrated processing of complex assets, especially those containing high-resolution PBR textures, results in a surge in read / write requests to specific areas of video memory. Frequent and intensive access patterns to local video memory regions severely strain the internal bandwidth and latency processing capabilities of the video memory. When the rendering time of a single frame exceeds its allocated budget, new frames cannot be delivered at the required frequency. The accumulated frame rendering time exceeding limits leads to a significant drop in the overall rendering frame rate. Ultimately, the virtual reality content presented to the user is degraded: resulting in decreased smoothness, significantly insufficient visual fidelity and detail, making it difficult for users to enjoy an immersive and high-quality experience. Summary of the Invention

[0004] The purpose of this invention is to provide a computer image generation method and system to solve the problem that existing computer image generation methods, when processing highly complex virtual reality assets, suffer from inaccurate resource allocation mechanisms, leading to unbalanced utilization of computing resources, resulting in decreased rendering frame rates and insufficient visual fidelity.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a computer image generation method, comprising: Obtain the user's current interactive behavior data in the virtual environment and the current rendering parameters of each image region in the virtual environment; Based on the current interaction behavior data, determine the user's demand for smooth rendering of the virtual environment and their sensitivity to visual details; Image region determination is performed on the rendering smoothness requirement and perceptual sensitivity to obtain the image adjustment region and image adjustment coefficient; By using image adjustment coefficients, the current rendering parameters of each image in the image adjustment region are adjusted to obtain all images after rendering adjustment of the virtual environment.

[0006] Furthermore, in the step of determining the image region based on the rendering smoothness requirement and perceptual sensitivity to obtain the image adjustment region and image adjustment coefficient, the step of obtaining the image adjustment coefficient includes: The rendering smoothness requirement and perception sensitivity are evaluated by image region determination to determine the dynamic overhead environment parameters and initial image adjustment coefficients in the virtual environment. The dynamic overhead rendering cost is determined based on the dynamic overhead environment parameters and the preset rendering cost library. The rendering cost ratio is obtained by comparing the dynamic overhead rendering cost with a preset cost threshold. The initial coefficients of the image are adjusted using the rendering cost ratio to obtain the image adjustment coefficients.

[0007] Furthermore, in the step of determining the image adjustment region and image adjustment coefficient based on the rendering smoothness requirement and perceptual sensitivity, the steps for obtaining the image adjustment region include: The rendering smoothness requirement and perception sensitivity are judged by image region, and the rendering recognition result of each image region containing the type of rendering object, semantic role and potential interaction possibility is determined; Based on the rendering recognition results, the image adjustment area is determined.

[0008] Furthermore, the step of adjusting the current rendering parameters of each image in the image adjustment region using image adjustment coefficients to obtain all images after rendering adjustment of the virtual environment includes: Based on the current interaction behavior data, determine the importance coefficient of each rendered image; Based on the importance coefficient of each rendered image, the interaction priority area within the user's attention area is determined. By using the interaction priority region and image adjustment coefficients, the current rendering parameters of each image in the image adjustment region are adjusted to obtain all images after rendering adjustment of the virtual environment.

[0009] Furthermore, the step of determining the importance coefficient of each rendered image based on the current interaction behavior data includes: Based on the current interaction behavior data, confirm the narrative progress information, currently activated task objective information, and user personalized preference information in the virtual environment; Coefficient analysis is performed on the narrative progress information, the task objective information, and the user personalized preference information to determine the importance coefficient of each rendered image.

[0010] Furthermore, the steps of confirming narrative progress information, currently activated task objective information, and user personalized preference information in the virtual environment based on the current interaction behavior data include: Confirm the event publication status in the current interaction behavior data, wherein the event publication status includes the event type, occurrence time, and the scene object identifier associated with the event; The event publication status is parsed to obtain the timestamp of the event publication status; By utilizing the timestamp of the event publication status and the current interaction behavior data, the virtual environment state information in the narrative state register can be extracted; Data analysis is performed on the virtual environment state information in the narrative state register to confirm the narrative progress information, currently active task objective information, and user personalized preference information in the virtual environment.

[0011] Further, the steps of performing data analysis on the virtual environment state information of the narrative state register to confirm the narrative progress information, currently active task objective information, and user personalized preference information in the virtual environment include: Data analysis is performed on the virtual environment state information of the narrative state register to confirm the original information of the narrative progress, the original information of the currently activated task objective, and the original information of the user's personalized preferences in the virtual environment; The original information of the narrative progress, the original information of the currently activated task objective, and the original information of the user's personalized preferences in the virtual environment are subjected to integrity checks and consistency verifications to obtain the narrative progress information, the currently activated task objective information, and the user's personalized preference information in the virtual environment.

[0012] Furthermore, based on the current interaction behavior data, the steps of determining the user's demand for smooth rendering of the virtual environment and their sensitivity to visual details include: Based on the current interaction behavior data, determine the image data of each frame of the user in the virtual environment within a certain period of time; Image analysis is performed on each frame of image data to determine the user's requirements for smooth rendering of the virtual environment and their sensitivity to image details.

[0013] Furthermore, the steps of performing image analysis on each frame of image data to determine the user's requirements for rendering smoothness of the virtual environment and their sensitivity to image details include: Image analysis is performed on each frame of image data to determine the smoothness parameters and perception parameters of each frame of image data; By using preset smoothness parameters, the smoothness parameters of each frame of image data are compared and analyzed to determine the user's rendering smoothness requirements for the virtual environment. By using preset perception parameters, the perception parameters of each frame of image data are compared and analyzed to determine the user's sensitivity to image details.

[0014] The present invention also provides a computer image generation system, the system comprising: The information acquisition module is used to acquire the user's current interactive behavior data in the virtual environment and the current rendering parameters of each image region in the virtual environment; The data determination module is used to determine the user's demand for smooth rendering of the virtual environment and sensitivity to visual details based on the current interaction behavior data. The coefficient judgment module is used to judge the image region based on the rendering smoothness requirement and perceptual sensitivity, and obtain the image adjustment region and image adjustment coefficient. The image adjustment module is used to adjust the current rendering parameters of each image in the image adjustment area using image adjustment coefficients, so as to obtain all images after rendering adjustment of the virtual environment.

[0015] Compared with the prior art, the computer image generation method and system of the present invention have the following advantages: This invention acquires user's current interactive behavior data and the current rendering parameters of each image region in a virtual environment. Based on the interactive behavior data, it determines the user's demand for rendering smoothness and their sensitivity to image detail. Image regions are then judged based on the rendering smoothness demand and sensitivity to perception, resulting in image adjustment regions and coefficients. These coefficients are then used to adjust the current rendering parameters of each image within the adjustment regions, ultimately yielding all images of the virtual environment after rendering adjustments. This allows for dynamic and intelligent adjustment of the virtual environment's rendering parameters based on the user's real-time perception and interactive needs, avoiding imbalances in computational resource utilization caused by inaccurate resource allocation. This effectively solves the problems of decreased rendering frame rate and insufficient visual fidelity, significantly improving the smoothness and immersion of virtual reality content, and providing users with a superior virtual experience. Attached Figure Description

[0016] To more clearly illustrate the specific embodiments of the present invention, the accompanying drawings used in the specific embodiments will be briefly described below. In all the drawings, the elements or parts are not necessarily drawn to scale.

[0017] Figure 1 This is a flowchart of a computer image generation method according to the present invention.

[0018] Figure 2This is a structural block diagram of a computer image generation system according to the present invention.

[0019] In the diagram: 210, Information Acquisition Module; 220, Data Determination Module; 230, Coefficient Judgment Module; 240, Image Adjustment Module.

[0020] The implementation and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0021] The following drawings disclose several embodiments of the present invention. For clarity, many practical details will be described in the following description. However, it should be understood that these practical details are not intended to limit the invention. That is, in some embodiments of the invention, these practical details are not essential. Furthermore, for the sake of simplicity, some conventional structures and components will be shown in the drawings in a simple schematic manner.

[0022] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0023] Furthermore, in this invention, the use of terms such as "first" and "second" is for descriptive purposes only and does not specifically refer to any order or sequence, nor is it intended to limit the invention. They are merely used to distinguish components or operations described using the same technical terms, and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of various embodiments can be combined with each other, but only if they are feasible for those skilled in the art. If a combination of technical solutions is contradictory or impossible to implement, such a combination should be considered nonexistent and not within the scope of protection claimed by this invention.

[0024] In the digital creative industry, virtual reality (VR) content provides users with immersive experiences. However, with the increasing pursuit of realism and immersion, VR projects introduce significantly more complex and detailed assets, such as high-resolution PBR textures and intricate geometric surfaces. Existing computer graphics generation methods employ resource allocation mechanisms based on preset rules and evaluation models for scenes of typical complexity. When faced with highly complex assets, their scene complexity evaluation modules often struggle to accurately identify the actual rendering overhead of these assets, leading to an imbalance in the allocation of computing resources (especially GPU cores) and video memory resources (especially the read / write bandwidth of video memory chips). This inaccurate resource allocation results in localized imbalances in the utilization of computing resources. Simultaneously, the cumulative effect of continuous congestion in the rendering pipeline and video memory data read latency causes the actual rendering time of a single frame to significantly exceed the preset time threshold. Consequently, the VR content presented to users suffers a double degradation: the reduced frame rate leads to decreased smoothness, while visual fidelity and detail are also significantly insufficient, making it difficult for users to enjoy an immersive and high-quality experience.

[0025] To further understand the content, features, and effects of this invention, the following embodiments are provided, and detailed descriptions are given below in conjunction with the accompanying drawings: Please see Figure 1 This invention provides a computer image generation method, comprising the following steps: S100. Acquire the user's current interactive behavior data in the virtual environment and the current rendering parameters of each image region in the virtual environment. The virtual environment refers to a simulated world created using computer technology that allows users to interact and perceive, such as a 3D scene in virtual reality (VR) or augmented reality (AR) applications. Current interactive behavior data refers to data generated when the user operates within the virtual environment, such as the user's gaze direction, movement trajectory, and gestures. An image region refers to a visual area in the virtual environment divided into different parts, each region potentially containing different rendering objects and complexities. Rendering parameters refer to various settings that affect image rendering quality and performance, such as texture resolution, level of detail (LOD), lighting effects, and shadow quality. This step can acquire the user's gaze direction and fixation point using eye-tracking sensors integrated into the virtual reality headset, and include this as part of the current interactive behavior data. Simultaneously, the current rendering parameters of each image region, such as texture resolution and LOD, can be directly read from the rendering engine. Alternatively, user interactive behavior data can be acquired by analyzing user input such as controller actions or voice commands within the virtual environment. In addition, the rendering status and parameters of each image region can be captured in real time in the rendering pipeline through the preset scene analysis module.

[0026] S200. Based on the current interaction behavior data, determine the user's demand for smooth rendering of the virtual environment and their sensitivity to visual details. The demand for smooth rendering refers to the user's expectation of the frequency and consistency of the virtual environment's visual updates. Perceptual sensitivity refers to the user's awareness of changes in visual details and a decrease in visual quality. This step can determine the user's demand for smooth rendering based on their movement speed and head rotation speed within the virtual environment. For example, when a user moves quickly or frequently turns their head, their demand for smooth rendering is considered high. Simultaneously, the user's sensitivity to visual details can be determined based on the time they spend in a specific area and their zooming actions in that area. For example, when a user stares at an area for a long time and zooms in, their sensitivity to details in that area is considered high. Alternatively, a machine learning model can be used. By analyzing a large amount of user interaction behavior data and corresponding user feedback, a predictive model can be trained. This model can predict the user's sensitivity to smooth rendering and visual details based on the current interaction behavior data.

[0027] S300. The rendering smoothness requirement and perceptual sensitivity are used to determine the image region, resulting in an image adjustment region and an image adjustment coefficient. The image adjustment region refers to a specific image region requiring rendering parameter adjustment. The image adjustment coefficient refers to the specific value or proportion used to adjust the rendering parameters. This step can match the rendering smoothness requirement and perceptual sensitivity with different image regions in the virtual environment based on a preset rule base. For example, for the central region currently being viewed by the user, if the user has a high perceptual sensitivity to image details, this region may be identified as an image adjustment region and assigned a higher image adjustment coefficient to improve its rendering quality. For regions at the edge of the user's line of sight, if the user has a high requirement for rendering smoothness, this region may be identified as an image adjustment region and assigned a lower image adjustment coefficient to reduce its rendering overhead. Alternatively, a fuzzy logic-based inference engine can be used, taking the rendering smoothness requirement and perceptual sensitivity as input, and using fuzzy rules and membership functions to derive the adjustment priority and adjustment magnitude of each image region, thereby obtaining the image adjustment region and image adjustment coefficient.

[0028] S400. Using image adjustment coefficients, adjust the current rendering parameters of each image in the image adjustment region to obtain all images after rendering adjustments for the virtual environment. This step can directly apply the image adjustment coefficients to the rendering parameters in the image adjustment region. For example, if the image adjustment coefficients indicate improved rendering quality, the texture resolution of the region can be increased, the level of model detail can be improved, or more complex lighting effects can be enabled. If the image adjustment coefficients indicate reduced rendering overhead, the texture resolution of the region can be reduced, the level of model detail can be simplified, or certain visual effects can be disabled. Alternatively, a rendering parameter adjustment strategy library can be maintained, containing specific rendering parameter adjustment schemes corresponding to different image adjustment coefficients. After obtaining the image adjustment coefficients, a corresponding adjustment scheme is selected from the strategy library and applied to the current rendering parameters of each image in the image adjustment region.

[0029] In this embodiment, the current interactive behavior data of the user in the virtual environment and the current rendering parameters of each image region in the virtual environment are introduced as the basis for dynamic adjustment. Specifically, firstly, the current interactive behavior data of the user in the virtual environment and the current rendering parameters of each image region in the virtual environment are obtained. This provides a real-time basis for subsequent dynamic adjustments. Subsequently, based on the current interactive behavior data, the user's rendering smoothness requirements and sensitivity to image details in the virtual environment are determined. By incorporating the user's personalized perception requirements into the rendering decision, rendering optimization becomes more aligned with the user experience. Next, image region judgment is performed based on the rendering smoothness requirements and perception sensitivity to obtain image adjustment regions and image adjustment coefficients. This identifies which image regions need adjustment, as well as the degree and direction of adjustment. For example, for regions that the user pays close attention to and is sensitive to details, their rendering quality may be improved; while for regions that the user does not pay attention to or have higher requirements for smoothness, their rendering overhead may be appropriately reduced. Finally, using the image adjustment coefficients, the current rendering parameters of each image in the image adjustment region are adjusted to obtain all images after rendering adjustments to the virtual environment. This invention can dynamically and intelligently adjust the rendering parameters of the virtual environment based on the user's real-time perception and interaction needs, avoiding the imbalance in computing resource utilization caused by inaccurate resource allocation. This effectively solves the problems of decreased rendering frame rate and insufficient visual fidelity, significantly improving the smoothness and immersion of virtual reality content, and providing users with a superior virtual experience.

[0030] In some embodiments of this application described above, the step of determining the image region and obtaining the image adjustment region and image adjustment coefficient based on the rendering smoothness requirement and perceptual sensitivity includes the following step: Image region determination is performed based on the rendering smoothness requirement and perceptual sensitivity to determine the dynamic overhead environment parameters and initial image adjustment coefficients in the virtual environment. The dynamic overhead environment parameters are real-time performance indicators reflecting the current operating state of the virtual environment, such as CPU utilization, GPU utilization, memory usage, network latency, and frame rate fluctuations, which directly affect the actual rendering cost and system load. The initial image adjustment coefficients are adjustment coefficients initially determined based on the rendering smoothness requirement and perceptual sensitivity without considering dynamic overhead; these may be baseline values ​​or preliminary estimates.

[0031] The dynamic overhead rendering cost is determined based on the dynamic overhead environment parameters and the preset rendering cost library. Specifically, the preset rendering cost library is a dataset storing the expected computational costs for different rendering operations, different image regions, or different combinations of rendering parameters. For example, it may contain the CPU / GPU time or memory consumption required to render a high-detail model, apply a complex shader, or process a specific particle effect. By matching and calculating the real-time acquired dynamic overhead environment parameters with the data in the preset rendering cost library, the dynamic overhead rendering cost generated by the current virtual environment during rendering adjustments can be obtained.

[0032] The rendering cost ratio is obtained by comparing the dynamic overhead rendering cost with a preset cost threshold. The preset cost threshold is a system-defined acceptable upper limit for rendering cost, used to measure whether the current rendering overhead is within a controllable range. The rendering cost ratio can be a dimensionless proportion, such as the ratio of dynamic overhead rendering cost to the preset cost threshold, reflecting the degree of deviation between the actual cost of the current rendering task and the expected cost.

[0033] The initial image adjustment coefficients are adjusted using the rendering cost ratio to obtain the image adjustment coefficients. If the rendering cost ratio is high (e.g., the actual cost far exceeds a threshold), the initial image adjustment coefficients may be adjusted downwards to reduce rendering quality or complexity, thereby reducing resource consumption. Conversely, if the rendering cost ratio is low, the initial image adjustment coefficients may be adjusted upwards to improve rendering quality and fully utilize available resources. In this way, the image adjustment coefficients can more intelligently adapt to the real-time performance of the virtual environment.

[0034] Specifically, when moving at high speed in a complex virtual city environment, based on the user's current interaction data, it is determined that the user has an extremely high demand for rendering smoothness, while their sensitivity to distant details is relatively low. At this point, initial image adjustment coefficients are generated, prioritizing smoothness. However, after determining these initial coefficients, it is further detected that the current GPU utilization has reached 90% (dynamic overhead environment parameter), and several highly complex NPCs (non-player characters) have entered the user's field of vision. Based on the preset rendering cost library, it is determined that the dynamic overhead rendering cost of the current scene far exceeds the preset cost threshold, resulting in a high rendering cost ratio. Using this rendering cost ratio, the previously determined initial image adjustment coefficients are adjusted downwards. Specifically, this might involve further reducing the texture resolution of distant buildings, reducing particle effects in non-interactive areas, or simplifying shadow calculations, thus obtaining the final image adjustment coefficients. Through these adjustments, even in complex dynamic scenes, a stable rendering frame rate can be ensured, avoiding stuttering, while optimizing resource allocation without affecting the core user experience.

[0035] This embodiment introduces dynamic overhead environment parameters and rendering costs, making the determination of image adjustment coefficients more intelligent and adaptive. Specifically, after obtaining the user's rendering smoothness requirements and sensitivity to image details in the virtual environment, and initially determining the initial image adjustment coefficients, the real-time operating status of the virtual environment, i.e., dynamic overhead environment parameters, is further considered. These parameters, combined with a preset rendering cost library, can accurately calculate the dynamic overhead rendering cost of the current rendering task. By comparing this dynamic overhead rendering cost with a preset cost threshold, the reasonableness of the resource consumption of the current rendering task can be evaluated. The resulting rendering cost ratio serves as a feedback mechanism to correct the initial image adjustment coefficients. For example, when the load is high and the dynamic overhead rendering cost exceeds the preset threshold, the rendering cost ratio will prompt the initial image adjustment coefficients to be adjusted downwards, thereby reducing rendering complexity and ensuring smoothness; conversely, when resources are sufficient, the rendering cost ratio may allow the initial image adjustment coefficients to be adjusted upwards to improve image details and meet the user's perceptual sensitivity. This ensures that the rendering strategy can respond to changes in system performance in real time, avoiding resource waste or performance bottlenecks that may result from fixed coefficients.

[0036] In some embodiments of this application described above, the step of determining the image adjustment region and image adjustment coefficient based on the rendering smoothness requirement and perceptual sensitivity includes the following steps in the image adjustment region acquisition step: The rendering smoothness requirement and perceptual sensitivity are assessed by image region judgment to determine the rendering recognition result for each image region, including the type of rendered object, semantic role, and potential interaction possibility. The rendering recognition result is a set of information obtained after in-depth analysis of the content of each image region in the virtual environment. Specifically, it includes the type of rendered object presented in the image region, such as a character, background building, interactive prop, or special effects particle; the semantic role, i.e., the function or significance of the object in the current virtual scene, such as a primary task objective, a secondary non-player character (NPC), environmental decoration, or a user-interactive element; and the potential interaction possibility, i.e., the probability or frequency of user interaction within the image region. This constitutes a comprehensive understanding of the content of the image region and its importance.

[0037] Based on the rendering recognition results, image adjustment regions are determined. This step involves identifying specific regions requiring rendering parameter adjustments based on a deep understanding of the image region content. For example, regions containing primary task objectives or highly potential interactive objects are typically of higher importance and are therefore more likely to be designated as image adjustment regions to ensure rendering quality and smoothness in critical areas. Conversely, regions containing static backgrounds or low-importance objects may be excluded from image adjustment regions to conserve rendering resources.

[0038] Specifically, when exploring an ancient ruin in a virtual reality game, the first step is to acquire data on the user's current interactive behavior in the virtual environment, as well as the current rendering parameters of each image region within the virtual environment. Next, based on the current interactive behavior data, the user's demand for smooth rendering of the virtual environment and their sensitivity to visual details are determined. Further, when judging image regions based on the demand for smooth rendering and perceptual sensitivity, the following steps are performed to determine image adjustment areas: First, a deep analysis is performed on each image region in the virtual environment to determine its rendering recognition result. For example, in one image region, an ancient scroll (the type of rendered object) is identified, its semantic role being a key prop in the current task, and the user is moving towards it, indicating a high potential for interaction. In another image region, distant background mountains (the type of rendered object) are identified, their semantic role being environmental decoration, and their potential for interaction is zero. Then, image adjustment areas are determined based on the rendering recognition results. Specifically, the area containing the ancient scroll, due to its high importance and high potential interactivity, is identified as an image adjustment area and needs to maintain high rendering quality. Meanwhile, distant background mountain areas may be excluded from the image adjustment area or assigned a lower rendering priority to save computing resources. This allows for fine-tuning of the current rendering parameters of the ancient scroll within the designated image adjustment area using image adjustment coefficients, such as enhancing its texture details or lighting effects, while maintaining lower rendering parameters for the background mountain areas. The result is a final image of the virtual environment after rendering adjustments, thus optimizing overall rendering performance while ensuring the core user experience.

[0039] This embodiment introduces rendering recognition results, enabling the determination of image adjustment areas to go beyond simply relying on the user's surface-level smoothness requirements and perceptual sensitivity, delving into the content level of the virtual environment. Specifically, by identifying the type, semantic role, and potential interaction possibilities of rendered objects within each image region, it can more accurately determine which areas have the greatest impact on the user experience. For example, when users have a high sensitivity to image details, rendering parameters will be adjusted first for areas containing important characters or interactive objects, as the details in these areas are crucial to the user experience. Simultaneously, when users have a high demand for rendering smoothness, rendering recognition results can be used to appropriately downgrade rendering in non-critical areas while ensuring the rendering quality of critical areas, thereby balancing rendering performance and visual effects overall. This content- and semantic-based region judgment makes rendering adjustments more precise and efficient, avoiding the resource waste or poor results associated with adjusting all areas to the same degree.

[0040] In some embodiments of this application described above, the step of adjusting the current rendering parameters of each image in the image adjustment region using image adjustment coefficients to obtain all images after rendering adjustment of the virtual environment includes: Based on the current interaction behavior data, an importance coefficient is determined for each rendered image. Specifically, the importance coefficient can be understood as a quantitative indicator measuring the importance of each image region in the virtual environment to the user's current experience or task objective. For example, by analyzing the user's eye-tracking data, controller or keyboard input, voice commands, and the triggering state of events in the virtual environment, the user's current most focused area or object can be determined. The determination of the importance coefficient can employ a machine learning model, which, after training, can output an importance score for each image region based on multi-dimensional interaction behavior data.

[0041] Based on the importance coefficient of each rendered image, an interaction priority region is determined within the user's attention area. The user attention area refers to the set of objects that the user is currently visually focused on or that could potentially interact with. The interaction priority region is the sub-region within the user attention area that has the highest probability of interaction or the greatest impact on the user experience. For example, if a user is looking at an NPC (non-player character) and preparing to interact with it, then the image area where the NPC is located and its surrounding interaction hotspots will be identified as the interaction priority region. The determination of the interaction priority region can be based on a threshold filtering of the importance coefficient or prediction using a pre-defined interaction model.

[0042] By utilizing interaction priority regions and image adjustment coefficients, the current rendering parameters of each image within the image adjustment region are adjusted to obtain all images after rendering adjustments to the virtual environment. When adjusting the rendering parameters, not only are the overall rendering smoothness requirements and perceptual sensitivity reflected by the image adjustment coefficients considered, but also the areas where the user is most focused and most likely to interact are given special consideration. For images within interaction priority regions, their rendering parameters may be adjusted higher to ensure more refined visual effects and a smoother interactive experience; while for non-interaction priority regions, appropriate downgrading can be performed based on the image adjustment coefficients to save rendering resources.

[0043] Specifically, when a user is exploring a complex scene in a virtual reality (VR) game, the process begins by acquiring data on the user's current interactive behavior within the virtual environment. This includes head posture, eye-tracking data (gaze direction), hand controller inputs (such as pointing or grasping actions), and in-game event triggering information (such as NPC dialogue initiation and quest objective updates). Next, based on this data, the importance coefficient of each rendered image is determined. For example, if the user's gaze lingers on an NPC for an extended period and their hand controller is pointing at that NPC, the importance coefficient of the image area containing that NPC will be significantly increased. Similarly, if a game quest prompts the user to find a specific item, the importance coefficient of the image area containing that item will also increase accordingly. Based on these importance coefficients, priority interaction areas within the user's focus area are further identified. For instance, if an NPC has a high importance coefficient and the user is engaged in dialogue, the area containing the NPC's facial expressions, gestures, and dialogue boxes will be identified as priority interaction areas. If the user is attempting to pick up an item, the item and its surrounding interactive area will be identified as priority interaction areas. Finally, using these priority interaction areas and the previously determined image adjustment coefficients, the current rendering parameters of each image within the image adjustment areas are adjusted. Specifically, for images in interaction-priority areas, their rendering parameters (such as texture resolution, lighting quality, and anti-aliasing level) may be adjusted to the highest level to ensure that users can clearly see details and receive smooth interactive feedback. For non-interaction-priority areas, such as regions in the scene background far from the user's line of sight, their rendering parameters can be appropriately reduced based on the image adjustment coefficients to save computing resources without affecting the core user experience. In this way, overall rendering performance can be optimized while ensuring the core user experience.

[0044] This embodiment avoids failing to fully consider the user's actual focus when adjusting rendering parameters by introducing an importance coefficient for each rendered image and further determining the interaction priority area within the user's attention region based on the importance coefficient. Specifically, when a user interacts in the virtual environment, their behavioral data is used to quantify the importance of different image regions. This identifies the areas where the user is most likely to interact or is most visually focused. Based on this, by determining the interaction priority area, limited rendering resources can be more intelligently allocated to areas with the greatest impact on user experience. For example, in the interaction priority area, rendering parameters can be adjusted higher to provide finer details and lower latency, ensuring the user receives optimal visual and interactive feedback at crucial moments. In non-interaction priority areas, appropriate rendering degradation can be performed based on the overall rendering smoothness requirements and perceptual sensitivity, combined with image adjustment coefficients, to optimize overall rendering performance and avoid unnecessary resource waste. This refined judgment of user attention and interaction priority makes the rendering adjustment process more targeted and efficient.

[0045] In some embodiments of this application described above, the step of determining the importance coefficient of each rendered image based on the current interaction behavior data includes: Based on the current interaction data, narrative progress information, currently active task objective information, and user personalized preference information in the virtual environment are identified. Specifically, narrative progress information refers to narrative-related data such as the current storyline, plot development stage, and key event triggering status in the virtual environment. For example, in a role-playing game, narrative progress information may include the progress of the current main quest, unlocked side quests, and key nodes in NPC (non-player character) dialogues. Its purpose is to identify narrative elements in the virtual environment that currently have high user attention, thereby assigning higher importance to image areas related to these elements. Currently active task objective information can be understood as the specific content and status of the task the user is performing or about to perform in the virtual environment. For example, in exploration games, task objective information may include finding specific items, reaching a designated location, and defeating an enemy. Its purpose is to guide the rendering system to prioritize image areas directly related to the current task objective, ensuring that the user can clearly perceive and complete the task. User personalized preference information specifically refers to data reflecting the user's personal preferences, such as historical behavior patterns, interest tags, and custom settings in the virtual environment. For example, users may prefer to explore specific types of areas, focus on certain characters or objects, or have a particular preference for a certain visual style. The goal is to dynamically adjust the importance of different image areas based on the user's unique interests, providing a more personalized visual experience.

[0046] The narrative progress information, task objective information, and user personalized preference information are analyzed using coefficients to determine the importance coefficient of each rendered image. This step uses a pre-defined algorithm or machine learning model to derive the comprehensive importance coefficient of each image region. For example, different weights can be assigned to each type of information, or a neural network model can be used to learn the complex relationship between user behavior and image importance. The aim is to fuse multi-dimensional information to form a unified and accurate image importance evaluation standard.

[0047] Specifically, when a user is in a virtual adventure game, the current task objective is to find lost treasure, and the narrative progress indicates that the user has just entered an ancient ruin area. Simultaneously, the user's personalized preference information indicates a high interest in historical ruins and puzzle elements. Based on the current interaction data, the following is confirmed: First, narrative progress information: the user is in the ancient ruin exploration stage, possibly involving specific ruin structures, murals, or runes. Second, currently active task objective information: the user needs to find lost treasure, meaning the treasure may be hidden in a specific corner of the ruins or can only be discovered by solving puzzles. Third, user personalized preference information: the user has a preference for historical ruins and puzzles. This information will be analyzed using coefficients. For example, a weighted model can be set: task objective information has the highest weight (e.g., 0.5), narrative progress information is second (e.g., 0.3), and personalized preference information is third (e.g., 0.2). Based on this, specific puzzle areas within the ruins, hidden corners that may contain treasure, and image areas such as murals with historical and cultural characteristics will be assigned higher importance coefficients. For example, the image area of ​​the stone tablet displaying the puzzle will have a significantly higher importance coefficient than the image area of ​​insignificant trees in the background. Therefore, when adjusting the rendering, image areas with high importance coefficients will receive more rendering resources, resulting in higher detail and a more refined visual effect, while other less important areas can have their rendering quality appropriately reduced to balance overall performance and user experience.

[0048] This embodiment incorporates narrative progression information, currently active task objective information, and user personalized preference information, and performs coefficient analysis on them to more comprehensively and accurately capture the user's true focus in the virtual environment. By integrating this multi-dimensional information, the process of determining the importance coefficient of each rendered image becomes more intelligent and contextualized. For example, when the user is at a key plot point, image areas related to the plot are given higher importance; when the user is performing a specific task, areas related to the task objective are given priority; at the same time, the user's personalized preferences also affect the weight of their focused areas. This ensures that rendering resources are more rationally allocated to areas that the user truly cares about, thereby improving rendering efficiency and user experience.

[0049] In some embodiments of this application described above, the step of confirming narrative progress information, currently activated task objective information, and user personalized preference information in the virtual environment based on the current interaction behavior data includes: The event publication status in the current interaction behavior data is confirmed. This event publication status includes the event type, occurrence time, and associated scene object identifier. Specifically, the event publication status refers to a structured record of a meaningful interaction or system event occurring in the virtual environment. This event publication status is designed to include key information such as event type, occurrence time, and associated scene object identifier. The event type indicates the nature of the event, such as a user click, task completion, or NPC dialogue; the occurrence time records the specific moment the event occurred for time series analysis; and the associated scene object identifier indicates the specific entity or area in the virtual environment involved in the event.

[0050] The event publication status is parsed to obtain the timestamp of the event publication status. The purpose of this step is to extract the timestamp information that can be used for subsequent processing from the raw event publication status data. The timestamp is usually a precise time mark that can uniquely identify the order and time of the event.

[0051] By utilizing the timestamp of the event's publication status and current interaction data, virtual environment state information can be extracted from the narrative state register. The narrative state register is a database or memory structure that dynamically stores and manages key information such as the current narrative thread, task progress, or user personalization settings within the virtual environment. By combining the event's timestamp with the user's real-time interaction data, the virtual environment state information most relevant to the current context can be accurately located and extracted.

[0052] Data analysis is performed on the virtual environment state information in the narrative state register to confirm the narrative progress information, currently active task objective information, and user personalized preference information within the virtual environment. This step identifies, integrates, and confirms the narrative progress information, currently active task objective information, and user personalized preference information from the extracted raw state information. For example, narrative progress can be confirmed by analyzing records of task completion in the narrative state register; task objectives can be confirmed by examining the currently active task list; and user personalized preferences can be confirmed by analyzing the user's choices and behavioral patterns in historical interactions.

[0053] Specifically, during the quest to find a lost artifact, when a user interacts with an NPC (non-player character) and obtains a clue about the artifact's location, an event publication status is generated. The event type is NPC dialogue, the occurrence time is the moment the dialogue ends, and the associated scene objects are the NPC and the clue item. This event publication status is parsed to obtain its timestamp. Subsequently, using this timestamp and user interaction data such as current perspective and movement, relevant information is extracted from the narrative state register. The narrative state register may contain the current quest progress (e.g., clue A obtained), the current stage of the main storyline (e.g., exploration stage two), and the user's preferences shown in past quests (e.g., preference for exploration over combat). By analyzing the extracted virtual environment state information, it can be confirmed that the current narrative progress is in the clue acquisition stage, the activated quest objective is to reach the clue-indicated location, and the user's personalized preference is immersive exploration. This precisely confirmed information is then used to calculate the importance coefficient of each rendered image; for example, image areas indicating clue locations are assigned a higher importance coefficient, thus receiving higher priority for detail preservation or smoothness assurance during rendering adjustments.

[0054] This embodiment introduces event publication states and a narrative state register. First, it abstracts complex user interactions in the virtual environment into event publication states with clear structure and time stamps, enabling the standardized capture and understanding of user intent and context. Second, by parsing the event publication states to obtain timestamps and combining them with current interaction data, it can accurately extract virtual environment state information highly relevant to the current moment and user behavior from the narrative state register. The narrative state register, as a centralized hub for managing dynamic information in the virtual environment, ensures the comprehensiveness and consistency of the extracted information. Finally, data analysis of the state information accurately and dynamically confirms narrative progress, task objectives, and user personalized preferences. This avoids direct and coarse judgments on raw, unstructured interaction data, significantly improving the accuracy and real-time nature of information confirmation and providing a solid data foundation for the subsequent accurate calculation of importance coefficients.

[0055] In some embodiments of this application described above, the step of performing data analysis on the virtual environment state information of the narrative state register to confirm the narrative progress information, currently active task objective information, and user personalized preference information in the virtual environment includes: Data analysis is performed on the virtual environment state information in the narrative state register to confirm the original information of narrative progress, the original information of the currently active task objective, and the original information of user personalization preferences in the virtual environment. Specifically, the original information of narrative progress refers to the preliminary extraction of unprocessed information about the virtual environment's storyline, plot development stage, etc., from the narrative state register. The original information of task objective refers to the preliminary identification of the user's currently active task or pending objective in the virtual environment, which has not yet been verified. The original information of user personalization preferences refers to the preliminary analysis of user interaction behavior data to obtain unrefined data on the user's preferences for specific content, style, or interaction methods. After preliminary confirmation, the original information may contain redundant, missing, or illogical data points.

[0056] The original information of the narrative progress, the currently active task objective, and the user's personalized preferences in the virtual environment is subjected to integrity checks and consistency verification to obtain the narrative progress information, the currently active task objective information, and the user's personalized preference information in the virtual environment. Integrity checks involve a comprehensive detection of missing data and structural verification of the aforementioned original information of narrative progress, task objective, and user personalized preferences. For example, it can check whether key fields are empty, whether the data format conforms to preset specifications, and whether the data records are complete. Its purpose is to ensure that all necessary information elements have been obtained, avoiding deviations in subsequent analysis due to incomplete data. Consistency verification involves logical cross-comparison and conflict detection of the original information after integrity checks. For example, it can verify whether the narrative progress matches the current scene object identifier, whether the task objective logically conflicts with the user's current location or completed events, and whether the user's personalized preferences are consistent with historical behavior data. Its purpose is to eliminate contradictions and irrationalities between data, ensuring that the confirmed narrative progress information, task objective information, and user personalized preference information are logically consistent and accurate.

[0057] Specifically, in a virtual reality role-playing game, it's necessary to confirm the user's narrative progress, current mission objectives, and personalized preferences. First, raw information about narrative progress (e.g., current chapter and completed story points), mission objectives (e.g., collecting item A and defeating enemy B), and user personalized preferences (e.g., preference for exploration and preference for combat) is extracted from the narrative state register. Then, the raw information undergoes an integrity check. For example, it checks if the current chapter field is empty, or if the mission to collect item A lacks information on the quantity of the target item. If missing information is found, attempts are made to supplement it from other related data sources, or it is marked as an pending exception. Next, consistency verification is performed. For example, if the raw narrative progress information indicates the user is in the late game, but the raw mission objective information contains a mission that only appears in the early game, this is an inconsistency. Logical judgment is applied, possibly correcting it through priority rules or context analysis, for example, marking the early mission as expired or invalid. Similarly, if the raw user personalized preference information shows the user prefers exploration, but their recent interaction data frequently shows combat, consistency verification is performed, possibly updating or correcting their preference information through weighted averaging or time decay algorithms. Through the aforementioned integrity checks and consistency verifications, refined and calibrated narrative progress information, currently active task objective information, and user personalized preference information within the virtual environment are ultimately obtained. This high-quality information will be used for subsequent coefficient analysis to more accurately determine the importance coefficients of each rendered image, thereby guiding the rendering system to make image adjustments that better match the user's current context and preferences.

[0058] This embodiment effectively avoids inaccuracies and inconsistencies that may arise from directly using raw data by introducing integrity checks and consistency verification. Specifically, integrity checks ensure that all key narrative progress, task objectives, and user preference data are in place, preventing misjudgments due to missing information. Building on this, consistency verification further logically verifies these data, eliminating contradictions and conflicts between them, thus guaranteeing that the confirmed narrative progress information, currently active task objective information, and user personalized preference information are highly reliable and accurate. This precise confirmation of key information allows for a more accurate determination of the importance coefficients for each rendered image, reflecting the user's true intentions and the actual state of the virtual environment, providing a solid data foundation for refined rendering adjustments.

[0059] In some embodiments of this application described above, the steps of determining the user's demand for smooth rendering of the virtual environment and their sensitivity to visual details based on the current interaction behavior data include: Based on the current interaction behavior data, determine each frame of image data of the user in the virtual environment within a certain period. The current interaction behavior data can be understood as all recordable information generated when the user operates in the virtual environment, such as user input commands, eye-tracking data, head posture, hand gestures, and navigation paths. It reflects the user's real-time interaction state and intention with the virtual environment. Each frame of image data of the user in the virtual environment within a certain period refers to the continuous sequence of images actually rendered and presented to the user within a specific time window of the user's interaction behavior. The image data contains the visual information that the user actually sees and experiences in the virtual environment. Determining each frame of image data can be achieved by real-time capture of the rendering pipeline output or reading from the rendering buffer.

[0060] Image analysis is performed on each frame of image data to determine the user's rendering smoothness requirements and sensitivity to image detail in the virtual environment. This step involves processing and evaluating the captured consecutive image frames to extract features related to rendering smoothness and image detail perception. For example, frame rate fluctuations, screen tearing, stuttering, texture detail, lighting effects, and geometric complexity can be analyzed. The aim is to quantify the user's actual experience of smoothness and detail from the presented visual effects. The user's rendering smoothness requirements refer to their expected level of screen update frequency and consistency in the current interactive context. For example, in fast-moving or intense combat scenes, users may have a higher demand for smoothness. The user's sensitivity to image detail refers to the degree of attention the user pays to visual details such as image texture, lighting, and model accuracy in the current context. For example, when observing specific objects or performing fine operations, users may have a higher sensitivity to detail.

[0061] This embodiment obtains the visual information flow actually experienced by the user by determining each frame of image data in the virtual environment over a period of time based on current interaction behavior data. Subsequently, by performing image analysis on each frame of image data, indicators related to the user's subjective feelings can be extracted from the objective visual presentation. This allows the determination of the user's rendering smoothness requirements and sensitivity to image detail to no longer rely solely on preset rules or simple interaction type judgments, but rather on a quantitative evaluation based on the visual feedback actually received by the user. By analyzing actual image data, the user's true needs and perceptual tendencies regarding rendering quality in different interaction scenarios can be captured more accurately.

[0062] In some embodiments of this application described above, the step of performing image analysis on each frame of image data to determine the user's requirements for the smoothness of virtual environment rendering and their sensitivity to image details includes: Image analysis is performed on each frame of image data to determine the smoothness and perceptual parameters for each frame. Specifically, this step aims to extract quantitative indicators related to rendering smoothness and the perception of image detail from the raw image data. Smoothness parameters may include, but are not limited to, frame rate (FPS), inter-frame latency, jitter rate, and screen tearing, directly reflecting the smoothness and coherence of image rendering. Perceptual parameters may include, but are not limited to, image sharpness, texture detail, color saturation, contrast, lighting effects, and anti-aliasing effects; these parameters affect the user's intuitive perception of the image's visual quality.

[0063] By using preset smoothness parameters, the smoothness parameters of each frame of image data are compared and analyzed to determine the user's requirements for the smoothness of rendering the virtual environment.

[0064] By using preset perceptual parameters, the perceptual parameters of each frame of image data are compared and analyzed to determine the user's sensitivity to image detail. Preset smoothness parameters and preset perceptual parameters are predefined benchmark values ​​or models used to measure the user's expectations or acceptable range for rendering smoothness and image detail. For example, preset smoothness parameters can be set as the minimum acceptable frame rate for different application scenarios (such as 30FPS or 60FPS), or a smoothness model built based on user historical preference data. Preset perceptual parameters can be thresholds for a series of image quality indicators, such as the minimum resolution of texture detail and the maximum tolerance for color deviation for a specific type of virtual environment. Specifically, various methods can be used to compare and analyze the smoothness parameters of each frame of image data using preset smoothness parameters. For example, the real-time frame rate can be compared with the preset minimum frame rate; if it is lower than the preset value, it indicates a smoothness problem. Alternatively, the deviation or matching degree between the real-time smoothness parameters and the preset smoothness parameters can be calculated to quantify the user's rendering smoothness requirements for the virtual environment. Similarly, by comparing and analyzing the perceptual parameters of each frame of image data using preset perceptual parameters, it is possible to assess whether the sharpness and texture details of the current image meet the user's expectations. Through comparative analysis, a quantifiable sensitivity of the user to image details can be obtained; for example, the user has a higher sensitivity to high-detail textures or a stronger perception of specific lighting effects.

[0065] This embodiment performs detailed image analysis on each frame of image data from the user in the virtual environment over a period of time, enabling a technical quantification of the user's actual needs and perceptions regarding rendering smoothness and image detail. Specifically, by extracting smoothness and perception parameters from each frame of image data, the abstract user experience is transformed into measurable technical indicators. Subsequently, by comparing and analyzing the real-time acquired parameters with preset benchmark parameters, it is possible to objectively determine whether the current rendering state meets the user's potential needs. For example, when the real-time frame rate is lower than the preset smoothness parameter, it can be identified that the user has a higher demand for smoothness; when the image detail parameters do not reach the preset perception parameter, it indicates that the user has a high sensitivity to image detail; thus providing a precise basis for subsequent rendering adjustments.

[0066] Based on any of the above embodiments, please refer to the computer image generation method. Figure 2 The present invention also provides a computer image generation system, which includes an information acquisition module 210, a data determination module 220, a coefficient judgment module 230 and an image adjustment module 240.

[0067] The information acquisition module 210 is used to acquire the user's current interactive behavior data in the virtual environment and the current rendering parameters of each image region in the virtual environment.

[0068] The data determination module 220 is used to determine the user's demand for smooth rendering of the virtual environment and sensitivity to visual details based on the current interaction behavior data.

[0069] The coefficient judgment module 230 is used to judge the image region based on the rendering smoothness requirement and perception sensitivity, and obtain the image adjustment region and image adjustment coefficient.

[0070] The image adjustment module 240 is used to adjust the current rendering parameters of each image in the image adjustment area using image adjustment coefficients, so as to obtain all images after rendering adjustment of the virtual environment.

[0071] In this embodiment, the information acquisition module 210 is responsible for collecting real-time user interaction data in the virtual environment and the current rendering parameters of each image region in the virtual environment, providing basic data for subsequent decision-making. The data determination module 220 analyzes and quantifies the user's needs for rendering smoothness and sensitivity to image details based on the interaction data. Subsequently, the coefficient judgment module 230 comprehensively considers the user's perceptual needs, performs image region judgment on the virtual environment, and identifies the image regions that need adjustment and the corresponding adjustment coefficients. Finally, the image adjustment module 240, based on the adjustment coefficients, performs fine-tuning of the rendering parameters of specific image regions to generate an optimized virtual environment image. This effectively solves the problem of unbalanced rendering resource allocation caused by the introduction of highly complex assets in existing technologies, significantly improving the smoothness and visual quality of the user's virtual reality experience.

[0072] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the present invention specification.

Claims

1. A computer image generation method characterized by, The method comprises the following steps: acquiring current interaction behavior data of a user in a virtual environment and current rendering parameters of each image area in the virtual environment; determining a rendering fluency requirement degree of the user for the virtual environment and a perception sensitivity of the user for picture details according to the current interaction behavior data; performing image area judgment on the rendering fluency requirement degree and the perception sensitivity to obtain an image adjustment area and an image adjustment coefficient; adjusting the current rendering parameters of each image in the image adjustment area by using the image adjustment coefficient to obtain all images after rendering adjustment of the virtual environment.

2. The computer image generation method of claim 1, wherein, In the step of performing image area judgment on the rendering fluency requirement degree and the perception sensitivity to obtain the image adjustment area and the image adjustment coefficient, the step of acquiring the image adjustment coefficient comprises: determining a dynamic overhead environment parameter in the virtual environment and an image adjustment initial coefficient by performing image area judgment on the rendering fluency requirement degree and the perception sensitivity; determining a dynamic overhead rendering cost according to the dynamic overhead environment parameter and a preset rendering cost library; comparing the dynamic overhead rendering cost with a preset cost threshold to obtain a rendering cost ratio; adjusting the image adjustment initial coefficient by using the rendering cost ratio to obtain the image adjustment coefficient.

3. The computer image generation method of claim 1, wherein, In the step of performing image area judgment on the rendering fluency requirement degree and the perception sensitivity to obtain the image adjustment area and the image adjustment coefficient, the step of acquiring the image adjustment area comprises: determining a rendering recognition result of a type, a semantic role and a potential interaction possibility of a rendering object contained in each image area by performing image area judgment on the rendering fluency requirement degree and the perception sensitivity; determining the image adjustment area according to the rendering recognition result.

4. The method of claim 1, wherein, The step of adjusting the current rendering parameters of each image in the image adjustment area by using the image adjustment coefficient to obtain all images after rendering adjustment of the virtual environment comprises: determining an important coefficient of each rendering image according to the current interaction behavior data; determining an interaction priority area in a user attention area based on the important coefficient of each rendering image; adjusting the current rendering parameters of each image in the image adjustment area by using the interaction priority area and the image adjustment coefficient to obtain all images after rendering adjustment of the virtual environment.

5. The method of claim 4, wherein, The step of determining the important coefficient of each rendering image according to the current interaction behavior data comprises: confirming narrative progress information, a currently activated task target information and user personalized preference information in the virtual environment according to the current interaction behavior data; performing coefficient analysis on the narrative progress information, the task target information and the user personalized preference information to determine the important coefficient of each rendering image.

6. The computer graphics generation method of claim 5, wherein, The step of confirming narrative progress information, a currently activated task target information and user personalized preference information in the virtual environment according to the current interaction behavior data comprises: confirming an event publishing state in the current interaction behavior data, the event publishing state containing an event type, an occurrence time and a scene object identifier associated with the event; performing analysis processing on the event publishing state to obtain a timestamp of the event publishing state; The virtual environment state information in the narrative state register is extracted by using the time stamp of the event publishing state and the current interaction behavior data; Data analysis is performed on the virtual environment state information of the narrative state register to confirm the narrative progress information, the currently activated task target information and the user personalized preference information in the virtual environment.

7. A computer graphics generation method according to claim 6, wherein, The step of performing data analysis on the virtual environment state information of the narrative state register to confirm the narrative progress information, the currently activated task target information and the user personalized preference information in the virtual environment includes: The step of performing data analysis on the virtual environment state information of the narrative state register to confirm the narrative progress information, the currently activated task target information and the user personalized preference information in the virtual environment includes: The step of performing data analysis on the virtual environment state information of the narrative state register to confirm the narrative progress information, the currently activated task target information and the user personalized preference information in the virtual environment includes:

8. The method of claim 1, wherein, The step of determining the rendering fluency requirement degree of the user for the virtual environment and the perception sensitivity of the user for picture details according to the current interaction behavior data includes: According to the current interaction behavior data, the user's every frame image data in the virtual environment within a period of time is determined. The step of performing image analysis on the every frame image data to determine the rendering fluency requirement degree of the user for the virtual environment and the perception sensitivity of the user for picture details includes:

9. The computer image generation method of claim 8, wherein, The step of performing image analysis on the every frame image data to determine the rendering fluency requirement degree of the user for the virtual environment and the perception sensitivity of the user for picture details includes: The step of performing image analysis on the every frame image data to determine the rendering fluency requirement degree of the user for the virtual environment and the perception sensitivity of the user for picture details includes: The step of performing image analysis on the every frame image data to determine the rendering fluency requirement degree of the user for the virtual environment and the perception sensitivity of the user for picture details includes: The system includes:

10. A computer image generation system characterized by, An information acquisition module is configured to acquire current interaction behavior data of a user in a virtual environment and current rendering parameters of each image region in the virtual environment; A data determination module is configured to determine a rendering fluency requirement degree of the user for the virtual environment and a perception sensitivity of the user for picture details according to the current interaction behavior data; A coefficient judgment module is configured to perform image region judgment on the rendering fluency requirement degree and the perception sensitivity to obtain an image adjustment region and an image adjustment coefficient; An image adjustment module is configured to adjust the current rendering parameters of each image in the image adjustment region by using the image adjustment coefficient to obtain all images after rendering adjustment of the virtual environment. ​

Images