Virtual scene construction method and system based on AIGC technology

Abstract

The invention discloses a virtual scene construction method and system based on the AI GC technology, and belongs to the technical field of virtual scene construction. The system comprises a data quality optimization module, a generative model optimization module, a physical simulation and interactivity module, a quality control and evaluation module, a multi-modal generation module, a real-time production and optimization module, a self-adaptive adjustment and personalization module, a post-processing and repairing module and a man-machine collaborative optimization module. By optimizing the generation efficiency and stability, after the problem of unstable quality is solved, the system can better meet the requirements of large-scale virtual scene generation, the stability, accuracy and innovativeness of the virtual scene construction system can be remarkably improved, the user experience is enhanced, the adaptability and diversity of the system are improved, and the method is suitable for popularization and application. Therefore, developers are helped to create more immersive and diversified virtual worlds, and more personalized virtual contents meeting requirements are brought to users.

Description

Technical Field

[0001] The present invention belongs to the technical field of virtual scene construction, and in particular relates to a virtual scene construction method and system based on AIGC technology. Background Art

[0002] With the advancement of science and technology and the popularization of technologies such as virtual reality (VR) and augmented reality (AR), users' needs for virtual scenes have become more and more diverse. AIGC technology can automatically generate virtual scenes of various styles, contents and complexity through advanced algorithms such as deep learning and generative adversarial networks to meet the needs of various industries for personalization and customization. Virtual scenes based on AIGC are not limited to static scenes, but can also include dynamic elements, interactive mechanisms and other content, further enhancing the interactivity and immersion of virtual environments. Traditional virtual scene construction relies on manual design and modeling, which is a complex and time-consuming process. Based on AIGC technology, realistic and demand-compliant virtual environments can be automatically generated, greatly improving creation efficiency and saving a lot of manpower and time costs. Especially in large-scale projects, the application of AIGC technology has significantly reduced the overall cost of virtual scene construction.

[0003] Although AIGC technology has shown strong capabilities in building virtual scenes, the quality and consistency of generated content are still problematic. Especially in complex or highly customized scenes, AI may not fully understand the detailed requirements, resulting in unnatural or inharmonious generated scenes. Such unstable generation results may affect the user's immersion and experience. In addition, many AIGC-based virtual scene construction systems have problems with physical properties (such as lighting, reflection, collision, etc.) and scene interactivity. AI may not fully consider physical laws when generating scenes, resulting in objects in the virtual world not being able to interact as naturally as in the real world. For example, objects may penetrate other objects, or the lighting and shadow effects may not match the actual situation, giving users an abrupt and unreal feeling. Summary of the invention

[0004] The technical problem to be solved by the present invention is to overcome the shortcomings of the above-mentioned prior art and provide a virtual scene construction method and system based on AIGC technology.

[0005] The technical solution adopted to solve the above technical problems is: a virtual scene construction system based on AIGC technology, which includes: a data quality optimization module, a generation model optimization module, a physical simulation and interactivity module, a quality control and evaluation module, a multimodal generation module, a real-time production and optimization module, an adaptive adjustment and personalization module, a post-processing and repair module, and a human-machine collaborative optimization module;

[0006] Data quality optimization module, used to improve the quality of input data and ensure the stability and high quality of generated content;

[0007] Generative model optimization module, used to generate model stability and detail presentation;

[0008] Physical simulation and interactivity module, used to generate virtual scenes in accordance with physical rules and interactive requirements;

[0009] Quality control and assessment module for consistent quality of generated content;

[0010] Multimodal generation module, used to generate virtual scenes that meet user needs;

[0011] Real-time production and optimization module, used to optimize the computational efficiency of the model;

[0012] Adaptive adjustment and personalization module, used to automatically adjust the generated content according to the context of user input;

[0013] The post-processing and repair module is used to post-process the generated scene and repair defects;

[0014] The human-computer collaborative optimization module is used to provide designers with interactive tools, allowing them to intervene in the generation process in real time and adjust the generated content.

[0015] The construction method of the virtual scene construction system based on AIGC technology includes the following specific steps:

[0016] S1, remove irrelevant and low-quality data through the data quality optimization module, and generate diversified data of different scenarios and elements;

[0017] S2, generate a rough framework through the generative model optimization module, and then gradually add details and enhance the physical and environmental simulation of the virtual scene;

[0018] S3, the physical simulation and interactivity module introduces weather and environmental change simulation to establish a generative model based on the actual environment;

[0019] S4, using image quality assessment algorithm to automatically evaluate the generated results and find quality problems;

[0020] S5, cross-domain generation capabilities through multimodal generation modules, such as combining virtual reality, augmented reality and sound elements;

[0021] S6. In large-scale generation tasks, the real-time production and optimization module ensures the stability and efficiency of the generation system through distributed computing and load balancing technology;

[0022] S7, Adaptive adjustment and personalization module allows users to select different scene styles and generate scenes of specific styles according to the settings;

[0023] S8, super-resolution technology in the post-processing and restoration module to improve the details and clarity of the generated images;

[0024] S9, designers and domain experts optimize and adjust the generated content by controlling the human-computer collaborative optimization module.

[0025] Furthermore, in step S1, the data quality optimization module performs unified standardization and normalization processing on different types of data.

[0026] Furthermore, the generative model optimization module in step S2 uses the generative model to capture the structure and details of the complex scene.

[0027] Furthermore, the physical simulation and interactivity module in step S3 uses a physical engine to ensure that the movement, lighting and collision of objects in the scene conform to the laws of physics.

[0028] Furthermore, the image quality assessment algorithm in S4 checks the spatial relationship and visual consistency between different elements in the scene during the generation process.

[0029] Furthermore, the multimodal generation module in S5 optimizes the model training and reasoning process by using GPU and TPU hardware acceleration technology.

[0030] Furthermore, the real-time production and optimization module in S6 automatically adjusts the content of the virtual scene through a context-aware generation method in combination with specific scene requirements and style requirements input by the user.

[0031] Furthermore, the adaptive adjustment and personalization module in S7 provides a user interactive control option, allowing the user to adjust elements in the scene in real time and obtain instant feedback.

[0032] Furthermore, the post-processing and repair module in the S8 uses AI-based post-processing technology to automatically repair geometric defects and object misalignments that occur during the generation process.

[0033] The beneficial effects of the present invention are as follows: (1) The present invention improves data quality, optimizes generation models, strengthens physical and environmental simulations, and other measures to make the quality of generated content more stable and consistent. Regardless of different scenes, styles, or environmental conditions, the system can continuously output high-quality virtual content, reduce fluctuations caused by model or data problems, and strengthen physical simulation and environmental changes. The authenticity of the virtual scene is greatly improved, and users can experience a more natural and immersive virtual world. For example, simulations such as weather changes and day and night cycles make the scene more dynamic and vivid, which provides users with a virtual experience closer to reality; (2) The present invention can significantly improve the speed and real-time performance of the generation process by accelerating model training and reasoning, adopting efficient hardware acceleration technology and distributed computing. Even in complex virtual scene generation tasks, the system can quickly respond to user needs and provide timely and high-quality generated content, ensuring that large-scale scene generation is no longer limited by performance bottlenecks. In addition, through multi-level quality control and consistency checks, unreasonable elements in the generated content (such as unreasonable object layout, physical simulation distortion, etc.) will be effectively identified and corrected. This avoids abrupt and uncoordinated phenomena in virtual scenes, making the generated content more reasonable and in line with actual requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a framework diagram of a virtual scene construction system based on AIGC technology of the present invention;

[0035] Figure 2 It is a flow chart of the virtual scene construction method based on AIGC technology of the present invention. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0037] like Figure 1-Figure 2 As shown, the virtual scene construction system based on AIGC technology of this embodiment includes: a data quality optimization module, a generation model optimization module, a physical simulation and interactivity module, a quality control and evaluation module, a multimodal generation module, a real-time production and optimization module, an adaptive adjustment and personalization module, a post-processing and repair module, and a human-machine collaborative optimization module;

[0038] Data quality optimization module, used to improve the quality of input data and ensure the stability and high quality of generated content;

[0039] Generative model optimization module, used to generate model stability and detail presentation;

[0040] Physical simulation and interactivity module, used to generate virtual scenes in accordance with physical rules and interactive requirements;

[0041] Quality control and assessment module for consistent quality of generated content;

[0042] Multimodal generation module, used to generate virtual scenes that meet user needs;

[0043] Real-time production and optimization module, used to optimize the computational efficiency of the model;

[0044] Adaptive adjustment and personalization module, used to automatically adjust the generated content according to the context of user input;

[0045] The post-processing and repair module is used to post-process the generated scene and repair defects;

[0046] The human-computer collaborative optimization module is used to provide designers with interactive tools, allowing them to intervene in the generation process in real time and adjust the generated content.

[0047] The construction method of the virtual scene construction system based on AIGC technology includes the following specific steps:

[0048] S1. Remove irrelevant and low-quality data through the data quality optimization module, generate diversified data for different scenarios and elements, and perform unified standardization and normalization on different types of data;

[0049] S2, through the generative model optimization module, a rough framework is generated, and then details are gradually added to enhance the physical and environmental simulation of the virtual scene. At the same time, the generative model is used to capture the structure and details of the complex scene; and, the generative model is used to capture the structure and details of the complex scene;

[0050] S3, physical simulation and interactivity module introduces weather and environmental change simulation to establish a generative model based on the actual environment. At the same time, the physical engine is used to ensure that the movement, lighting and collision of objects in the scene conform to the laws of physics;

[0051] S4, automatically evaluate the generated results using image quality assessment algorithms to identify quality issues and check the spatial relationship and visual consistency between different elements in the scene;

[0052] S5. Through the cross-domain generation capabilities of the multimodal generation module, such as combining virtual reality, augmented reality and sound elements, and using GPU and TPU hardware acceleration technology to optimize the model training and reasoning process;

[0053] S6. In large-scale generation tasks, the real-time production and optimization module ensures the stability and efficiency of the generation system through distributed computing and load balancing technology, and automatically adjusts the content of the virtual scene through context-aware generation methods based on the specific scene requirements and style requirements input by users;

[0054] S7, Adaptive adjustment and personalization module allows users to select different scene styles and generate scenes of specific styles according to the settings, providing user interactive control options, allowing users to adjust elements in the scene in real time and obtain instant feedback;

[0055] S8, super-resolution technology in the post-processing and restoration module improves the details and clarity of the generated images, and uses AI-based post-processing technology to automatically repair geometric defects and object dislocations that occur during the generation process;

[0056] S9, designers and domain experts optimize and adjust the generated content by controlling the human-computer collaborative optimization module.

[0057] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention.

Claims

1. A virtual scene construction system based on AIGC technology, characterized by: The system includes: a data quality optimization module, a generation model optimization module, a physical simulation and interactivity module, a quality control and evaluation module, a multimodal generation module, a real-time production and optimization module, an adaptive adjustment and personalization module, a post-processing and repair module, and a human-machine collaborative optimization module; Data quality optimization module, used to improve the quality of input data and ensure the stability and high quality of generated content; Generative model optimization module, used to generate model stability and detail presentation; Physical simulation and interactivity module, used to generate virtual scenes in accordance with physical rules and interactive requirements; Quality control and assessment module for consistent quality of generated content; Multimodal generation module, used to generate virtual scenes that meet user needs; Real-time production and optimization module, used to optimize the computational efficiency of the model; Adaptive adjustment and personalization module, used to automatically adjust the generated content according to the context of user input; The post-processing and repair module is used to post-process the generated scene and repair defects; The human-computer collaborative optimization module is used to provide designers with interactive tools, allowing them to intervene in the generation process in real time and adjust the generated content.

2. The method for constructing a virtual scene construction system based on AIGC technology according to claim 1, characterized in that: The specific steps include: S1, remove irrelevant and low-quality data through the data quality optimization module, and generate diversified data of different scenarios and elements; S2, generate a rough framework through the generative model optimization module, and then gradually add details and enhance the physical and environmental simulation of the virtual scene; S3, the physical simulation and interactivity module introduces weather and environmental change simulation to establish a generative model based on the actual environment; S4, using image quality assessment algorithm to automatically evaluate the generated results and find quality problems; S5, cross-domain generation capabilities through multimodal generation modules, such as combining virtual reality, augmented reality and sound elements; S6. In large-scale generation tasks, the real-time production and optimization module ensures the stability and efficiency of the generation system through distributed computing and load balancing technology; S7, Adaptive adjustment and personalization module allows users to select different scene styles and generate scenes of specific styles according to the settings; S8, super-resolution technology in the post-processing and restoration module to improve the details and clarity of the generated images; S9, designers and domain experts optimize and adjust the generated content by controlling the human-computer collaborative optimization module.

3. The method for constructing a virtual scene construction system based on AIGC technology according to claim 2, characterized in that: In step S1, the data quality optimization module performs unified standardization and normalization processing on different types of data.

4. The method for constructing a virtual scene construction system based on AIGC technology according to claim 3 is characterized in that: The generative model optimization module in step S2 uses the generative model to capture the structure and details of the complex scene.

5. The method for constructing a virtual scene construction system based on AIGC technology according to claim 4, characterized in that: The physical simulation and interactivity module in step S3 uses a physical engine to ensure that the movement, lighting and collision of objects in the scene conform to the laws of physics.

6. The method for constructing a virtual scene construction system based on AIGC technology according to claim 5, characterized in that: The image quality assessment algorithm in S4 checks the spatial relationship and visual consistency between different elements in the scene during the generation process.

7. The method for constructing a virtual scene construction system based on AIGC technology according to claim 6, characterized in that: The multimodal generation module in the S5 optimizes the model training and reasoning processes by using GPU and TPU hardware acceleration technology.

8. The method for constructing a virtual scene construction system based on AIGC technology according to claim 7, characterized in that: The real-time production and optimization module in S6 automatically adjusts the content of the virtual scene through a context-aware generation method in combination with the specific scene requirements and style requirements input by the user.

9. The method for constructing a virtual scene construction system based on AIGC technology according to claim 8, characterized in that: The adaptive adjustment and personalization module in the S7 provides user interactive control options, allowing users to adjust elements in the scene in real time and obtain instant feedback.

10. The method for constructing a virtual scene construction system based on AIGC technology according to claim 9, characterized in that: The post-processing and repair module in the S8 uses AI-based post-processing technology to automatically repair geometric defects and object misalignments that occur during the generation process.

Images