Digital human generation method and device, electronic equipment and computer readable storage medium

Abstract

The invention relates to the technical field of 3D, and provides a digital human generation method and device, electronic equipment and a computer readable storage medium, and the method comprises the steps: obtaining a multi-view figure image; performing calculation processing on the multi-view figure image based on a preset motion structure recovery algorithm to obtain an initialized point cloud; performing three-dimensional Gaussian distribution rendering processing on the initialized point cloud to obtain a three-dimensional Gaussian body; performing classification processing on the three-dimensional Gaussian body to obtain a plurality of three-dimensional Gaussian kernels; performing integration processing on a preset physical parameter set and the plurality of three-dimensional Gaussian kernels to obtain a plurality of discrete particle clouds; and solving each discrete particle cloud based on a preset substance point algorithm to obtain a three-dimensional Gaussian digital human. According to the technical scheme, the face of the digital human can be finer, the action is more natural, the clothes have high non-rigid expression ability, and better use experience is brought to a user.

Description

Technical Field

[0001] The embodiments of the present application relate to but are not limited to the field of 3D technology, and in particular to a method, device, electronic device and computer-readable storage medium for generating a digital human. Background Art

[0002] At present, with the continuous development of social economy and the continuous advancement of science and technology, people's living standards are constantly improving; the financial industry is also constantly developing in the direction of intelligence, among which virtual digital people are increasingly used in the financial industry, such as virtual anchors, virtual customer service or film and television characters; However, the current virtual digital people's movements are not natural enough, facial modeling is not refined enough, and the non-rigid feature expression ability of clothing is insufficient, which brings a bad user experience to users. Summary of the invention

[0003] The following is a summary of the subject matter described in detail herein. This summary is not intended to limit the scope of the claims.

[0004] In order to solve the problems mentioned in the above background technology, the embodiments of the present application provide a digital human generation method, device, electronic device and computer-readable storage medium, which can make the digital human's face more refined, the movements more natural, improve the non-rigid expression ability of clothing, and bring users a better user experience.

[0005] In a first aspect, an embodiment of the present application provides a method for generating a digital human, comprising:

[0006] Acquire multi-view character images;

[0007] The multi-view character image is processed by calculation based on a preset motion structure recovery algorithm to obtain an initialization point cloud;

[0008] Performing three-dimensional Gaussian distribution rendering processing on the initialization point cloud to obtain a three-dimensional Gaussian volume;

[0009] Classifying the three-dimensional Gaussian body to obtain a plurality of three-dimensional Gaussian kernels;

[0010] Integrating a preset physical parameter set with the plurality of three-dimensional Gaussian kernels to obtain a plurality of discrete particle clouds;

[0011] Each discrete particle cloud is solved based on a preset material point algorithm to obtain a three-dimensional Gaussian digital human.

[0012] According to some embodiments of the present application, the step of calculating and processing the multi-view character image based on a preset motion structure recovery algorithm to obtain an initialization point cloud includes:

[0013] Performing feature point detection on the multi-view character image to obtain image feature points;

[0014] Performing feature description on the image feature points to obtain image feature description information;

[0015] Performing feature matching processing on the image feature points according to the image feature description information to obtain matching feature points;

[0016] determining motion estimation information according to the matching feature points;

[0017] Perform triangulation based on the matching feature points and the motion estimation information to obtain three-dimensional coordinate information;

[0018] The initialization point cloud is constructed according to the three-dimensional coordinate information.

[0019] According to some embodiments of the present application, performing three-dimensional Gaussian distribution rendering processing on the initialization point cloud to obtain a three-dimensional Gaussian volume includes:

[0020] Preprocessing the initialization point cloud to obtain a preprocessed point cloud;

[0021] Extracting initial feature points from the preprocessed point cloud;

[0022] Constructing an initial Gaussian distribution model according to the initial feature points, wherein the initial Gaussian distribution model carries mean information and covariance information;

[0023] Performing surface reconstruction on the initial Gaussian distribution model according to the mean information and the covariance information to obtain a surface reconstructed Gaussian distribution model;

[0024] Performing surface smoothing processing on the surface reconstructed Gaussian distribution model to obtain a smoothed Gaussian distribution model;

[0025] The smoothed Gaussian distribution model is rendered to obtain the three-dimensional Gaussian volume.

[0026] According to some embodiments of the present application, the classifying process of the three-dimensional Gaussian volume to obtain a plurality of three-dimensional Gaussian kernels includes:

[0027] Determine segmentation intensity information, segmentation gradient information and shape feature information according to the three-dimensional Gaussian volume;

[0028] The three-dimensional Gaussian volume is segmented according to the segmentation intensity information, the segmentation gradient information and the shape feature information to obtain a plurality of the three-dimensional Gaussian kernels.

[0029] According to some embodiments of the present application, each of the three-dimensional Gaussian kernels carries classification information, and the integration of the preset physical parameter set with the multiple three-dimensional Gaussian kernels to obtain multiple discrete particle clouds includes:

[0030] determining a target physical parameter from the physical parameter set according to the classification information carried by the three-dimensional Gaussian kernel;

[0031] The target physical parameter and the corresponding three-dimensional Gaussian kernel are fused to obtain the discrete particle cloud.

[0032] According to some embodiments of the present application, solving each discrete particle cloud based on a preset material point algorithm to obtain a three-dimensional Gaussian digital human includes:

[0033] Obtaining state information of the discrete particle cloud and constructing a background grid;

[0034] Interpolating and mapping the state information onto the grid nodes of the background grid;

[0035] Performing physical quantity calculation processing on the grid nodes of the background grid to obtain force values ​​and momentum;

[0036] updating the physical information of the background grid according to the force value and the momentum;

[0037] The state information of the discrete particle cloud is updated according to the updated physical information to obtain the three-dimensional Gaussian digital human.

[0038] According to some embodiments of the present application, determining a target physical parameter from the physical parameter set according to the classification information carried by the three-dimensional Gaussian kernel includes:

[0039] Determining a parameter selection type according to the classification information;

[0040] The target physical parameter is selected from the physical parameter set according to the parameter selection type.

[0041] In a second aspect, the embodiment of the present application further provides a digital human generation device, including:

[0042] A first processing module, used for acquiring multi-view character images;

[0043] A second processing module is used to calculate and process the multi-view character image based on a preset motion structure recovery algorithm to obtain an initialization point cloud;

[0044] A third processing module is used to perform three-dimensional Gaussian distribution rendering processing on the initialization point cloud to obtain a three-dimensional Gaussian volume;

[0045] A fourth processing module, used for classifying the three-dimensional Gaussian volume to obtain a plurality of three-dimensional Gaussian kernels;

[0046] A fifth processing module, used for integrating a preset physical parameter set with the plurality of three-dimensional Gaussian kernels to obtain a plurality of discrete particle clouds;

[0047] The sixth processing module is used to solve each discrete particle cloud based on a preset material point algorithm to obtain a three-dimensional Gaussian digital human.

[0048] In a third aspect, an embodiment of the present application further provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the digital human generation method as described in the first aspect above is implemented.

[0049] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are used to execute the digital human generation method as described in the first aspect above.

[0050] The digital human generation method according to the embodiment provided in the present application has at least the following beneficial effects: in the process of digital human generation, firstly, a multi-view character image is obtained; then, the multi-view character image is calculated and processed based on a preset motion structure recovery algorithm to obtain an initialization point cloud; then, the initialization point cloud is subjected to three-dimensional Gaussian distribution rendering processing to obtain a three-dimensional Gaussian body; then, the three-dimensional Gaussian body is classified to obtain multiple three-dimensional Gaussian kernels; then, a preset physical parameter set is integrated with multiple three-dimensional Gaussian kernels to obtain multiple discrete particle clouds; finally, each discrete particle cloud is solved based on a preset material point algorithm to obtain a three-dimensional Gaussian digital human; through the above technical solution, a preset physical parameter set is integrated with multiple three-dimensional Gaussian kernels to obtain multiple discrete particle clouds, so that the movement and expression of the generated digital human can be closer to nature, the face of the digital human can be more refined, the movement is more natural, the non-rigid expression ability of the clothing is improved, and a better user experience is brought to the user. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The accompanying drawings are used to provide further understanding of the technical solution of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the technical solution of the present application and do not constitute a limitation on the technical solution of the present application.

[0052] Figure 1 is a flow chart of a method for generating a digital human provided by an embodiment of the present application;

[0053] Figure 2is a specific flow chart of step S200 provided in one embodiment of the present application;

[0054] Figure 3 is a specific flow chart of step S300 provided in one embodiment of the present application;

[0055] Figure 4 is a specific flow chart of step S400 provided in one embodiment of the present application;

[0056] Figure 5 is a specific flow chart of step S500 provided in one embodiment of the present application;

[0057] Figure 6 is a specific flow chart of step S600 provided in one embodiment of the present application;

[0058] Figure 7 is a specific flow chart of step S510 provided by an embodiment of the present application;

[0059] Figure 8 is a schematic diagram of a digital human generation device provided by an embodiment of the present application;

[0060] Fig. 9 It is a schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application 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 application and are not used to limit the present application.

[0062] It should be noted that although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification and claims and the above drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0063] It should be noted that, unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0064] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0065] AI is a new technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence. Artificial intelligence is a branch of computer science. Artificial intelligence attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Research in this field includes robots, language recognition, image recognition, natural language processing and expert systems. Artificial intelligence can simulate the information process of human consciousness and thinking. Artificial intelligence is also a theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0066] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0067] Artificial intelligence is AI. AI is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0068] The servers involved in artificial intelligence technology can be independent servers or cloud servers that provide basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), as well as big data and artificial intelligence platforms.

[0069] The present application provides a method, device, electronic device and computer-readable storage medium for generating a digital human. In the process of generating a digital human, firstly, a multi-view character image is obtained; then, the multi-view character image is calculated and processed based on a preset motion structure recovery algorithm to obtain an initialization point cloud; then, the initialization point cloud is rendered with a three-dimensional Gaussian distribution to obtain a three-dimensional Gaussian body; then, the three-dimensional Gaussian body is classified to obtain a plurality of three-dimensional Gaussian kernels; then, a preset physical parameter set is integrated with the plurality of three-dimensional Gaussian kernels to obtain a plurality of discrete particle clouds; finally, each discrete particle cloud is solved based on a preset material point algorithm to obtain a three-dimensional Gaussian digital human; through the above technical solution, a preset physical parameter set is integrated with the plurality of three-dimensional Gaussian kernels to obtain a plurality of discrete particle clouds, so that the movement and expression of the generated digital human can be closer to nature, the digital human can be more refined, the movement can be more natural, and a better user experience can be brought to the user.

[0070] The digital human generation method provided in the embodiment of the present application relates to the field of 3D technology. The digital human generation method provided in the embodiment of the present application can be applied to a terminal, can also be applied to a server side, and can also be software running in a terminal or a server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or a server cluster or a distributed system composed of multiple physical servers, and can also be configured as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the digital human generation method, etc., but is not limited to the above forms.

[0071] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0072] It should be noted that in each specific implementation of the present application, when it comes to the need to perform relevant processing based on data related to user identity or characteristics such as user information, user behavior data, user historical data, and user location information, the user's permission or consent will be obtained first, and the collection, use, and processing of these data will comply with relevant laws, regulations, and standards. In addition, when the embodiment of the present application needs to obtain the user's sensitive personal information, the user's separate permission or consent will be obtained through a pop-up window or by jumping to a confirmation page. After clearly obtaining the user's separate permission or consent, the necessary user-related data for the normal operation of the embodiment of the present application will be obtained.

[0073] The embodiments of the present application are further described below in conjunction with the accompanying drawings.

[0074] like Figure 1 As shown, Figure 1 It is a flowchart of a digital human generation method provided by an embodiment of the present application. The digital human generation method includes but is not limited to steps S100 to S600.

[0075] Step S100, obtaining multi-view character images;

[0076] Step S200, calculating and processing the multi-view character image based on a preset motion structure recovery algorithm to obtain an initialization point cloud;

[0077] Step S300, performing three-dimensional Gaussian distribution rendering processing on the initialization point cloud to obtain a three-dimensional Gaussian volume;

[0078] Step S400, classifying the three-dimensional Gaussian volume to obtain a plurality of three-dimensional Gaussian kernels;

[0079] Step S500, integrating a preset physical parameter set with a plurality of three-dimensional Gaussian kernels to obtain a plurality of discrete particle clouds;

[0080] Step S600, solving each discrete particle cloud based on a preset material point algorithm to obtain a three-dimensional Gaussian digital human.

[0081] It should be noted that, in the process of generating a digital human, firstly, a multi-view character image is obtained; then, the multi-view character image is calculated and processed based on a preset motion structure recovery algorithm to obtain an initialization point cloud; then, the initialization point cloud is subjected to three-dimensional Gaussian distribution rendering processing to obtain a three-dimensional Gaussian body; then, the three-dimensional Gaussian body is classified to obtain multiple three-dimensional Gaussian kernels; then, the preset physical parameter set is integrated with the multiple three-dimensional Gaussian kernels to obtain multiple discrete particle clouds; finally, each discrete particle cloud is solved based on a preset material point algorithm to obtain a three-dimensional Gaussian digital human; through the above technical solution, the preset physical parameter set is integrated with the multiple three-dimensional Gaussian kernels to obtain multiple discrete particle clouds, so that the movement and expression of the generated digital human can be closer to nature, the face of the digital human can be more refined, the movement can be more natural, the non-rigid expression ability of the clothing can be improved, and a better user experience can be brought to the user.

[0082] It is worth noting that the multi-view character image in the embodiment of the present application is an image obtained by shooting the target person from multiple angles, and the permission of the target person is obtained in advance during the shooting process, and the user's permission or consent will be obtained before obtaining the multi-view character image, and the collection, use and processing of these data will comply with relevant laws, regulations and standards. In addition, when the embodiment of the present application needs to obtain the user's sensitive personal information, the user's separate permission or consent will be obtained through a pop-up window or jump to a confirmation page. After clearly obtaining the user's separate permission or consent, the necessary user-related data for the normal operation of the embodiment of the present application is obtained. Exemplarily, in order to complete the construction of a digital person to explain the relevant knowledge of insurance to customers, it is necessary to first collect and process the multi-view character images of the target person, so as to make prerequisite preparations for the subsequent construction of the digital person.

[0083] It is worth noting that the initialization point cloud can be obtained by calculating and processing the multi-view character image based on the preset motion structure recovery algorithm. After the initialization point cloud is obtained, the premise is prepared for the subsequent three-dimensional Gaussian volume construction. Point cloud is a data structure composed of a large number of three-dimensional coordinate points, each of which contains spatial position and possible attributes such as color and intensity. These point cloud data can come from various sensors, such as lidar or depth camera, and are widely used in autonomous driving, robot navigation, three-dimensional reconstruction and other fields. Point cloud processing technology mainly includes point cloud filtering, key point extraction, feature and feature description, point cloud registration, point cloud segmentation and classification, target recognition and retrieval, change detection, three-dimensional reconstruction and point cloud data management. The challenges faced by point cloud technology include large data volume, data quality, real-time processing, etc. For example, point cloud data is usually large in volume and requires effective storage and processing methods. In addition, point cloud data may be affected by noise and incompleteness, requiring robust preprocessing algorithms. In real-time applications, such as autonomous driving, point cloud processing algorithms need to be fast and accurate. The application scenarios of point cloud technology are very wide, including geographic information systems, robot navigation, industrial detection, cultural heritage protection, virtual reality and games, medical imaging, etc. In these applications, point cloud technology can provide accurate three-dimensional spatial information to help realize the automation and intelligence of complex tasks. With the development of technology, point cloud processing and recognition technology still has a lot of room for improvement in processing speed, robustness and expansion of application areas. Future research may focus on developing more efficient algorithms, improving the robustness and stability of algorithms, and applying point cloud technology to more fields.

[0084] It is worth noting that 3D Gaussian is a technique for 3D scene representation and rendering. It describes the geometric and radiation properties of the scene by distributing Gaussian functions in 3D space. Each Gaussian function defines a distribution with a specific mean and covariance matrix, and these parameters can describe a local area in the scene. 3D Gaussian can be used for a variety of applications, including but not limited to: 3D reconstruction: By applying 3D Gaussian to the representation of the scene, high-quality 3D reconstruction can be performed, including reconstruction of dynamic scenes and 3D reconstruction of challenging inputs. 3D editing: 3D Gaussian provides an intuitive way to edit 3D models, including geometry editing, appearance editing, and physical simulation. Rendering: 3D Gaussian can be used for efficient rendering, which is achieved by rasterizing the Gaussian ellipsoid into an image. Downstream applications: including segmentation and understanding, geometry reconstruction, digital humans, 3D / 4D generation, etc.

[0085] It is worth noting that discrete particle clouds usually refer to a series of discontinuous collections of particles distributed in three-dimensional space. In the field of computational science and engineering, discrete particle clouds may refer to particle methods used to simulate complex systems, such as particle swarm optimization and ant colony optimization. These algorithms solve optimization problems such as task scheduling and path planning by simulating the behavior of biological groups in nature. In these applications, the concept of discrete particle clouds is used to represent the space of possible solutions and to search for the optimal solution through iteration. In materials science and nanotechnology, discrete particle clouds may involve a collection of granular materials, whose interactions and organizational structures have a decisive influence on the properties of the materials.

[0086] It is worth noting that in the process of generating digital humans, multiple discrete particle clouds can be obtained by integrating the preset physical parameter set and multiple three-dimensional Gaussian kernels; according to the preset physical parameter set, the movements of the subsequently generated digital humans can be more natural, the faces of the digital humans can be more refined, the non-rigid expression ability of the clothes can be improved, the clothes can be made to flutter, and a better user experience can be brought to users.

[0087] It is worth noting that numerical simulation methods are mainly divided into two categories: Lagrangian method and Euler method. In the Lagrangian method, the computational grid deforms with the material, which is convenient for tracking material interfaces and introducing material models related to deformation history. However, when it comes to large deformation problems, numerical solutions may be difficult due to grid distortion. In the Euler method, the computational grid is fixed in space, which avoids grid distortion problems, but it is not easy to track material interfaces, and nonlinear convection terms may cause difficulties in numerical solutions. The material point method is developed from the particle grid method, which uses Lagrangian and Euler dual descriptions to discretize objects into particles moving in a spatial grid. Each particle carries all material information, such as mass, velocity, strain, and stress, which is convenient for tracking material interfaces and introducing material models related to deformation history. The material point method is suitable for analyzing large deformation and flow problems, avoiding grid distortion problems, and is particularly good in the numerical analysis of hypervelocity collisions and explosions. Compared with similar meshless methods, the material point method has higher computational efficiency, especially in large deformation and high strain rate problems.

[0088] In some embodiments, Figure 2 As shown, the above step S200 may include but is not limited to steps S210 to S260.

[0089] Step S210, performing feature point detection on the multi-view character image to obtain image feature points;

[0090] Step S220, performing feature description on the image feature points to obtain image feature description information;

[0091] Step S230, performing feature matching processing on the image feature points according to the image feature description information to obtain matching feature points;

[0092] Step S240, determining motion estimation information according to the matching feature points;

[0093] Step S250, performing triangulation based on the matching feature points and the motion estimation information to obtain three-dimensional coordinate information;

[0094] Step S260, constructing an initialization point cloud according to the three-dimensional coordinate information.

[0095] It should be noted that, in the process of calculating and processing the multi-view character images based on the preset motion structure recovery algorithm, first, feature point detection is performed on the multi-view character images to obtain image feature points; then, feature description is performed on the image feature points to obtain image feature description information; then, feature matching processing is performed on the image feature points according to the image feature description information to obtain matching feature points; then, motion estimation information is determined based on the matching feature points; then, triangulation processing is performed based on the matching feature points and the motion estimation information to obtain three-dimensional coordinate information; finally, an initialization point cloud can be constructed based on the three-dimensional coordinate information; through the above technical solution, the motion structure recovery algorithm is used to convert the multi-view character images into initialization point clouds, which makes the prerequisite preparation for the subsequent construction of three-dimensional Gaussian digital humans.

[0096] It is worth noting that the motion structure recovery algorithm is a computer vision technology that estimates the 3D structure of the scene and the motion of the camera by analyzing a series of images taken from different perspectives. This technology has a wide range of applications in close-range 3D reconstruction, robot navigation, intelligent positioning, augmented reality and other fields. The strategies of the motion structure recovery algorithm can be divided into incremental, global and hierarchical. The incremental motion structure recovery algorithm gradually adds images to the reconstruction, and each addition performs feature matching and triangulation; the global motion structure recovery algorithm attempts to process all images at once; the hierarchical motion structure recovery algorithm first groups the images, then reconstructs them within the group, and finally performs global optimization. In practical applications, the challenges faced by the motion structure recovery algorithm include the accuracy of feature point matching, camera calibration, conversion between sparse and dense reconstruction, and optimization and post-processing. In order to improve the accuracy and efficiency of the motion structure recovery algorithm, researchers have proposed a variety of improvement methods, such as feature matching algorithms based on scene geometric structure constraints, rotation estimation methods based on Bayesian inference networks, and random maximum spanning tree algorithms to filter outliers.

[0097] In some embodiments, Figure 3 As shown, the above step S300 may include but is not limited to steps S310 to S360.

[0098] Step S310, preprocessing the initialization point cloud to obtain a preprocessed point cloud;

[0099] Step S320, extracting initial feature points from the preprocessed point cloud;

[0100] Step S330, constructing an initial Gaussian distribution model according to the initial feature points, wherein the initial Gaussian distribution model carries mean information and covariance information;

[0101] Step S340, performing surface reconstruction on the initial Gaussian distribution model according to the mean information and the covariance information to obtain a surface reconstructed Gaussian distribution model;

[0102] Step S350, performing surface smoothing processing on the surface reconstructed Gaussian distribution model to obtain a smoothed Gaussian distribution model;

[0103] Step S360: Rendering the smoothed Gaussian distribution model to obtain a three-dimensional Gaussian volume.

[0104] It should be noted that, in the process of performing three-dimensional Gaussian distribution rendering processing on the initialization point cloud to obtain a three-dimensional Gaussian body, the initialization point cloud is first preprocessed to obtain a preprocessed point cloud; then the initial feature points are extracted from the preprocessed point cloud; then the initial Gaussian distribution model is constructed based on the initial feature points, wherein the initial Gaussian distribution model carries mean information and covariance information; then the initial Gaussian distribution model is surface reconstructed based on the mean information and covariance information to obtain a surface reconstructed Gaussian distribution model; then the surface reconstructed Gaussian distribution model is surface smoothed to obtain a smoothed Gaussian distribution model; finally, the smoothed Gaussian distribution model is rendered to obtain the corresponding three-dimensional Gaussian body; through the above technical solution, after the three-dimensional Gaussian body is obtained by rendering, the three-dimensional Gaussian body can be classified to obtain multiple three-dimensional Gaussian kernels, which makes the prerequisite preparation for the subsequent construction of digital humans.

[0105] It is worth noting that the smoothed Gaussian distribution model usually refers to the model of the Gaussian distribution (also known as the normal distribution) used in machine learning and statistics, which plays a central role in various algorithms and theories. The probability density function of the Gaussian distribution is symmetric and centered at the mean, and most data fall within one standard deviation of the mean. This distribution is widely used in machine learning, including but not limited to: Linear regression and logistic regression: In these algorithms, it is usually assumed that the observed data is generated by a Gaussian distribution, which simplifies the model and allows efficient parameter estimation. Bayesian inference: In Bayesian machine learning, the Gaussian distribution is often used as a prior distribution for model parameters. Clustering: Gaussian mixture models can simulate complex data distributions and are often used in image segmentation and data compression. Anomaly detection: Gaussian distribution is often used in anomaly detection algorithms, where the goal is to identify rare events or outliers in the data. Principal component analysis: It finds the directions of maximum variance in the data, which correspond to the principal components. Kernel methods: Gaussian kernels are used in kernelized machine learning algorithms such as support vector machines and Gaussian processes to define the similarity between data points.

[0106] In some embodiments, Figure 4 As shown, the above step S400 may include but is not limited to step S410 to step S420.

[0107] Step S410, determining segmentation intensity information, segmentation gradient information and shape feature information according to the three-dimensional Gaussian volume;

[0108] Step S420 , segmenting the three-dimensional Gaussian volume according to the segmentation intensity information, the segmentation gradient information and the shape feature information to obtain a plurality of three-dimensional Gaussian kernels.

[0109] It should be noted that in the process of classifying the three-dimensional Gaussian body to obtain multiple three-dimensional Gaussian kernels, the segmentation intensity information, segmentation gradient information and shape feature information are first determined based on the three-dimensional Gaussian body; then, based on the segmentation intensity information, segmentation gradient information and shape feature information, the three-dimensional Gaussian body can be segmented to obtain multiple three-dimensional Gaussian kernels, which is a prerequisite for the subsequent construction of digital humans.

[0110] It is worth noting that the 3D Gaussian kernel is a Gaussian distribution in 3D space, which has a wide range of applications in image processing, signal processing, and machine learning. In 3D space, the Gaussian kernel can be used to smooth data, perform blurring, or act as a filter; its mathematical expression usually contains a location parameter (mean) and a covariance matrix, which describes the extension and shape of the Gaussian distribution in all directions. When the Gaussian kernel is convolved with an image, the image blurring effect can be achieved. In the frequency domain, the convolution of the Gaussian kernel is equivalent to multiplying by the coefficients of the Fourier transform, which makes the Gaussian smoothing operation more efficient in the frequency domain. In the field of 3D reconstruction and rendering, 3D Gaussian splatting is a technique that uses a Gaussian ellipsoid to represent the scene. This method achieves efficient rendering by rasterizing the Gaussian ellipsoid into an image, and can be used for tasks such as dynamic reconstruction, geometry editing, and physical simulation. In practical applications, the 3D Gaussian kernel can be adjusted in shape and size through various parameterization methods to adapt to different processing requirements. For example, the extension of the Gaussian kernel in different directions can be controlled by adjusting the values ​​in the covariance matrix to achieve anisotropic Gaussian distribution. In summary, the 3D Gaussian kernel is a very useful tool when dealing with 3D data, as it provides a flexible way to describe and manipulate distributions in space.

[0111] In some embodiments, Figure 5 As shown, each three-dimensional Gaussian kernel carries classification information, and the above step S500 may include but is not limited to steps S510 to S520.

[0112] Step S510, determining target physical parameters from a physical parameter set according to classification information carried by the three-dimensional Gaussian kernel;

[0113] Step S520: fusing the target physical parameters and the corresponding three-dimensional Gaussian kernel to obtain a discrete particle cloud.

[0114] It should be noted that in the process of integrating the preset physical parameter set with multiple three-dimensional Gaussian kernels, the target physical parameters are first determined from the physical parameter set based on the classification information carried by the three-dimensional Gaussian kernel; then the target physical parameters and the corresponding three-dimensional Gaussian kernel are fused to obtain a discrete particle cloud; by fusing the physical parameter set into the three-dimensional Gaussian kernel, the simulation effect can be closer to the real physical world, so that the facial expressions of the constructed digital people are more natural, and the movements can be closer to nature, which brings a better user experience to users.

[0115] For example, in the process of constructing a three-dimensional digital human to explain relevant knowledge of financial products to users, the preset physical parameter set can be fused with multiple three-dimensional Gaussian kernels to make the movements of the constructed three-dimensional digital human more natural, the clothes have better non-rigid expression capabilities, and the clothes flutter.

[0116] In some embodiments, Figure 6 As shown, the above step S600 may include but is not limited to steps S610 to S650.

[0117] Step S610, obtaining state information of discrete particle cloud and constructing background grid;

[0118] Step S620, interpolating and mapping the state information to the grid nodes of the background grid;

[0119] Step S630, performing physical quantity calculation processing on the grid nodes of the background grid to obtain force values ​​and momentum;

[0120] Step S640, updating the physical information of the background grid according to the force value and momentum;

[0121] Step S650, updating the state information of the discrete particle cloud according to the updated physical information to obtain a three-dimensional Gaussian digital human.

[0122] It should be noted that, in the process of solving each discrete particle cloud based on the preset material point algorithm to obtain a three-dimensional Gaussian digital human, first the state information of the discrete particle cloud is obtained and the background grid is constructed; then the state information is interpolated and mapped to the grid nodes of the background grid; then the physical quantity calculation processing of the grid nodes of the background grid can obtain the force value and momentum; then the physical information of the background grid is updated according to the force value and momentum; finally, the state information of the discrete particle cloud is updated according to the updated physical information to obtain the three-dimensional Gaussian digital human; through the above technical solution, the movements of the three-dimensional Gaussian digital human can be more natural, and the facial expressions can be more natural, thereby bringing a good user experience.

[0123] It is worth noting that, through the above technical solution, giving physical properties to the 3D Gaussian can make the 3D Gaussian digital human have better performance for non-rigid structures, effectively improving the problems of human body movements and clothing stiffness; the movement of the 3D Gaussian digital human conforms to the laws of physics, making the model have a more natural visual effect, and can simulate the movement of the 3D Gaussian digital human more realistically, including muscle deformation, joint movement, clothing fluttering, etc., making the 3D Gaussian digital human more natural during simulation. Segmenting and classifying the 3D Gaussian and adaptively giving different materials different physical parameters can achieve higher simulation quality and accuracy.

[0124] In some embodiments, Figure 7 As shown, the above step S510 may include but is not limited to steps S511 to S512.

[0125] Step S511, determining the parameter selection type according to the classification information;

[0126] Step S512: selecting a target physical parameter from a physical parameter set according to the parameter selection type.

[0127] It should be noted that, in the process of determining the target physical parameters from the physical parameter set according to the classification information carried by the three-dimensional Gaussian kernel, the parameter selection type is first determined according to the classification information; then, the target physical parameters can be selected from the physical parameter set according to the parameter selection type, in order to prepare for the subsequent integration processing of the three-dimensional Gaussian kernel.

[0128] It is worth noting that the target physical parameters are selected from the physical parameter set according to the parameter selection type, and then the target physical parameters and the corresponding three-dimensional Gaussian kernel can be fused, so that the subsequent movements of the three-dimensional digital human can be more natural.

[0129] In addition, if Figure 8 As shown, an embodiment of the present application further provides a digital human generation device 10, comprising:

[0130] The first processing module 100 is used to obtain multi-view character images;

[0131] The second processing module 200 is used to calculate and process the multi-view character image based on a preset motion structure recovery algorithm to obtain an initialization point cloud;

[0132] The third processing module 300 is used to perform three-dimensional Gaussian distribution rendering processing on the initialization point cloud to obtain a three-dimensional Gaussian volume;

[0133] A fourth processing module 400 is used to classify the three-dimensional Gaussian volume to obtain a plurality of three-dimensional Gaussian kernels;

[0134] A fifth processing module 500 is used to integrate a preset physical parameter set with a plurality of three-dimensional Gaussian kernels to obtain a plurality of discrete particle clouds;

[0135] The sixth processing module 600 is used to solve each discrete particle cloud based on a preset material point algorithm to obtain a three-dimensional Gaussian digital human.

[0136] It should be noted that, in the process of generating a digital human, firstly, a multi-view character image is obtained; then, the multi-view character image is calculated and processed based on a preset motion structure recovery algorithm to obtain an initialization point cloud; then, the initialization point cloud is subjected to three-dimensional Gaussian distribution rendering processing to obtain a three-dimensional Gaussian body; then, the three-dimensional Gaussian body is classified to obtain multiple three-dimensional Gaussian kernels; then, the preset physical parameter set is integrated with the multiple three-dimensional Gaussian kernels to obtain multiple discrete particle clouds; finally, each discrete particle cloud is solved based on a preset material point algorithm to obtain a three-dimensional Gaussian digital human; through the above technical solution, the preset physical parameter set is integrated with the multiple three-dimensional Gaussian kernels to obtain multiple discrete particle clouds, so that the action and expression of the generated digital human can be closer to nature, the digital human can be more refined, the action can be more natural, and a better user experience can be brought to the user.

[0137] The specific implementation of the digital human generation device 10 is basically the same as the specific implementation of the above-mentioned digital human generation method, and will not be repeated here.

[0138] In addition, if Fig. 9 As shown, an embodiment of the present application further provides an electronic device 700 , which includes: a memory 720 , a processor 710 , and a computer program stored in the memory 720 and executable on the processor 710 .

[0139] The processor 710 and the memory 720 may be connected via a bus or in other ways.

[0140] The non-transitory software programs and instructions required to implement the digital human generation methods of the above embodiments are stored in the memory 720 , and when executed by the processor 710 , the digital human generation methods of the above embodiments are executed.

[0141] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0142] In addition, an embodiment of the present application further provides a computer-readable storage medium, which stores computer-executable instructions, and the computer-executable instructions are executed by a processor 710 or a controller, for example, by a processor 710 in the above-mentioned device embodiment, so that the above-mentioned processor 710 can execute the digital human generation method in the above-mentioned embodiment.

[0143] The above embodiments may be used in combination, and modules with the same name in different embodiments may be the same or different.

[0144] The above describes specific embodiments of the present application, and other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily have to be performed in the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0145] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and computer-readable storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0146] The apparatus, device, computer-readable storage medium and method provided in the embodiments of the present application correspond to each other. Therefore, the apparatus, device and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the corresponding apparatus, device and computer storage medium will not be repeated here.

[0147] In the 1990s, improvements to a technology could be clearly distinguished as hardware improvements (for example, improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the method flow). However, with the development of technology, many improvements to the method flow today can be regarded as direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that an improvement in a method flow cannot be implemented using a hardware entity module. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system on a PLD by programming it themselves, without having to ask a chip manufacturer to design and produce a dedicated integrated circuit chip. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages ​​and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.

[0148] The controller can be implemented in any appropriate manner, for example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (such as software or firmware) that can be executed by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in a purely computer-readable program code manner, the controller can be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, this controller can be considered as a hardware component, and the devices included therein for implementing various functions can also be regarded as structures within the hardware component. Or even, the devices for implementing various functions can be regarded as both software modules for implementing the method and structures within the hardware component.

[0149] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0150] For the convenience of description, the above devices are described in terms of functions and are divided into various units. Of course, when implementing the embodiments of the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0151] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the embodiments of the present application may adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the embodiments of the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0152] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0153] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0154] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0155] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0156] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0157] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0158] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0159] In the embodiments of the present application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Among them, A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c can be single or multiple.

[0160] Embodiments of the present application may be described in the general context of computer executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. Embodiments of the present application may also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0161] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0162] The above is only the embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for generating a digital human, characterized in that: The method comprises: Acquire multi-view character images; Based on a preset motion structure recovery algorithm, the multi-view character image is processed to obtain an initialization point cloud; Performing three-dimensional Gaussian distribution rendering processing on the initialization point cloud to obtain a three-dimensional Gaussian volume; Classifying the three-dimensional Gaussian body to obtain a plurality of three-dimensional Gaussian kernels; Integrating a preset physical parameter set with the plurality of three-dimensional Gaussian kernels to obtain a plurality of discrete particle clouds; Each discrete particle cloud is solved based on a preset material point algorithm to obtain a three-dimensional Gaussian digital human.

2. The method for generating a digital human according to claim 1, characterized in that: The method of calculating and processing the multi-view character image based on a preset motion structure recovery algorithm to obtain an initialization point cloud includes: Performing feature point detection on the multi-view character image to obtain image feature points; Performing feature description on the image feature points to obtain image feature description information; Performing feature matching processing on the image feature points according to the image feature description information to obtain matching feature points; determining motion estimation information according to the matching feature points; Performing triangulation based on the matching feature points and the motion estimation information to obtain three-dimensional coordinate information; The initialization point cloud is constructed according to the three-dimensional coordinate information.

3. The method for generating a digital human according to claim 1, characterized in that: The performing three-dimensional Gaussian distribution rendering processing on the initialization point cloud to obtain a three-dimensional Gaussian volume includes: Preprocessing the initialization point cloud to obtain a preprocessed point cloud; Extracting initial feature points from the preprocessed point cloud; Constructing an initial Gaussian distribution model according to the initial feature points, wherein the initial Gaussian distribution model carries mean information and covariance information; Performing surface reconstruction on the initial Gaussian distribution model according to the mean information and the covariance information to obtain a surface reconstructed Gaussian distribution model; Performing surface smoothing processing on the surface reconstructed Gaussian distribution model to obtain a smoothed Gaussian distribution model; The smoothed Gaussian distribution model is rendered to obtain the three-dimensional Gaussian volume.

4. The method for generating a digital human according to claim 1, characterized in that: The classifying process of the three-dimensional Gaussian volume to obtain a plurality of three-dimensional Gaussian kernels includes: Determine segmentation intensity information, segmentation gradient information and shape feature information according to the three-dimensional Gaussian volume; The three-dimensional Gaussian volume is segmented according to the segmentation intensity information, the segmentation gradient information and the shape feature information to obtain a plurality of the three-dimensional Gaussian kernels.

5. The method for generating a digital human according to claim 1, characterized in that: Each of the three-dimensional Gaussian kernels carries classification information, and the preset physical parameter set is integrated with the multiple three-dimensional Gaussian kernels to obtain multiple discrete particle clouds, including: determining a target physical parameter from the physical parameter set according to the classification information carried by the three-dimensional Gaussian kernel; The target physical parameter and the corresponding three-dimensional Gaussian kernel are fused to obtain the discrete particle cloud.

6. The method for generating a digital human according to claim 1, characterized in that: The method solves each discrete particle cloud based on a preset material point algorithm to obtain a three-dimensional Gaussian digital human, including: Obtaining state information of the discrete particle cloud and constructing a background grid; Interpolating and mapping the state information onto the grid nodes of the background grid; Performing physical quantity calculation processing on the grid nodes of the background grid to obtain force values ​​and momentum; updating the physical information of the background grid according to the force value and the momentum; The state information of the discrete particle cloud is updated according to the updated physical information to obtain the three-dimensional Gaussian digital human.

7. The method for generating a digital human according to claim 5, characterized in that: The step of determining a target physical parameter from the physical parameter set according to the classification information carried by the three-dimensional Gaussian kernel comprises: Determining a parameter selection type according to the classification information; The target physical parameter is selected from the physical parameter set according to the parameter selection type.

8. A digital human generation device, characterized in that: include: A first processing module, used for acquiring multi-view character images; A second processing module is used to calculate and process the multi-view character image based on a preset motion structure recovery algorithm to obtain an initialization point cloud; A third processing module is used to perform three-dimensional Gaussian distribution rendering processing on the initialization point cloud to obtain a three-dimensional Gaussian volume; A fourth processing module, used for classifying the three-dimensional Gaussian volume to obtain a plurality of three-dimensional Gaussian kernels; A fifth processing module, used for integrating a preset physical parameter set with the plurality of three-dimensional Gaussian kernels to obtain a plurality of discrete particle clouds; The sixth processing module is used to solve each discrete particle cloud based on a preset material point algorithm to obtain a three-dimensional Gaussian digital human.

9. An electronic device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the digital human generation method as described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing computer-executable instructions, characterized in that: The computer executable instructions are used to execute the digital human generation method described in any one of claims 1 to 7.

Images