Intelligent virtual article synthesis method and computer readable storage medium

Abstract

The invention provides an intelligent virtual article synthesis method and a computer readable storage medium, and relates to the technical field of virtual reality. The method comprises the steps of obtaining an article three-dimensional model corresponding to a virtual article in response to a synthesis operation; performing discretization conversion on the three-dimensional model of the article, and determining an article voxel model; based on the article modal information, article voxel units in the article three-dimensional model are paired, and an article voxel unit set is determined; synthesizing the article modal information corresponding to the article voxel unit set to generate fusion modal information; and constructing a fusion voxel unit according to the fusion modal information, and constructing a synthetic article model through the fusion voxel unit. According to the technical scheme, the flexibility of virtual article synthesis in a virtual reality scene can be effectively improved, the applicability and reliability of virtual article synthesis are improved, and technical support is provided for diversified synthesis of complex virtual articles.

Description

Background Art

[0002] In the context of the rapid development of virtual reality technology, the generation and operation technology of virtual objects has become an important part of promoting innovation in virtual reality applications. With the widespread application of virtual reality devices in games, education, industrial design and other fields, users' demand for flexible interaction and customization of virtual objects continues to increase. Especially in virtual reality scenes, the synthesis operation of multiple virtual objects, as an important interaction method, can effectively enhance the user's immersion and the diversification of virtual objects.

[0003] Traditional virtual item synthesis technologies are mostly based on direct model operations or static synthesis rules. This method has low reliance on the computing power of the device, but it is difficult to cope with complex virtual scenes and dynamic requirements. For example, static rules often lack sufficient flexibility when facing personalized needs proposed by users, and cannot adjust the synthesis strategy in real time according to changes in the scene. In addition, these methods usually use direct geometric operations and lack a deep understanding of the functionality and interactivity of virtual items, resulting in the synthesis results being unable to meet the actual application requirements in terms of functionality.

[0004] In general, although the application prospects of virtual reality technology are broad, the existing technologies generally have certain limitations in flexibility, real-time and resource adaptability in the synthesis function of virtual objects, and it is difficult to fully meet the needs of users for the synthesis of complex virtual objects in dynamic scenes.

[0005] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0006] The purpose of the embodiments of the present disclosure is to provide an intelligent virtual item synthesis method and a computer-readable storage medium, thereby effectively improving the flexibility of virtual item synthesis in a virtual reality scene, improving the applicability and reliability of virtual item synthesis, and improving the operational efficiency and response speed of virtual item synthesis.

[0007] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by the practice of the present disclosure.

[0008] According to a first aspect of an embodiment of the present disclosure, a method for synthesizing an intelligent virtual item is provided, which is applied to a virtual reality device, wherein the virtual reality device generates a virtual reality scene when running a virtual reality application program, and the method comprises:

[0009] In response to a synthesis operation on at least two virtual objects in the virtual reality scene, obtaining a three-dimensional object model corresponding to the virtual object;

[0010] Discretize the three-dimensional model of the object to determine a voxel model of the object, wherein the voxel model of the object includes object voxel units and object modal information associated with each of the object voxel units;

[0011] Pairing the object voxel units in the three-dimensional model of the object based on the object modal information to determine an object voxel unit set;

[0012] synthesizing the object modal information corresponding to the object voxel unit set to generate fused modal information;

[0013] A fused voxel unit is constructed according to the fused modal information, and a synthetic object model is built by using the fused voxel unit.

[0014] In some example embodiments of the present disclosure, based on the aforementioned scheme, the synthesizing the item modal information corresponding to the item voxel unit set to generate fused modal information includes: determining device resource information of the virtual reality device, the device resource information including at least device computing resources and device network resources; if the device resource information is greater than or equal to a preset resource threshold, synthesizing the item modal information corresponding to the item voxel unit set by a pre-trained synthesis model based on a generative adversarial network to generate fused modal information; if the device resource information is less than the preset resource threshold, synthesizing the item modal information corresponding to the item voxel unit set by a preset synthesis voxel library to generate fused modal information.

[0015] In some example embodiments of the present disclosure, based on the aforementioned scheme, the synthesis model based on the generative adversarial network includes a generator and a discriminator; the pre-trained synthesis model based on the generative adversarial network is used to synthesize the item modal information corresponding to the item voxel unit set to generate fused modal information, including: constructing an item modal feature vector according to the item modal information corresponding to the item voxel unit set; inputting the item modal feature vector into the generator of the synthesis model based on the generative adversarial network to generate fused modal information; evaluating the fused modal information using the discriminator of the synthesis model based on the generative adversarial network, and optimizing the generator through adversarial training until the latest generated fused modal information is evaluated successfully, and outputting the fused modal information.

[0016] In some example embodiments of the present disclosure, based on the aforementioned scheme, the item modal information corresponding to the item voxel unit set is synthesized through a preset synthetic voxel library to generate fused modal information, including: clustering the item modal information corresponding to the item voxel unit set to obtain multiple modal center vectors; performing similarity matching between the modal center vector and the fused modal information corresponding to each fused voxel unit in the preset synthetic voxel library to determine the fused modal information of the item modal information.

[0017] In some example embodiments of the present disclosure, based on the aforementioned scheme, the discretization conversion of the object three-dimensional model to determine the object voxel model includes: parsing the three-dimensional model file corresponding to the object three-dimensional model to determine the model geometry data of the three-dimensional model, the model geometry data including vertex coordinates, edge connection relationships and normal vectors; discretizing the object three-dimensional model into multiple object voxel units according to preset voxel parameters and the model geometry data; determining the object modal information associated with each of the object voxel units through the three-dimensional model file; and constructing the object voxel model based on the voxel position coordinates of the object voxel unit and the object modal information.

[0018] In some example embodiments of the present disclosure, based on the aforementioned scheme, the item modal information includes at least geometric features, material features and functional features; the geometric features include the voxel position coordinates and neighborhood connection characteristics of the item voxel units; the material features include the color, transparency and glossiness of the item voxel units; the functional features include the item functional attributes of the item voxel units in the virtual reality scene, and the item functional attributes include operability and interaction logic.

[0019] In some example embodiments of the present disclosure, based on the aforementioned scheme, the item voxel units in the three-dimensional model of the item are paired based on the item modal information to determine a set of item voxel units, including: performing similarity matching on the item voxel units of the three-dimensional model of the item based on the geometric features of the item voxel units to determine geometric candidate voxel units; comparing the geometric candidate voxel units using the material features of the item voxel units to screen out material candidate voxel units that match the material characteristics; prioritizing the material candidate voxel units according to the functional features of the item voxel units, and determining a final paired set of item voxel units based on a priority threshold.

[0020] In some example embodiments of the present disclosure, based on the aforementioned scheme, constructing a fused voxel unit according to the fused modal information, and building a synthetic object model through the fused voxel unit, includes: obtaining a preset standard voxel unit, and adjusting and setting the standard voxel unit through the fused modal information to obtain a fused voxel unit; constructing a fused voxel grid based on the spatial connection relationship of the fused voxel unit, and performing surface optimization and material mapping on the fused voxel grid to obtain a synthetic object model.

[0021] In some example embodiments of the present disclosure, based on the aforementioned scheme, the method includes: responding to an adjustment instruction for the synthetic object model, adjusting the fused modal information through the adjustment instruction to obtain new fused modal information; generating a new fused voxel unit according to the new fused modal information; and rebuilding the synthetic object model through the new fused voxel unit and rendering it in real time.

[0022] According to a second aspect of an embodiment of the present disclosure, there is provided an electronic device, comprising: a processor; and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, any one of the above-mentioned intelligent virtual item synthesis methods is implemented.

[0023] According to a third aspect of an embodiment of the present disclosure, there is provided a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the method for synthesizing an intelligent virtual item according to any one of the above items is implemented.

[0024] The technical solution provided by the embodiments of the present disclosure may have the following beneficial effects:

[0025] The intelligent virtual item synthesis method in the exemplary embodiment of the present disclosure can dynamically adapt to the characteristic requirements of virtual items in different scenes by performing specific structural processing on virtual item models, effectively avoiding the single reliance on static rules in related technologies, so that the synthesis strategy of virtual items can be flexibly adjusted according to actual operations; different item models in virtual reality scenes can be accurately parsed into basic units with multi-dimensional attributes through specific conversion and extraction of associated information. This unitized processing method provides basic support for the diversified combination of complex virtual items; by performing multi-level decomposition and integration of multi-dimensional information of virtual items, the resource consumption of direct calculation of highly complex models can be reduced; through the application of characteristic association processing and adaptive data structure, the synthesis operation can be efficiently completed in the virtual reality device, overcoming the reliance of traditional geometric operations on high computing resources, so that portable virtual reality devices can also synthesize virtual items under limited computing power, thereby improving operation efficiency and response speed, and meeting the needs of real-time interaction. The system can meet the needs of virtual objects in different scenes; through systematic analysis and comprehensive matching of the geometry, material and function of virtual objects, it can ensure that the generated virtual objects have functional consistency and material continuity while having reasonable structure; especially in geometric property matching, through detailed analysis of position and connection characteristics, the morphological distortion problem caused by direct operation in related technologies is solved; in material property processing, through the refined management of associated data, the visual consistency of the final synthesized objects in the virtual scene is guaranteed; in functional property optimization, through the inherent association of interaction logic, the operability and use value of virtual objects in the scene are significantly improved; through the processing method for the resource characteristics of virtual reality devices, efficient computing and lightweight computing are balanced, and the applicability bottleneck of high resource consumption methods in resource-constrained environments is effectively avoided, thereby realizing the universal expansion of technical means. Therefore, in the case of unstable network environment or device performance limitations, the stability and integrity of virtual object synthesis operations can still be guaranteed.

[0026] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.

[0028] Figure 1 The flowchart of the intelligent virtual item synthesis method according to some embodiments of the present disclosure is schematically shown.

[0029] Figure 2 A schematic diagram of a process for generating fused modality information according to some embodiments of the present disclosure is schematically shown.

[0030] Figure 3 A flowchart of constructing an object voxel model according to some embodiments of the present disclosure is schematically shown.

[0031] Figure 4 The flowchart of pairing item voxel units to obtain an item voxel unit set according to some embodiments of the present disclosure is schematically shown.

[0032] Figure 5 A schematic diagram of the structure of a computer system of an electronic device according to some embodiments of the present disclosure is schematically shown.

[0033] Figure 6 A schematic diagram of a computer-readable storage medium according to some embodiments of the present disclosure is schematically shown.

[0034] In the drawings, the same or corresponding reference numerals represent the same or corresponding parts. DETAILED DESCRIPTION

[0035] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with this specification. Instead, they are merely examples of devices and methods consistent with some aspects of this specification as detailed in the appended claims.

[0036] Furthermore, the drawings are only schematic illustrations and are not necessarily drawn to scale. The block diagrams shown in the drawings are only functional entities and do not necessarily correspond to physically separate entities. That is, these functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0037] In this example embodiment, first, a method for synthesizing an intelligent virtual item is provided. The method for synthesizing an intelligent virtual item can be applied to a virtual reality device or a terminal device supporting virtual reality technology. When the virtual reality device runs a virtual reality application, a virtual reality scene is generated. Of course, the method for synthesizing an intelligent virtual item can also be used in a server. This embodiment does not specifically limit this. The following description will take a virtual reality device or a terminal device supporting virtual reality technology as an example to illustrate the method. Figure 1The following schematically shows a flow chart of a method for synthesizing an intelligent virtual item according to some embodiments of the present disclosure. Figure 1 As shown, the intelligent virtual item synthesis method may include the following steps:

[0038] Step S110, in response to a synthesis operation on at least two virtual objects in the virtual reality scene, obtaining a three-dimensional object model corresponding to the virtual object;

[0039] Step S120, discretizing the three-dimensional model of the object to determine an object voxel model, wherein the object voxel model includes object voxel units and object modal information associated with each of the object voxel units;

[0040] Step S130, pairing the object voxel units in the three-dimensional model of the object based on the object modal information to determine an object voxel unit set;

[0041] Step S140, synthesizing the object modal information corresponding to the object voxel unit set to generate fused modal information;

[0042] Step S150, constructing a fused voxel unit according to the fused modal information, and building a synthetic object model through the fused voxel unit.

[0043] According to the intelligent virtual item synthesis method in this example embodiment, by performing specific structural processing on the virtual item model, it is possible to dynamically adapt to the characteristic requirements of virtual items in different scenarios, effectively avoiding the single reliance on static rules in related technologies, so that the synthesis strategy of virtual items can be flexibly adjusted according to actual operations; different item models in virtual reality scenarios can be accurately parsed into basic units with multi-dimensional attributes through specific conversion and extraction of associated information. This unitized processing method provides basic support for the diversified combination of complex virtual items; by performing multi-level decomposition and integration of the multi-dimensional information of virtual items, the resource consumption of direct calculation of highly complex models can be reduced; through the application of characteristic association processing and adaptive data structure, the synthesis operation can be efficiently completed in the virtual reality device, overcoming the dependence of traditional geometric operations on high computing resources, so that portable virtual reality devices can also synthesize virtual items under limited computing power, thereby improving operation efficiency and response speed, and meeting the needs of real-time interaction. The system can meet the needs of virtual objects in different scenes; through systematic analysis and comprehensive matching of the geometry, material and function of virtual objects, it can ensure that the generated virtual objects have functional consistency and material continuity while having reasonable structure; especially in geometric property matching, through detailed analysis of position and connection characteristics, the morphological distortion problem caused by direct operation in related technologies is solved; in material property processing, through the refined management of associated data, the visual consistency of the final synthesized objects in the virtual scene is guaranteed; in functional property optimization, through the inherent association of interaction logic, the operability and use value of virtual objects in the scene are significantly improved; through the processing method for the resource characteristics of virtual reality devices, efficient computing and lightweight computing are balanced, and the applicability bottleneck of high resource consumption methods in resource-constrained environments is effectively avoided, thereby realizing the universal expansion of technical means. Therefore, in the case of unstable network environment or device performance limitations, the stability and integrity of virtual object synthesis operations can still be guaranteed.

[0044] Next, the intelligent virtual item synthesis method in this example embodiment will be further described.

[0045] In step S110, in response to a synthesis operation on at least two virtual objects in the virtual reality scene, a three-dimensional object model corresponding to the virtual object is obtained.

[0046] In an example embodiment of the present disclosure, a synthesis operation refers to an operation instruction initiated by a user through an input device of a virtual reality device (such as a controller, a gesture capture system, or a voice instruction). For example, the synthesis operation may include specific instruction information for selecting a virtual item and triggering the synthesis operation.

[0047] The three-dimensional model of an object can be used as a digital representation of a virtual object, usually stored in the form of a three-dimensional mesh, including vertex data, edge data, face data, and related texture coordinates and material information. The interface layer of the virtual reality application can capture the operation and parse the object identifier selected by the user, call the resource management module of the virtual object, and obtain the corresponding three-dimensional model file from the pre-loaded virtual object database. These files may be stored in a standard format (such as .obj file format, .fbx file format, etc.) or a custom format, and need to be read by a dedicated parser for subsequent processing. Optionally, the synthesis operation target can be determined directly through user gesture recognition, or the object synthesis range can be specified by clicking or dragging in the scene, thereby providing a variety of interaction methods.

[0048] In step S120, the three-dimensional model of the object is discretized to determine an object voxel model, wherein the object voxel model includes object voxel units and object modal information associated with each of the object voxel units.

[0049] In an exemplary embodiment of the present disclosure, the object voxel model refers to a three-dimensional discrete representation based on a regular grid, which represents complex geometric structures in a simpler form by decomposing continuous three-dimensional geometric data into discrete voxel units. The implementation of discretization can generally include three stages: three-dimensional grid analysis, regular grid division and data association. For example, the geometric data of the object three-dimensional model, including the connection relationship between vertices, edges and faces, can be extracted through three-dimensional grid analysis, and these data are used as the basis for the subsequent voxelization of space division; then, the three-dimensional space can be divided into regular grid units according to the preset voxel resolution, and the size of each unit depends on the setting of the voxel resolution. A higher resolution provides a more refined structural representation, while a lower resolution can optimize the computing performance; finally, the grid unit can be associated with the object modal information, which includes geometric characteristics (such as voxel position and neighborhood connectivity), material characteristics (such as color, transparency and glossiness) and functional characteristics (such as interactive logic and operability), and each voxel unit is given a unique modal characteristic description through a characteristic association algorithm.

[0050] In an optional implementation, the three-dimensional model of the object can be discretized based on an irregular grid, or simplified using a direct point cloud conversion method. Through the above steps, the operability of the object geometric data can be significantly improved, and a unified data basis can be provided for subsequent voxel unit pairing and modal information processing.

[0051] In step S130, the object voxel units in the object three-dimensional model are paired based on the object modal information to determine an object voxel unit set.

[0052] In an exemplary embodiment of the present disclosure, item voxel unit pairing refers to a process of matching item voxel units of two virtual items according to item modality information and generating an association relationship.

[0053] The pairing can first be implemented based on geometric characteristics, by calculating the positional similarity and neighborhood connectivity similarity between voxel units to screen out preliminary pairing candidate units; secondly, the candidate units can be further screened using material characteristics, which include parameters such as voxel color, transparency, and gloss, and a more accurate matching result can be obtained by calculating the weighted similarity of multi-dimensional characteristics; finally, the screened units can be prioritized according to functional characteristics, with priority given to units that are closely related in interaction logic, to ensure that the generated voxel set can retain the core functions of the original object.

[0054] It is understandable that local pairing optimization can also be performed based on the global optimization strategy of the graph matching algorithm, or by specifying a specific area through user interaction, which is not specifically limited in this example embodiment. Through the above steps, the accuracy and consistency of pairing can be effectively ensured, laying the foundation for the subsequent generation of fusion modality information.

[0055] In step S140, the object modal information corresponding to the object voxel unit set is synthesized to generate fused modal information.

[0056] In an exemplary embodiment of the present disclosure, fused modal information refers to a comprehensive characteristic description obtained by jointly processing the modal characteristics of a set of object voxel units, and its generation process depends on the specific implementation of the modal characteristic synthesis algorithm.

[0057] Through morphological synthesis based on geometric characteristics, the spatial position and neighborhood connection characteristics of voxel units can be interpolated or reconstructed to ensure the continuity of the synthesized structure; through fusion strategies based on material characteristics, material parameters can be weighted averaged or selectively retained to ensure that the final material effect is visually consistent; through optimization algorithms based on functional characteristics, the interaction logic can be integrated to reasonably integrate the functional attributes of the two objects into the new voxel unit.

[0058] It is understandable that the rules engine can also be used to predefine the modal information synthesis rules, or the modal characteristic reasoning generation can be realized through the synthesis model based on the generative adversarial network. Through the above steps, it is possible to generate fused modal information that meets the multi-dimensional consistency of structure, material and function, and provide data support for the construction of fused voxel units.

[0059] In step S150, a fused voxel unit is constructed according to the fused modality information, and a synthetic object model is built by using the fused voxel unit.

[0060] In an example embodiment of the present disclosure, the construction of fused voxel units may include feature allocation, unit reorganization, and structural optimization. First, the geometric position, material parameters, and functional characteristics of the voxel unit can be reallocated by fusion modal information to ensure the integrity of the unit characteristics in multiple dimensions; second, the fused voxel unit can be constructed into a voxel grid with a continuous topological structure by using a reorganization algorithm through spatial connectivity analysis; finally, the voxel grid can be converted into a three-dimensional surface model through surface optimization and material mapping, and the surface material and texture effect can be adjusted according to the fusion modal information.

[0061] In an optional implementation, a multi-resolution reorganization strategy can be used to perform multi-level processing on the fused voxel grid, or a dynamic optimization algorithm can be used to achieve real-time model updating. Through the above steps, a synthetic object model with coordinated multi-dimensional characteristics is finally generated, and can present an efficient and stable use effect in a virtual reality scene.

[0062] Next, the contents of step S110 to step S150 are described in detail.

[0063] In an exemplary embodiment of the present disclosure, Figure 2 The steps in are used to synthesize the modal information of the items corresponding to the set of voxel units of the items and generate fused modal information. Figure 2 As shown, it may specifically include:

[0064] Step S210, determining device resource information of the virtual reality device, wherein the device resource information at least includes device computing resources and device network resources;

[0065] Step S220: if the device resource information is greater than or equal to a preset resource threshold, synthesizing the item modal information corresponding to the item voxel unit set by using a pre-trained synthesis model based on a generative adversarial network to generate fused modal information;

[0066] Step S230: If the device resource information is less than the preset resource threshold, the item modal information corresponding to the item voxel unit set is synthesized through a preset synthetic voxel library to generate fused modal information.

[0067] Among them, the device resource information is used to determine whether the virtual reality device can support highly complex object modal information synthesis operations. The device computing resources may include the device's processor performance (such as the frequency and core number corresponding to the central processing unit and image processing unit), available memory capacity, and real-time computing load. The device computing resources can be obtained in real time through the operating system resource monitoring interface, and the resource status description can be dynamically generated. The device network resources may include the network bandwidth, delay, and packet loss rate between the current device and the cloud server. The actual network status can be calculated through multiple communication requests through the network status monitoring module.

[0068] Generative Adversarial Networks (GAN) can include a generator and a discriminator. The generator generates new modal information based on the input object modal feature vector, and the discriminator distinguishes the generated modal information from the real modal information, thereby improving the output quality of the generator through adversarial training.

[0069] The pre-trained synthetic model based on the generative adversarial network can be pre-trained on a variety of virtual item data sets, and its model parameters can adapt to the modal characteristics of various types of items. In the specific implementation process, the modal information of the item voxel unit set can be converted into a vector form and input into the generator. The generator infers new fused modal information based on the internal model parameters and optimizes the generator output through the feedback of the discriminator. Of course, lightweight generative networks (such as MobileGAN) can also be used to adapt to resource-constrained devices.

[0070] The synthetic voxel library is a set of fused voxel units pre-generated based on common modal characteristics, which covers a variety of geometric forms, material characteristics and functional characteristics, as an alternative to modal information synthesis. In the specific implementation, the modal information of the object voxel unit set can be clustered first, and the modal information with similar characteristics can be classified into one category; then the modal center vector of each category is matched with the standard fused modal information in the synthetic voxel library for similarity, and the closest fused modal information is screened out and replaced The generation step. Optionally, the synthetic voxel library can also be managed through distributed storage or some commonly used modal information can be cached locally on the device to improve query efficiency.

[0071] Through the above-mentioned resource information acquisition steps, the device resource status can be accurately judged in real time, providing a basis for the subsequent selection of synthesis strategies; through the synthesis method of the synthesis model based on the generative adversarial network, fused modal information with complex modal characteristics can be efficiently generated to meet the needs of high-precision synthesis; through the synthesis method of the voxel library, rapid response can be achieved under low-resource conditions while ensuring the rationality of the modal characteristics.

[0072] Optionally, the synthesis model based on the generative adversarial network includes a generator and a discriminator; the following steps may be performed to synthesize the object modal information corresponding to the object voxel unit set through the pre-trained synthesis model based on the generative adversarial network to generate fused modal information. Specifically, the following steps may be performed:

[0073] An item modal feature vector can be constructed based on the item modal information corresponding to a set of item voxel units, and the item modal feature vector can be input into a generator of a synthetic model based on a generative adversarial network to generate fused modal information, and the fused modal information can be evaluated by a discriminator of the synthetic model based on the generative adversarial network, and the generator can be optimized through adversarial training until the latest generated fused modal information is successfully evaluated and the fused modal information is output.

[0074] Among them, the generator can generate fused modal information with comprehensive characteristics based on the input item modal feature vector. Its network structure can usually include multiple layers of fully connected layers or convolutional layers, and can process multi-dimensional feature vectors. The discriminator can compare the generated fused modal information with the real modal information to evaluate whether the generated result meets the target requirements. This can be achieved through a binary classification task. Its network structure is usually a symmetric convolutional network to efficiently capture the differences between modal information. Through adversarial training between the generator and the discriminator, the output capacity of the generator can be continuously optimized.

[0075] When the modal feature vector of the item can be input into the generator, the construction of the feature vector is based on the multi-dimensional characteristics of the modal information. For example, the geometric feature vector can record the spatial position of each voxel unit, the neighborhood connectivity and other parameters, the material feature vector includes the color, transparency and glossiness values, and the functional feature vector represents the interaction logic in scalar form. The above modal information is encoded into a fixed-dimensional feature vector through a feature encoder and used as the input of the generator; optionally, an autoencoder can be used to reduce the dimensionality of the modal information to optimize the feature vector generation.

[0076] After the fused modal information output by the generator is verified by the discriminator, if the evaluation fails, the generator can update the model parameters according to the feedback of the discriminator until the generated fused modal information meets the requirements of the discriminator. Through this two-way optimization mechanism of the generative adversarial network, the quality and stability of modal information synthesis can be continuously improved.

[0077] In an exemplary embodiment of the present disclosure, the following steps may be used to synthesize the object modal information corresponding to the object voxel unit set through a preset synthetic voxel library to generate fused modal information, which may specifically include:

[0078] The modal information of the items corresponding to the set of item voxel units can be clustered to obtain multiple modal center vectors; the modal center vectors are similarly matched with the fused modal information corresponding to each fused voxel unit in the preset synthetic voxel library to determine the fused modal information of the item modal information.

[0079] Among them, the modal information of the object can generate multiple modal center vectors through clustering processing. The modal center vector refers to the mathematical expression of a group of similar modal information, which can represent the characteristics of the modal information of this group. The clustering algorithm can choose K-means clustering, hierarchical clustering or density-based spatial clustering of applications with noise (DBSCAN). The specific selection can be determined according to the distribution characteristics of the modal information of the object voxel unit. For example, K-means clustering can calculate the Euclidean distance of the modal information and repeatedly adjust the cluster center position to minimize the intra-group variance, thereby obtaining the optimal clustering result.

[0080] When performing similarity matching between the modal center vector and the fused modal information in the synthetic voxel library, a multi-dimensional weighted similarity calculation of the characteristic vector may be used. For example, the matching algorithm may include cosine similarity, Euclidean distance, or dynamic time warping (DTW) to measure the characteristic difference between the modal center vector and the synthetic voxel.

[0081] By setting the matching threshold, the fused modal information closest to the object voxel modal information can be screened out, and after the match is successful, the fused modal information in the synthetic voxel library can be directly assigned to the current modal unit without further generation calculation, thereby significantly reducing the computing burden of the device. Optionally, the matching process can be accelerated through the modal feature index tree or the approximate nearest neighbor algorithm can be used to improve efficiency.

[0082] By performing similarity matching between the modal center vector and the fused modal information corresponding to each fused voxel unit in the preset synthetic voxel library, the fused modal information of the object modal information can be determined. This allows the preset synthetic voxel library to be effectively utilized to achieve rapid generation of fused modal information while maintaining the advantages of the synthetic results in terms of feature consistency and computational efficiency.

[0083] In an exemplary embodiment of the present disclosure, Figure 3 The steps in the discretization conversion of the object 3D model are realized to determine the object voxel model. Figure 3 As shown, it may specifically include:

[0084] Step S310, parsing the 3D model file corresponding to the 3D model of the object to determine the model geometry data of the 3D model, wherein the model geometry data includes vertex coordinates, edge connection relationships, and normal vectors;

[0085] Step S320, discretizing the object three-dimensional model into a plurality of object voxel units according to preset voxel parameters and the model geometry data;

[0086] Step S330, determining the object modal information associated with each of the object voxel units through the three-dimensional model file;

[0087] Step S340: constructing an object voxel model based on the voxel position coordinates of the object voxel unit and the object modal information.

[0088] Among them, when the three-dimensional model of the object is discretized to determine the voxel model of the object, the three-dimensional model file needs to be fully parsed to obtain the basic data for constructing the voxel grid. The three-dimensional model file can be in .obj, .fbx and other file formats, and the three-dimensional model file can contain vertex coordinates, edge connection relationships, normal vectors, and material related information. The parsing process can extract the above data in sequence through the file reading module. The vertex coordinates can be used to define key points in the three-dimensional space, and the edge connection relationship and normal vector provide support for constructing surface topology and voxel mapping.

[0089] The three-dimensional model of the object can be divided into regular grids according to preset voxel parameters. By dividing the three-dimensional space into voxel units, the continuous geometric data can be discretized into regular structures. Voxel parameters can include voxel resolution and grid boundaries. Voxel resolution can determine the size and accuracy of each voxel unit, and the boundary is used to define the spatial range of the voxel grid. In the implementation process, a spatial partitioning algorithm (for example, an octree partitioning algorithm) can be used to quickly partition the model geometric data, and the voxel unit to which each vertex belongs can be determined according to the position of the vertex coordinates.

[0090] During the discretization process, each voxel unit can be associated with the relevant object modal information by parsing the material and functional information attached to the 3D model file. For example, material information can include color, transparency, and glossiness, and is assigned to the voxel unit through direct mapping or interpolation calculation; functional information can be assigned to the corresponding voxel unit by marking the interactive logical attributes of the object (such as grabbing, triggering). Optionally, point cloud-based discretization processing or dynamic meshing strategies can be used to adapt to different application scenarios and model complexity.

[0091] By discretizing the three-dimensional model of the object, the generated voxel model of the object not only retains the geometric structure of the original object, but also provides necessary support for subsequent voxel pairing and modal information synthesis through the association of the object's modal information.

[0092] Optionally, the modal information of an item refers to a multidimensional data set describing the characteristics of the voxel unit of the item, which may include geometric features, material features, and functional features. Among them, the geometric features can be used to represent the position and neighborhood relationship of the voxel unit in three-dimensional space, and may specifically include the spatial coordinates, boundary attributes, and connection relationship with adjacent voxels of the voxel. Through the geometric features, the spatial distribution and structural integrity of the voxel unit can be evaluated.

[0093] Material features can describe the visual and physical properties of voxel units, including model color (RGB value), transparency (scalar between 0 and 1) and glossiness (reflectivity parameter). Material features usually come from the material information in the 3D model file. If the original data is missing, it can be generated through material interpolation or default assignment.

[0094] Functional features are the operational properties of voxel units in a virtual reality scene, and commonly include operability (for example, operability can include whether it can be grasped or rotated) and interaction logic (for example, interaction logic can be the type of triggering event). Functional features can be generated by parsing the functional description file of the object model or preset rules, or dynamically added through user-defined interaction logic.

[0095] Optionally, the material features can be simplified into a single color attribute to reduce computational complexity, or functional features can be automatically generated through a machine learning model to improve dynamic adaptability.

[0096] In an exemplary embodiment of the present disclosure, Figure 4 The steps in the embodiment are to pair the object voxel units in the object 3D model based on the object modal information and determine the object voxel unit set. Figure 4 As shown, it may specifically include:

[0097] Step S410, performing similarity matching on the object voxel units of the three-dimensional model of the object based on the geometric features of the object voxel units, and determining geometric candidate voxel units;

[0098] Step S420, comparing the geometric candidate voxel units with the material characteristics of the object voxel units, and screening out the material candidate voxel units that match the material characteristics;

[0099] Step S430 , prioritizing the material candidate voxel units according to the functional characteristics of the item voxel units, and determining a final paired item voxel unit set based on a priority threshold.

[0100] Among them, when pairing voxel units based on object modal information, preliminary screening can be carried out through geometric characteristics. Specifically, the spatial coordinates of the voxel units can be used to calculate the position similarity, and the continuity of the geometric structure can be evaluated through the neighborhood connection relationship, so as to screen out geometric candidate voxel units. For example, the geometric matching algorithm can use Euclidean distance or dynamic neighborhood matching strategy, and this example embodiment does not specifically limit this.

[0101] When further screening the geometric candidate voxel units, material characteristics can be used for comparison. Material matching can be done by calculating the weighted similarity of attributes such as color, transparency, and glossiness to screen out voxel units with matching material characteristics. In specific implementation, the normalized distance calculation formula can be used to measure the similarity between different characteristics.

[0102] The screened voxel units can be prioritized according to the functional characteristics, and the voxel units with higher correlation in interaction logic are retained first. The priority sorting algorithm can assign weights to the voxel units according to the importance of the functional characteristics, and screen the final set of paired voxel units according to the weight values.

[0103] Optionally, all voxel units can be globally optimized based on a graph matching algorithm, or voxel pairing rules can be directly specified by user input. Figure 4 The generated voxel unit set can simultaneously meet the multi-dimensional requirements of geometry, material and functional characteristics, laying the foundation for subsequent modal information synthesis.

[0104] In an exemplary embodiment of the present disclosure, the following steps may be used to construct a fused voxel unit according to the fused modality information, and to build a synthetic object model by using the fused voxel unit, which may specifically include:

[0105] A preset standard voxel unit can be obtained, and the standard voxel unit can be adjusted and set by fusing modal information to obtain a fused voxel unit; a fused voxel grid can be constructed based on the spatial connection relationship of the fused voxel unit, and the fused voxel grid can be surface optimized and material mapped to obtain a synthetic object model.

[0106] Among them, when constructing a fused voxel unit according to the fused modal information, the fused modal information can be mapped through a standard voxel unit template. The standard voxel unit template is a preset regularized unit whose structure and attributes can adapt to the modal characteristics of most objects. When the fused modal information is used to adjust the standard voxel unit, the geometric characteristics can be used to modify the spatial position and neighborhood connection relationship of the voxel, the material characteristics can be used to update parameters such as color and glossiness, and the functional characteristics can be used for the interactive logic assignment of the voxel.

[0107] When building a synthetic object model by fusing voxel units, a fused voxel grid can be constructed based on the spatial connection relationship of the voxel units, and the topology optimization algorithm can be used to ensure the continuity and integrity of the grid structure. Subsequently, the fused voxel grid can be surface optimized, and a smooth surface structure can be generated using surface fitting technology; at the same time, the material properties in the fused modal information can be applied to the grid surface through material mapping technology to improve the visual performance.

[0108] Optionally, a multi-resolution fusion strategy can be used to improve the adaptability, or an adaptive mesh reconstruction algorithm can be used to further optimize the surface effect. By fusing modal information to construct a fused voxel unit, and by fusing voxel units to build a synthetic object model, a synthetic object model with coordinated multi-dimensional characteristics of geometry, material and function can be generated, which is suitable for the diverse needs in virtual reality scenes.

[0109] In an exemplary embodiment of the present disclosure, the synthetic item model may be readjusted by the following steps, which may specifically include:

[0110] It can respond to adjustment instructions for the synthetic object model, adjust the fusion modal information through the adjustment instructions to obtain new fusion modal information, and generate new fusion voxel units based on the new fusion modal information, and then the synthetic object model can be rebuilt through the new fusion voxel units and rendered in real time.

[0111] When an adjustment instruction of the composite object model is detected, the fused modal information can be updated through the adjustment instruction. The adjustment instruction can include geometric shape modification (such as scaling, rotation), material attribute adjustment (such as color change, transparency change) and functional attribute modification (such as interactive logic redefinition). The instruction parsing module can parse the user input and convert the adjustment content into an incremental modification of the fused modal information.

[0112] Based on the updated fusion modality information, the fused voxel units can be regenerated and the object model can be updated. The update process can include voxel position adjustment, material attribute redistribution, and dynamic update of functional characteristics to ensure the compatibility of the adjusted voxel units with the original structure. Subsequently, the synthetic object model can be rebuilt with the updated voxel units and rendered in real time in a virtual reality scene.

[0113] Optionally, a more intuitive adjustment interface can be provided through a parameterized controller, or a real-time feedback mechanism can be used to optimize the interactive experience of the adjustment results. Adjusting the synthetic item model through the user's adjustment instructions can dynamically adapt to user needs and improve the interactivity and flexibility of virtual item synthesis.

[0114] It should be noted that, although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.

[0115] In addition, in an exemplary embodiment of the present disclosure, an electronic device capable of implementing the above-mentioned intelligent virtual item synthesis method is also provided.

[0116] Those skilled in the art will appreciate that various aspects of the present disclosure may be implemented as systems, methods or program products. Therefore, various aspects of the present disclosure may be specifically implemented in the following forms, namely: complete hardware embodiments, complete software embodiments (including firmware, microcode, etc.), or embodiments combining hardware and software aspects, which may be collectively referred to herein as "circuits", "modules" or "systems".

[0117] Refer to the following Figure 5 hereinafter describes an electronic device 500 according to such an embodiment of the present disclosure. Figure 5 The electronic device 500 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0118] like Figure 5 As shown, the electronic device 500 is in the form of a general computing device. The components of the electronic device 500 may include but are not limited to: at least one processing unit 510, at least one storage unit 520, a bus 530 connecting different system components (including the storage unit 520 and the processing unit 510), and a display unit 540.

[0119] The storage unit stores program codes, which can be executed by the processing unit 510, so that the processing unit 510 performs the steps according to various exemplary embodiments of the present disclosure described in the above “Exemplary Method” section of this specification. For example, the processing unit 510 can perform the following steps: Figure 1In step S110, in response to a synthesis operation on at least two virtual objects in the virtual reality scene, a three-dimensional object model corresponding to the virtual object is obtained; in step S120, the three-dimensional object model is discretized to determine an object voxel model, wherein the object voxel model includes an object voxel unit and object modal information associated with each of the object voxel units; in step S130, the object voxel units in the three-dimensional object model are paired based on the object modal information to determine an object voxel unit set; in step S140, the object modal information corresponding to the object voxel unit set is synthesized to generate fused modal information; in step S150, a fused voxel unit is constructed according to the fused modal information, and a synthetic object model is constructed through the fused voxel unit.

[0120] The storage unit 520 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 521 and / or a cache memory unit 522 , and may further include a read-only memory unit (ROM) 523 .

[0121] The storage unit 520 may also include a program / utility 524 having a set (at least one) of program modules 525, such program modules 525 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0122] Bus 530 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0123] The electronic device 500 may also communicate with one or more external devices 570 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 500, and / or may communicate with any device that enables the electronic device 500 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 550. Furthermore, the electronic device 500 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 560. As shown, the network adapter 560 communicates with other modules of the electronic device 500 via a bus 530. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 500, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0124] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the embodiment of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the embodiment of the present disclosure.

[0125] In an exemplary embodiment of the present disclosure, a computer-readable storage medium is also provided, on which a program product capable of implementing the above method of the present specification is stored. In some possible embodiments, various aspects of the present disclosure may also be implemented in the form of a program product, which includes a program code, and when the program product is run on a terminal device, the program code is used to enable the terminal device to perform the steps according to various exemplary embodiments of the present disclosure described in the above "Exemplary Method" section of the present specification.

[0126] refer to Figure 6 As shown, a program product 600 for implementing the above-mentioned intelligent virtual item synthesis method according to an embodiment of the present disclosure is described, which can adopt a portable compact disk read-only memory (CD-ROM) and include program code, and can be run on a terminal device, such as a personal computer. However, the program product of the present disclosure is not limited thereto. In this document, a readable storage medium can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, an apparatus or a device.

[0127] The program product may use any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0128] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, in which readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Readable signal media may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0129] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.

[0130] Program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0131] In addition, the above-mentioned figures are only schematic illustrations of the processes included in the method according to the exemplary embodiments of the present disclosure, and are not intended to be limiting. It is easy to understand that the processes shown in the above-mentioned figures do not indicate or limit the time sequence of these processes. In addition, it is also easy to understand that these processes can be performed synchronously or asynchronously, for example, in multiple modules.

[0132] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the embodiment of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a touch terminal, or a network device, etc.) to execute the method according to the embodiment of the present disclosure.

[0133] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are to be considered exemplary only, and the true scope and spirit of the present disclosure are indicated by the claims.

[0134] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A method for synthesizing intelligent virtual items, characterized in that: Applied to a virtual reality device, the virtual reality device generates a virtual reality scene when running a virtual reality application program, the method comprising: In response to a synthesis operation on at least two virtual objects in the virtual reality scene, obtaining a three-dimensional object model corresponding to the virtual object; Discretize the three-dimensional model of the object to determine a voxel model of the object, wherein the voxel model of the object includes object voxel units and object modal information associated with each of the object voxel units; Pairing the object voxel units in the three-dimensional model of the object based on the object modal information to determine an object voxel unit set; synthesizing the object modal information corresponding to the object voxel unit set to generate fused modal information; A fused voxel unit is constructed according to the fused modal information, and a synthetic object model is built by using the fused voxel unit.

2. The intelligent virtual item synthesis method according to claim 1, characterized in that: The synthesizing the object modal information corresponding to the object voxel unit set to generate fused modal information includes: Determine device resource information of the virtual reality device, wherein the device resource information includes at least device computing resources and device network resources; If the device resource information is greater than or equal to a preset resource threshold, synthesizing the item modal information corresponding to the item voxel unit set through a pre-trained synthesis model based on a generative adversarial network to generate fused modal information; If the device resource information is less than the preset resource threshold, the item modal information corresponding to the item voxel unit set is synthesized through a preset synthetic voxel library to generate fused modal information.

3. The intelligent virtual item synthesis method according to claim 2, characterized in that: The synthetic model based on the generative adversarial network includes a generator and a discriminator; The synthesizing the object modal information corresponding to the object voxel unit set by using the pre-trained synthesis model based on the generative adversarial network to generate fused modal information includes: Constructing an item modal feature vector according to item modal information corresponding to the item voxel unit set; Inputting the modal feature vector of the item into the generator of the synthetic model based on the generative adversarial network to generate fused modal information; The fused modal information is evaluated using a discriminator of the synthetic model based on the generative adversarial network, and the generator is optimized through adversarial training until the latest generated fused modal information is successfully evaluated and the fused modal information is output.

4. The method for synthesizing intelligent virtual items according to claim 2, characterized in that: The synthesizing the object modal information corresponding to the object voxel unit set through a preset synthetic voxel library to generate fused modal information includes: Clustering the object modal information corresponding to the object voxel unit set to obtain multiple modal center vectors; The modal center vector is matched with the fused modal information corresponding to each fused voxel unit in a preset synthetic voxel library for similarity, so as to determine the fused modal information of the object modal information.

5. The intelligent virtual item synthesis method according to claim 1, characterized in that: The step of discretizing the three-dimensional model of the object to determine a voxel model of the object includes: Parsing a three-dimensional model file corresponding to the three-dimensional model of the object to determine model geometry data of the three-dimensional model, wherein the model geometry data includes vertex coordinates, edge connection relationships, and normal vectors; Discretizing the object three-dimensional model into a plurality of object voxel units according to preset voxel parameters and the model geometry data; Determining the object modal information associated with each of the object voxel units through the three-dimensional model file; An object voxel model is constructed based on the voxel position coordinates of the object voxel unit and the object modal information.

6. The intelligent virtual item synthesis method according to claim 1, characterized in that: The object modal information includes at least geometric features, material features and functional features; The geometric features include voxel position coordinates and neighborhood connectivity characteristics of the object voxel unit; The material characteristics include the color, transparency and glossiness of the object voxel unit; The functional features include the item functional attributes of the item voxel unit in the virtual reality scene, and the item functional attributes include operability and interaction logic.

7. The intelligent virtual item synthesis method according to claim 1 or 6, characterized in that: The step of pairing the object voxel units in the three-dimensional model of the object based on the object modal information to determine the object voxel unit set includes: Based on the geometric features of the object voxel units, similarity matching is performed on the object voxel units of the three-dimensional model of the object to determine geometric candidate voxel units; Comparing the geometric candidate voxel units with the material characteristics of the object voxel units, and screening out the material candidate voxel units that match the material characteristics; The material candidate voxel units are prioritized according to the functional characteristics of the item voxel units, and a final paired item voxel unit set is determined based on a priority threshold.

8. The intelligent virtual item synthesis method according to claim 1, characterized in that: The step of constructing a fused voxel unit according to the fused modality information and building a synthetic object model through the fused voxel unit includes: Acquire a preset standard voxel unit, and adjust and set the standard voxel unit according to the fusion modality information to obtain a fusion voxel unit; A fused voxel grid is constructed based on the spatial connection relationship of the fused voxel units, and surface optimization and material mapping are performed on the fused voxel grid to obtain a synthetic object model.

9. The intelligent virtual item synthesis method according to claim 1, characterized in that: The method comprises: In response to an adjustment instruction for the composite object model, adjusting the fused modal information according to the adjustment instruction to obtain new fused modal information; generating a new fused voxel unit according to the new fused modality information; The synthetic object model is rebuilt through the new fused voxel unit and rendered in real time.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the intelligent virtual item synthesis method as described in any one of claims 1 to 9 is implemented.

Images