Virtual object generation method, apparatus and system

Abstract

Provided in the present application are a virtual object generation method, an apparatus and a system. The method comprises: acquiring a user requirement; on the basis of the user requirement, determining a generation requirement, the generation requirement being used for generating a virtual object; decomposing the generation requirement into at least one subtask, each subtask among the at least one subtask corresponding to one requirement domain, and each requirement domain comprising at least one requirement function; determining a requirement function corresponding to the virtual object in at least one requirement domain; and invoking a function model of the requirement function corresponding to the virtual object in the at least one requirement domain to generate the virtual object. In the present solution, decomposing the generation requirement into multiple layers on the basis of the requirement domain and the requirement function can refine the generation requirement and ensure accurate invoking of the model, thereby generating more accurate virtual objects of higher quality.

Description

Method, device and system for generating virtual object

[0001] The present application claims priority to the Chinese patent application No. 202411375363.3, filed on September 29, 2024, and entitled "Method, device and system for generating virtual object", the whole content of which is incorporated herein by reference. TECHNICAL FIELD

[0002] The present application relates to the field of artificial intelligence, and more particularly, to a method, device and system for generating virtual object. BACKGROUND

[0003] The continuous progress of computer technology has driven the development of virtual object generation technology. For example, as an important branch of virtual object, digital human relies on virtualization, natural language processing (NLP), computer vision (CV) and other technologies to achieve a broad application prospect. Digital human, i.e., a virtual human figure with digital appearance, refers to a comprehensive product existing in a non-physical world, which needs to be created and used by a computer and has multiple human characteristics (e.g., appearance characteristics, human performance ability, interactive ability, etc.). Driven by digital technology, digital human can have human-like behaviors, including language, expression, and action expression capabilities. However, the current technology generates virtual objects based on general models, and the generation result is uncontrollable, resulting in poor generation quality of virtual objects.

[0004] Therefore, how to improve the generation quality of virtual objects is a problem to be solved. SUMMARY

[0005] The present application provides a method, device and system for generating virtual object, which can realize accurate calling of models and improve the generation quality of virtual objects.

[0006] In a first aspect, a method for generating virtual object is provided, which includes: obtaining user demand; determining generation demand according to the user demand, the generation demand being used for generating virtual object; decomposing the generation demand into at least one subtask, each subtask in the at least one subtask corresponding to a demand field, and each demand field including at least one demand function; determining a demand function corresponding to the virtual object under at least one demand field; calling a function model of the demand function corresponding to the virtual object under the at least one demand field to generate the virtual object.

[0007] Based on the above scheme, the generation requirement is multi-layered decomposed based on the demand field and the demand function, the refinement of the generation requirement can be realized, the accurate calling of the model is guaranteed, and a more accurate and higher quality virtual object is generated.

[0008] In combination with the first aspect, in some implementations of the first aspect, the determining the generation requirement according to the user requirement comprises: determining the generation requirement according to the user requirement by using a large language model (LLM) and a virtual object knowledge base.

[0009] Based on the above scheme, the rationality of the user requirement can be further judged by the LLM and the virtual object knowledge base, and the final generation requirement is determined. It should be noted that the LLM is a general model with general knowledge, but it may not have specific knowledge (for example, related knowledge of virtual objects). Therefore, the LLM can be added with specific knowledge by externally adding specific knowledge (for example, a virtual object knowledge base), so that the LLM does not need to be retrained.

[0010] In combination with the first aspect, in some implementations of the first aspect, the user requirement comprises a first feature; the determining the generation requirement according to the user requirement comprises: in a case where the virtual object knowledge base does not comprise the first feature, outputting a correction prompt, the correction prompt being used to prompt to modify the first feature to a second feature, wherein the second feature is a feature comprised in the virtual object knowledge base; acquiring a confirmation message of the correction prompt; and determining the generation requirement according to the confirmation message of the correction prompt, the generation requirement comprising the second feature.

[0011] Based on the above scheme, when the virtual object knowledge base does not comprise the first feature, it indicates that the virtual object involved in the user requirement is not within the scope of prior knowledge, and the user requirement is an “unreasonable” requirement. Therefore, the user requirement can be corrected according to the content of the virtual object knowledge base, and the generation requirement is obtained. The second feature involved in the generation requirement is in the virtual object knowledge base, and therefore the generation requirement is a “reasonable” requirement. Therefore, the above scheme can improve the rationality of the requirement, thereby improving the generation quality of the virtual object.

[0012] In combination with the first aspect, in some implementations of the first aspect, the determining the generation requirement according to the user requirement comprises: outputting a selection prompt, the selection prompt being used to prompt to select a third feature, wherein the third feature is a feature comprised in the virtual object knowledge base and not comprised in the user requirement; acquiring a confirmation message of the selection prompt; and determining the generation requirement according to the confirmation message of the selection prompt, the generation requirement comprising the third feature.

[0013] Based on the above scheme, the user demand can also be supplemented according to the virtual object knowledge base, the user is prompted to add reasonable features, the content of the generated demand is enriched, and the finally generated virtual object has higher quality.

[0014] With reference to the first aspect, in some implementations of the first aspect, before the generating demand is decomposed into at least one subtask, the method further includes: outputting a confirmation prompt, the confirmation prompt being used to prompt confirmation of the generating demand; and obtaining a confirmation message of the confirmation prompt.

[0015] Based on the above scheme, before the virtual object is generated, the user can also filter the features involved in the generating demand, thereby improving the controllability of the user side on the virtual object generation result.

[0016] With reference to the first aspect, in some implementations of the first aspect, the determining of the demand function corresponding to the virtual object in at least one demand field includes: determining the demand function corresponding to the virtual object in each demand field; presenting a selection interface, the selection interface including the demand function corresponding to the virtual object in each demand field; and obtaining the demand function corresponding to the virtual object in the at least one demand field selected through the selection interface.

[0017] The second aspect provides a virtual object generation device, which includes: an interaction unit configured to obtain a user demand; and a processing unit including an intelligent agent (Agent) and a function Agent.

[0018] The intelligent Agent is configured to: determine a generating demand according to the user demand, the generating demand being used to generate a virtual object; and decompose the generating demand into at least one subtask, each subtask in the at least one subtask corresponding to a demand field, and each demand field including at least one demand function.

[0019] The function Agent is configured to: determine a demand function corresponding to the virtual object in at least one demand field; and call a function model of the demand function corresponding to the virtual object in the at least one demand field to generate the virtual object.

[0020] With reference to the second aspect, in some implementations of the second aspect, the processing unit further includes a large language model (LLM) and a virtual object knowledge base.

[0021] The determining of the generating demand according to the user demand includes: the intelligent Agent determining the generating demand by using the LLM and the virtual object knowledge base according to the user demand.

[0022] The decomposing the generation requirement into at least one subtask comprises: the intelligent agent decomposing the generation requirement into at least one subtask by using the LLM.

[0023] With reference to the second aspect, in some implementations of the second aspect, the user requirement comprises a first feature.

[0024] The interaction unit is configured to: in a case where the virtual object knowledge base does not comprise the first feature, output a correction prompt, the correction prompt being used to prompt modification of the first feature to a second feature, wherein the second feature is a feature comprised in the virtual object knowledge base; and acquire a confirmation message of the correction prompt.

[0025] The intelligent agent is configured to: according to the confirmation message of the correction prompt, determine the generation requirement, the generation requirement comprising the second feature.

[0026] With reference to the second aspect, in some implementations of the second aspect, the interaction unit is configured to: output a selection prompt, the selection prompt being used to prompt selection of a third feature, wherein the third feature is a feature comprised in the virtual object knowledge base and not comprised in the user requirement; and acquire a confirmation message of the selection prompt.

[0027] The intelligent agent is configured to: according to the confirmation message of the selection prompt, determine the generation requirement, the generation requirement comprising the third feature.

[0028] With reference to the second aspect, in some implementations of the second aspect, the interaction unit is configured to: output a confirmation prompt, the confirmation prompt being used to prompt confirmation of the generation requirement; and acquire a confirmation message of the confirmation prompt.

[0029] With reference to the second aspect, in some implementations of the second aspect, the functional agent is configured to: determine a requirement function corresponding to the virtual object under each requirement field.

[0030] The interaction unit is configured to: present a selection interface, the selection interface comprising a requirement function corresponding to the virtual object under each requirement field; and acquire a requirement function corresponding to the virtual object under the at least one requirement field selected through the selection interface.

[0031] With reference to the second aspect, in some implementations of the second aspect, the interaction unit is further configured to: present the virtual object.

[0032] It should be understood that the beneficial effects of the second aspect and the various implementations of the second aspect are as described in the first aspect and the various implementations of the first aspect, and will not be described here.

[0033] In a third aspect, a computing device is provided, comprising a processor and a memory, and optionally, an input / output interface. The processor is configured to control the input / output interface to transmit and receive information. The memory is configured to store a computer program. The processor is configured to invoke and run the computer program stored in the memory, so as to execute the method in the first aspect or any possible implementation of the first aspect.

[0034] Optionally, the processor can be a general-purpose processor, which can be implemented by hardware or software. When implemented by hardware, the processor can be a logic circuit, an integrated circuit, or the like. When implemented by software, the processor can be a general-purpose processor, which reads software codes stored in the memory to implement the processor. The memory can be integrated in the processor or exist independently outside the processor.

[0035] In a fourth aspect, a computing device cluster is provided, comprising at least one computing device, each computing device comprising a processor and a memory. The processor of the at least one computing device is configured to execute instructions stored in the memory of the at least one computing device, so as to execute the method in the first aspect or any possible implementation of the first aspect.

[0036] In a fifth aspect, a chip is provided, which acquires instructions and executes the instructions to implement the method in the first aspect or any possible implementation of the first aspect.

[0037] Optionally, as an implementation, the chip comprises a processor and a data interface. The processor reads instructions stored in a memory through the data interface and executes the method in the first aspect or any possible implementation of the first aspect.

[0038] Optionally, as an implementation, the chip further comprises a memory, which stores instructions. The processor is configured to execute the instructions stored in the memory. When the instructions are executed, the processor is configured to execute the method in the first aspect or any possible implementation of the first aspect.

[0039] In a sixth aspect, a computer program product comprising instructions is provided. When the instructions are run by a computing device, the computing device is caused to execute the method in the first aspect or any possible implementation of the first aspect.

[0040] In a seventh aspect, a computer program product comprising instructions is provided. When the instructions are run by a computing device cluster, the computing device cluster is caused to execute the method in the first aspect or any possible implementation of the first aspect.

[0041] In an eighth aspect, a computer-readable storage medium is provided, including computer program instructions, when the computer program instructions are executed by a computing device, the computing device performs the method in the first aspect and any one of the implementation manners of the first aspect.

[0042] By way of example, such computer-readable storage can include one or more of the following: read-only memory (ROM), programmable ROM (PROM), erasable PROM (EPROM), Flash memory, electrically EPROM (EEPROM), and hard drive.

[0043] Optionally, as an implementation manner, the storage medium can be a non-volatile storage medium.

[0044] In a ninth aspect, a computer-readable storage medium is provided, including computer program instructions, when the computer program instructions are executed by a computing device cluster, the computing device cluster performs the method in the first aspect and any one of the implementation manners of the first aspect.

[0045] By way of example, such computer-readable storage can include one or more of the following: read-only memory (ROM), programmable ROM (PROM), erasable PROM (EPROM), Flash memory, electrically EPROM (EEPROM), and hard drive.

[0046] Optionally, as an implementation manner, the storage medium can be a non-volatile storage medium.

[0047] In a tenth aspect, a computing device cluster is provided, including at least one computing device, each computing device including a processor and a memory, the processor of the at least one computing device is configured to execute instructions stored in the memory of the at least one computing device, so that the computing device cluster performs the method in the first aspect and any one of the implementation manners of the first aspect. BRIEF DESCRIPTION OF DRAWINGS

[0048] FIG. 1 is a schematic diagram of a training process of reinforcement learning.

[0049] FIG. 2 is a schematic diagram of an architecture 200 of a virtual object generation system suitable for embodiments of the present application.

[0050] FIG. 3 is a schematic diagram of an architecture 300 of a virtual object generation system suitable for embodiments of the present application.

[0051] FIG. 4 is a schematic diagram of a method 400 of virtual object generation suitable for embodiments of the present application.

[0052] FIG. 5 is a schematic diagram of a method 500 of virtual object generation suitable for embodiments of the present application.

[0053] FIG. 6 is a flowchart 600 of a digital human generation suitable for embodiments of the present application.

[0054] FIG. 7 is a schematic block diagram of a virtual object generation apparatus 700 provided by embodiments of the present application.

[0055] FIG. 8 is a schematic diagram of an architecture of a computing device 800 provided by embodiments of the present application.

[0056] FIG. 9 is a schematic diagram of an architecture of a computing device cluster provided by embodiments of the present application.

[0057] FIG. 10 is a schematic diagram of a connection between computing devices 800A and 800B through a network provided by embodiments of the present application. DETAILED DESCRIPTION

[0058] The technical solutions in the present application will be described below with reference to the accompanying drawings.

[0059] Artificial intelligence (AI) is the use of 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. In other words, artificial intelligence is a branch of computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a similar way to human intelligence. Artificial intelligence is the design principle and implementation method of various intelligent machines, so that machines have the functions of perception, reasoning and decision-making. The research in the field of artificial intelligence includes robots, natural language processing, computer vision, decision-making and reasoning, human-computer interaction, recommendation and search, AI basic theory, etc.

[0060] The basic principle of AI is to combine massive data with super strong operation processing capability and intelligent algorithm to establish an AI model for solving specific problems, so that the AI model can automatically induce and learn potential patterns or features from data, thereby realizing a thinking mode close to human.

[0061] An AI model, also referred to as an AI algorithm (or AI operator), is a general term for a mathematical algorithm constructed based on the principle of artificial intelligence, and is also the basis for solving a specific problem by using AI. According to different specific methods and / or technologies for realizing artificial intelligence, the AI model can also be referred to as a machine learning model, a deep learning model, or a reinforcement learning model.

[0062] For the convenience of understanding the embodiments of the present application, first, the terms involved in the present application are briefly explained.

[0063] It should be understood that the relevant concept explanation may be limited due to the specific circumstances of the embodiments of the present application, but it does not mean that the present application can only be limited to this specific situation. The specific circumstances of different embodiments may also be different, which is not limited here.

[0064] 1. Neural network

[0065] The neural network is a model, and the neural network can be composed of neural units, and the neural unit can refer to an operation unit with X s and intercept 1 as input. The output of the operation unit can be:

[0066]

[0067] wherein s = 1, 2, … n, n is a natural number greater than 1, W s is the weight of X s , b is the bias of the neural unit. f is the activation function of the neural unit, which is used to introduce non-linear characteristics into the neural network and convert the input signal in the neural unit into an output signal. The output signal of the activation function can be used as the input of the next convolutional layer. The activation function can be a Relu function. The neural network is a network formed by connecting a plurality of the above single neural units, that is, the output of one neural unit can be the input of another neural unit, and the input of each neural unit can be connected with the local receptive field of the previous layer, thereby extracting the features of the local receptive field, wherein the local receptive field can be a region composed of a plurality of neural units.

[0068] Based on the above, the neural network can be specifically understood as being composed of neurons and having an input layer, a hidden layer, and an output layer. Generally, the first layer is the input layer, the last layer is the output layer, and the layers in between are the hidden layers. The input of the neural network module can be a single frame image or an image sequence. The neural network with many layers of hidden layers is referred to as a deep neural network (DNN). The work of each layer in the neural network can be expressed by a mathematical expression The working of each layer in a neural network can be understood as a transformation of an input space (a set of input vectors) to an output space (i.e., the row space of a matrix to the column space of a matrix) through five operations on the input space, including: 1, dimensionality increase / decrease; 2, scaling up / down; 3, rotation; 4, translation; and 5, "bending". The operations of 1, 2, and 3 are completed by W, the operation of 4 is completed by "+b", and the operation of 5 is completed by "a()". The term "space" is used here because the objects being classified are not individual things, but a class of things, and the space refers to the set of all individuals of this class. Here, W is the weight matrix of each layer of the neural network, and each value in the matrix represents the weight value of a neuron in the layer. The matrix W determines the spatial transformation of the input space to the output space described above, that is, W of each layer of the neural network controls how the space is transformed. The purpose of training a neural network is to obtain the weight matrix of all layers of the trained neural network. Therefore, the training process of a neural network is essentially learning the way to control the spatial transformation, and more specifically, learning the weight matrix.

[0069] 2, Large Model

[0070] A large model is used to provide a large model service. A large model refers to a neural network model containing a super large scale of parameters. The large model can also be referred to as a foundation model. The large model plays an important role in many fields and applications, and common application scenarios include natural language processing, computer vision, speech recognition and synthesis, recommendation systems, financial risk control, intelligent dialogue systems, games, artificial intelligence, and medical health, etc. In the future, the large model will evolve towards multi-modal, and the multi-modal large model can process data of multiple modalities, such as processing natural language, text, pictures, or videos, etc.

[0071] A large language model (LLM) is a type of large model, which is a neural network model with a large number of parameters trained on a large amount of data, and can understand and generate natural language text. Specifically, a large language model is usually based on neural network technology, and learns the grammar, semantics and context information of language by training on a large amount of text data. During the training process, the model continuously optimizes the parameters to improve the understanding and generation ability of the text. Due to the strong understanding ability of large language models for natural language, they have been widely used in many fields to solve natural language understanding and generation problems. Large language models have a wide range of applications in the field of artificial intelligence, such as natural language processing, machine translation, dialogue systems, etc.

[0072] 3, Prompt

[0073] Prompt is an interaction method based on natural language processing. It enables communication between users and machines by parsing natural language. The main implementation of prompt is to convert natural language into machine-recognizable instructions by building a corresponding corpus and semantic parsing model. Prompt can generate finer-grained tasks by piecing together prompts from the original text based on context, thought processes, or cue words, guiding larger models to better complete tasks.

[0074] The design principles of prompts include: writing clear and specific instructions, and giving the model time to think.

[0075] The key design elements of a prompt include: Instruction, Context, Constraint, Example, and Input. Instruction provides the task instructions and states the specific goals. Context allows the AI ​​(or large model) to understand the task from the user's perspective and provides context around the task. Constraint specifies the steps to complete the task (e.g., a chain of thought, COT), clearly defines the language style, and specifies the output template. Example provides concrete examples. It should be noted that this division of design elements in a prompt is merely an example; other division methods are not limited.

[0076] 4. Artificial Intelligence Agent (AI Agent)

[0077] An AI Agent is an intelligent entity capable of perceiving its environment, making decisions, and executing actions. Typically based on machine learning and artificial intelligence technologies, it possesses autonomy and adaptability, enabling it to learn and improve autonomously in specific tasks or domains. AI Agents can not only quickly search for relevant information based on user needs but also leverage their learning capabilities to provide increasingly personalized and thoughtful solutions. In an LLM-based autonomous agent system (LLM Agent), the LLM acts as the agent's brain, collaborating with several key components. Specifically, the task execution process of an LLM Agent can include planning, memory, tool use, and action.

[0078] (1) Planning

[0079] Sub-goal decomposition: The agent breaks down large tasks into smaller, manageable sub-goals, thus enabling it to effectively handle complex tasks.

[0080] Reflection and Improvement: Agents can self-criticize and reflect on their past actions, learn from mistakes and improve in subsequent steps, thereby improving the quality of the final result.

[0081] (2) Memory

[0082] Short-term memory: Contextual learning is short-term memory learning that utilizes models.

[0083] Long-term memory: Provides agents with the ability to retain and recall long-term information, which can be achieved using external vector storage and retrieval.

[0084] (3) Tool usage

[0085] For information about missing model weights, the agent learns to call external APIs to obtain additional information, including current information, code execution capabilities, and access to proprietary information sources.

[0086] (4) Action

[0087] The action module is the part of the intelligent agent that actually executes decisions or responses. Faced with different tasks, the intelligent agent system has a complete set of action strategies, allowing it to select the actions to be performed during decision-making. Examples include memory retrieval, reasoning, learning, and programming.

[0088] 5. Reinforcement learning (RL)

[0089] RL, also known as reinforcement learning, evaluative learning, or enhancement learning, is used to describe and solve the problem of agents learning strategies to maximize rewards or achieve specific goals during their interactions with the environment.

[0090] Reinforcement learning is a learning process where an agent learns through trial and error. Rewards obtained from interactions with the environment guide behavior, with the goal of maximizing the agent's reward. Reinforcement learning does not require a training dataset. Instead of telling the reinforcement learning system how to produce the correct action, the reinforcement signals (rewards) provided by the environment evaluate the quality of the action. Because the external environment provides limited information, the agent must learn through its own experience. In this way, the agent acquires knowledge in the action-reward environment and improves its actions to adapt to the environment.

[0091] Figure 1 is a schematic diagram of the training process of reinforcement learning.

[0092] As shown in FIG. 1, reinforcement learning mainly includes four elements: an agent, an environment state, an action, and a reward, wherein the input of the agent is the state and the output is the action.

[0093] In the current technology, the training process of reinforcement learning is: through multiple interactions between the agent and the environment, the action, state, and reward of each interaction are obtained; the multiple sets of information (action, state, and reward) are used as training data to train the agent once. The above process is used to train the agent for the next round until the convergence condition is met.

[0094] Among them, the process of obtaining the action, state, and reward of one interaction is shown in FIG. 1. The current state s(t) of the environment is input to the agent to obtain the action a(t) output by the agent. The reward r(t) of this interaction is calculated according to the relevant performance indicators of the environment under the action a(t). At this point, the action a(t), state s(t), and reward r(t) of this interaction are obtained. The action a(t), state s(t), and reward r(t) of this interaction are recorded for subsequent use to train the agent. The next state s(t+1) of the environment under the action a(t) is also recorded to facilitate the next interaction between the agent and the environment.

[0095] Among them, the agent refers to an entity that can think and interact with the environment. For example, the agent can be a computer system or a part of a computer system in a certain environment. The agent can perceive the environment according to its own perception, follow existing instructions or through autonomous learning, and communicate and cooperate with other agents to autonomously complete the set goals in the environment. The agent can be a software or a combination of software and hardware entity.

[0096] 6、Multi-modal technology

[0097] As a subfield of AI, multi-modal technology is used to realize the simultaneous processing of multiple types of data. Different data types can become different modalities. For example, multi-modal data can include text, images, audio, video, and sensor data, etc. Multi-modal technology aims to better understand and interpret complex real-world scenarios, providing users with more rich and three-dimensional information presentation.

[0098] 7、Digital human technology

[0099] A digital human is a virtual human avatar with a digitalized appearance, which exists in a non-physical world, is created and used through a computer, and has multiple human characteristics (e.g., appearance characteristics, human performance capabilities, interaction capabilities, etc.). Driven by digitalization technology, a digital human can have human-like behaviors, including language, expressions, and action expression capabilities. For example, a realistic virtual human can be generated by simulating human voice, facial expressions, and body movements through AIGC technology. According to the dimensions of the task image, a digital human can be divided into 2D cartoon digital human, 2D real person digital human, 3D cartoon digital human, and 3D hyper-realistic digital human, etc.

[0100] The above describes related terms related to the embodiments of the present application. The system architecture applicable to the embodiments of the present application is described below with reference to FIGS. 3 and 4.

[0101] FIG. 2 is a schematic diagram of an architecture 200 of a virtual object generation system applicable to the embodiments of the present application.

[0102] As shown in FIG. 2, the data acquisition device 260 is configured to acquire training data and store the training data in the database 230. The training device 220 trains a target model / rule 201 based on the training data maintained in the database 230.

[0103] It should be noted that the data acquisition device 260 and the execution device 210 and the training device 220 can be the same or different devices.

[0104] The training device 220 how to obtain the target model / rule 201 based on the training data will be described in more detail below. The target model / rule 201 can be used to implement the image processing method provided by the embodiments of the present application.

[0105] It should be noted that in actual applications, the training data maintained in the database 230 can come from the data acquisition device 260 or from other devices. The training device 220 does not necessarily train the target model / rule 201 based entirely on the training data maintained in the database 230. It can also obtain training data from the cloud or other places for model training. The above description should not be considered as a limitation of the embodiments of the present application. The target model / rule 201 trained by the training device 220 can be applied to different systems or devices, for example, the execution device 210 shown in FIG. 2.

[0106] It should be understood that the execution device 210 can be a terminal. Illustratively, the terminal can be, for example, a digital camera, a surveillance camera, a mobile phone, a personal computer (PC), a notebook computer, a server, a tablet computer, a smart television, a vehicle terminal, a mobile internet device (MID), a wearable device, a virtual reality (VR) device, an augmented reality (AR) device, a wireless terminal in industrial control, a wireless terminal in self driving, a wireless terminal in remote medical surgery, a wireless terminal in a smart grid, a wireless terminal in transportation safety, a wireless terminal in a smart city, a wireless terminal in a smart home, and the like, and can also be an edge device (for example, a box carrying a chip with processing capability) and the like; the execution device 210 can also be a server or a cloud and the like.

[0107] The execution device 210 is configured with an I / O interface 212 for data interaction with an external device, and a user can input data to the I / O interface 212 through a client device 240. In addition, the input data can be input by the user, can also be uploaded by the user through a shooting device, or can come from a database, which is not limited in the present application.

[0108] As a possible embodiment, the execution device 210 and the training device 220 are different processors deployed on different physical devices, such as servers or servers in a cluster. For example, the execution device 210 can be a graphic processing unit (GPU), a central processing unit (CPU), other general-purpose processor, a digital signal processing (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The training device 220 can be a graphics processing unit (GPU), a neural network processing unit (NPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the program execution of the scheme of the present application. The training device 220 can configure the trained neural network to the plurality of execution devices 210. Each execution device 210 implements the function of super-resolution reconstruction, etc. by using the trained neural network.

[0109] In another possible embodiment, the execution device 210 and the training device 220 are deployed on the same physical device, or the execution device 210 and the training device 220 are the same physical device. The computing device can configure the trained neural network to itself, and implement the function of super-resolution reconstruction, etc. by using the trained neural network.

[0110] The training device 220 is configured to train the neural network by using the training data until the loss function in the neural network converges, and the loss function value is less than a certain threshold, so that the neural network training is completed, so that the neural network reaches a certain accuracy. Or, all the training data in the database 230 are used for training, and then the neural network training is completed, so that the trained neural network has the function of super-resolution reconstruction, etc. Further, the training device 220 configures the trained neural network to the execution device 210. The execution device 210 is configured to implement the function of processing application data according to the trained neural network.

[0111] The preprocessing module 213 is configured to perform preprocessing on the input data received by the I / O interface 212. During the preprocessing of the input data performed by the execution device 210, or during the processing performed by the computing module 211 of the execution device 210, the execution device 210 can call data, codes, etc. in the data storage system 250 for corresponding processing, and can also store the data, instructions, etc. obtained by the corresponding processing in the data storage system 250. Finally, the I / O interface 212 returns the processing result to the client device 240, thereby providing the user.

[0112] It should be understood that FIG. 2 is only a schematic diagram of a system architecture, and the positional relationship between the devices, components, modules, etc. shown in the figure does not constitute any limitation. For example, in FIG. 2, the data storage system 250 is an external memory relative to the execution device 210, and in other cases, the data storage system 250 can also be placed in the execution device 210.

[0113] FIG. 3 is a schematic diagram of an architecture 300 of a virtual object generation system suitable for embodiments of the present application.

[0114] As shown in FIG. 3, the architecture 300 includes a model layer, an Agent layer, and a user layer. The model layer can include an LLM and at least one functional model, the Agent layer can include at least one AI Agent and at least one functional Agent, and the user layer is configured to implement information interaction between the user and the Agent layer and the user and the model layer.

[0115] In the requirement generation phase, the user requirements can be obtained through the user layer, and the final requirements can be determined through the interaction of the user layer, the Agent layer, and the model layer. In the requirement decomposition and virtual object generation phase, the requirement decomposition can be performed through the model layer and the Agent layer, and the final virtual object generation can be implemented through the model layer.

[0116] It should be understood that the system architecture and business scenarios described in the embodiments of the present application are for more clearly illustrating the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application. Those skilled in the art can know that, with the evolution of system architecture and the appearance of new business scenarios, the technical solutions provided by the embodiments of the present application are also applicable to similar technical problems.

[0117] With the development of AI technology, virtual object generation technology has become increasingly mature and popular in recent years. Taking digital human generation technology as an example, the platform service provided by a digital human manufacturer can support users to generate virtual characters using various forms of input. The digital human generation scheme on the platform side mainly includes the following two kinds:

[0118] (1) User uploads voice, selects pre-made model / upload video, interface selects clothing, hair, skin color, etc. human information to generate digital human video with action and speech. The disadvantage of this scheme is that the user operation steps are fixed and cumbersome, and the operation freedom is low, and the types of generated digital humans are limited.

[0119] (2) User uses natural language to control the generation of digital human through dialogue, including generated voice, lip shape, action, clothing, hair, skin color, etc. Based on this way, the user can generate the desired digital human by interacting with the system using natural language. Through the ability of the language model, the user's natural language is analyzed, and after analyzing the user's command, the generation task of the digital human is completed by calling a special model (for example, a dressing model generates clothing, a lip shape model is responsible for lip shape tasks, an action model is responsible for action generation, and a voice model is responsible for analyzing voice). The final generation of the digital human is completed by combining the results of multiple models. The disadvantage of this scheme is that the quality of the generated digital human depends strictly on the accuracy of the user interface input, and the user cannot control the generation of the details of the digital human, resulting in weak controllability of the digital human generation process. The use of general models to generate digital humans cannot guarantee the effect.

[0120] For example, the user requires the generation of a digital human with blue skin, which is obviously not a reasonable digital human image. The digital human model has not learned the experience of a digital human with blue skin, and the generation quality is poor. For another example, the user cannot control whether to generate voice, and if the prompt is not specified, the user cannot control the type of generated voice, and the process of generating voice is uncontrollable. For another example, the user wants to generate a digital human wearing complex clothing, but due to the use of a general dressing model, the generation effect is poor.

[0121] As can be seen from the above, in the current multi-modal dialogue digital human generation system, the user can generate the desired digital human by interacting with the system using natural language. Through the LLM, the user's natural language is analyzed, and after analyzing the user's command, the generation task of the digital human is completed by calling a special model, and the final generation of the digital human is completed by combining the results of multiple special models.

[0122] In this case, the generation result of the virtual object is realized based on a general model, and the generation result is uncontrollable, resulting in poor generation quality of the virtual object. In addition, the generation of the virtual object depends on the input information of the user, and unreasonable input information will lead to the generation of unreasonable virtual objects, thereby reducing the generation quality of the virtual object. In addition, since the user level control cannot control the generation of virtual object details, the controllability of the virtual object generation process is weak.

[0123] Therefore, the virtual object generation method provided by the present application can realize accurate calling of the model, thereby improving the generation quality of the virtual object. In addition, in the process of generating the virtual object, the demand optimization can be realized through the demand rationality judgment, thereby further improving the generation quality of the virtual object.

[0124] The technical solutions provided by the present application will be described in detail below with reference to the accompanying drawings.

[0125] FIG. 4 is a schematic diagram of a virtual object generation method 400 suitable for the embodiments of the present application.

[0126] It should be understood that FIG. 4 shows the steps or operations of the virtual object generation method, but these steps or operations are only to help those skilled in the art to understand the embodiments of the present application, and are not intended to limit the embodiments of the present application to the specific values or specific scenarios shown in FIG. 4. Those skilled in the art can obviously make various equivalent modifications or changes according to the examples given below in FIG. 4, and such modifications and changes also fall within the scope of the embodiments of the present application. The embodiments of the present application can also perform other operations or variations of the operations in FIG. 4.

[0127] It should be understood that the embodiments shown below do not particularly limit the specific structure of the execution subject of the method provided by the embodiments of the present application, as long as the execution subject can perform virtual object generation according to the method provided by the embodiments of the present application by running the program in which the code of the method provided by the embodiments of the present application is recorded. For example, the method provided by the embodiments of the present application can be executed by a processing device, which can be the execution device in FIG. 2, or a component (such as a processor, a chip, or a chip system, etc.) of the execution device. The execution device can be a cloud device, a terminal device (such as a mobile phone), etc. Of course, the method provided by the embodiments of the present application can also be executed by a system including a cloud device and a terminal device.

[0128] Optionally, the method can be processed by the CPU in the execution device, or can be processed by the CPU and the GPU together, or can not use the GPU, but use other processors suitable for neural network calculation, which is not limited by the present application.

[0129] As shown in FIG. 4, the method 400 can include the following steps:

[0130] S401: Obtain a user demand.

[0131] Optionally, the user demand is proposed by the user in a direct or indirect manner, for generating a first virtual object. For example, the user demand can be directly input by the user, or can be a result obtained after processing the demand input by the user, which is not limited by the present application.

[0132] S402: determining a generation requirement according to the user requirement, the generation requirement being used to generate a virtual object.

[0133] It should be understood that the user requirement can be the same as the generation requirement, or can also be different from the generation requirement. Therefore, the first virtual object can be the same as the virtual object, or can also be different from the virtual object. The present application does not limit this.

[0134] Optionally, the determining the generation requirement according to the user requirement comprises: determining the generation requirement according to the user requirement, by using a large language model (LLM) and a virtual object knowledge base.

[0135] It should be noted that the LLM is a general model with general knowledge, but can not have specific knowledge (for example, related knowledge of virtual objects). Therefore, the LLM can be added with specific knowledge by externally adding specific knowledge (for example, a virtual object knowledge base), so as to avoid retraining the LLM.

[0136] In this case, the rationality of the user requirement can be further judged by the LLM and the virtual object knowledge base to determine the final generation requirement.

[0137] For example, step S402 can be implemented based on an intelligent agent. Specifically, the intelligent agent can use the prior knowledge of the virtual object knowledge base, call the LLM to filter and correct the user requirement, and return the corrected requirement to the user.

[0138] Optionally, the user requirement includes a first feature; and the determining the generation requirement according to the user requirement comprises: in a case where the virtual object knowledge base does not include the first feature, outputting a correction prompt, the correction prompt being used to prompt to modify the first feature to a second feature, wherein the second feature is a feature included in the virtual object knowledge base; obtaining a confirmation message of the correction prompt; and determining the generation requirement according to the confirmation message of the correction prompt, the generation requirement including the second feature.

[0139] In this case, when the virtual object knowledge base does not include the first feature, it indicates that the virtual object involved in the user requirement is not within the scope of the prior knowledge, and the user requirement is an "unreasonable" requirement. Therefore, the user can be prompted to correct the user requirement according to the content of the virtual object knowledge base to obtain the generation requirement. The second feature involved in the generation requirement is in the virtual object knowledge base, and therefore the generation requirement is a "reasonable" requirement. Therefore, the above scheme can improve the rationality of the requirement, thereby improving the generation quality of the virtual object.

[0140] It should be noted that the above step of determining the "reasonable generation requirement" can be performed at least once. In other words, the method 400 can also interact with the user multiple times based on the user's demand to ensure the rationality of the generation requirement.

[0141] Optionally, the determining the generation requirement according to the user requirement comprises: outputting a selection prompt, the selection prompt being used to prompt selection of a third feature, wherein the third feature is a feature included in the virtual object knowledge base and not included in the user requirement; obtaining a confirmation message of the selection prompt; and determining the generation requirement according to the confirmation message of the selection prompt, the generation requirement including the third feature.

[0142] In this case, the user requirement can also be supplemented according to the virtual object knowledge base, the user is prompted to add reasonable features, and the content of the generation requirement is enriched, so that the finally generated virtual object has higher quality.

[0143] It should be noted that the above step of determining the "supplemental generation requirement" can be performed at least once. In other words, the method 400 can also interact with the user multiple times based on the user's demand to expand the user's demand and obtain a generation requirement with rich content.

[0144] S403: decompose the generation requirement into at least one subtask, each subtask in the at least one subtask corresponding to a requirement field, and each requirement field including at least one requirement function.

[0145] For example, step S403 can be implemented based on an intelligent Agent and an LLM. Specifically, the intelligent Agent can be used to instruct the LLM to decompose the generation requirement to obtain at least one subtask, and guide the functional Agent to perform subtask decomposition based on the decomposed subtask.

[0146] Optionally, the requirement field and the requirement function are implemented based on a multi-layer architecture. That is, the features of the virtual object can be classified and layered according to different fields (or dimensions), and each field can be further divided into a multi-layer structure corresponding to different requirement functions. For example, the mapping relationship between the requirement field and the requirement function can be determined based on an Agent (or the mapping relationship between the requirement field and the requirement function is stored by the Agent); or the mapping relationship between the requirement field and the requirement function can be determined based on a list (or the mapping relationship between the requirement field and the requirement function is stored by the list), or the mapping relationship between the requirement field and the requirement function can be determined in other ways, which are not limited by the embodiments of the present application.

[0147] Exemplarily, for the digital human generated scene, the demand field can be the different dimensional features constituting the digital human image. For example, the demand field can include action, skin color, voice, and / or dress-up, etc. The demand function can be the subdivided features constituting each demand field. For example, the demand function corresponding to the "action" in the demand field can include running, dancing, and / or walking, etc.; the demand function corresponding to the "skin color" in the demand field can include yellow skin, white skin, and / or black skin, etc.; the demand function corresponding to the "voice" in the demand field can include Chinese voice and / or English voice, etc.; the demand function corresponding to the "dress-up" in the demand field can include men's clothing and / or women's clothing, etc.

[0148] It should be noted that the number of layers of the multi-layer architecture is not limited by the embodiments of the present application. Exemplarily, the demand function "men's clothing" can further include the demand functions of shirt, T-shirt, and / or long pants, etc.; and the demand function "women's clothing" can further include the demand functions of dress and / or skirt, etc.

[0149] It should be understood that the above examples are only illustrated by taking the digital human generated scene as an example, and do not constitute a limitation on the content of the virtual object. For example, the virtual object applicable to the present application can also be a virtual plant or a virtual article, etc., which is not limited by the present application.

[0150] Optionally, the mapping relationship between the demand field and the demand function can be obtained by training the virtual object knowledge base.

[0151] Optionally, before the generation demand is decomposed into at least one subtask, the method further includes: outputting a confirmation prompt, the confirmation prompt being used to prompt confirmation of the generation demand; and obtaining a confirmation message of the confirmation prompt.

[0152] In this case, before the virtual object is generated, the user can also filter the features involved in the generation demand, thereby improving the controllability of the user side on the virtual object generation result.

[0153] S404: determining the demand function corresponding to the virtual object under at least one demand field.

[0154] Optionally, after the demand function corresponding to the generation demand is determined, the user can filter the demand function, thereby finally determining the demand function involved in the virtual object.

[0155] For example, the determining of the demand function corresponding to the virtual object in the at least one demand field comprises: determining the demand function corresponding to the virtual object in each demand field; presenting a selection interface including the demand function corresponding to the virtual object in each demand field; and obtaining the demand function corresponding to the virtual object in the at least one demand field selected through the selection interface.

[0156] S405: calling a function model of the demand function corresponding to the virtual object in the at least one demand field to generate the virtual object.

[0157] In this case, the multi-layer decomposition of the generated demand based on the demand field and the demand function can realize the refinement of the generated demand, guarantee the accurate calling of the model, and thus generate a more accurate and higher-quality virtual object.

[0158] FIG. 5 is a schematic diagram of a virtual object generation method 500 suitable for the embodiments of the present application.

[0159] It should be noted that the embodiments of the present application take a digital human as an example for illustration, and do not constitute a limitation on the content of the virtual object. For example, the virtual object suitable for the present application can also be a virtual plant or a virtual article, and the present application does not limit this.

[0160] As shown in (a) of FIG. 5, the method 500 can include two steps of “digital human demand generation” and “digital human demand decomposition and digital human generation”, which will be described in detail below.

[0161] Step 1: Digital human demand generation

[0162] As shown in (b) of FIG. 5, the “digital human demand generation” step involves the interaction among the user layer, the Agent layer and the model layer. The user layer can include a user dialogue box 510, the Agent layer can include an intelligent Agent 520 and a digital human knowledge base 521, and the model layer can include an LLM 530.

[0163] User dialogue box 510: used for obtaining user demand and realizing information interaction between the digital human generation system and the user. It should be understood that the embodiments of the present application do not limit the specific form of the user dialogue box 510, i.e., the user can realize information interaction with the Agent layer and the model layer through text, voice or other means based on the user dialogue box 510.

[0164] Intelligent Agent 520: the intelligent Agent 520 can utilize the prior knowledge of the digital human knowledge base 521 to call the LLM 530 to filter and correct the user demand, and return the corrected demand to the user through the user dialogue box 510.

[0165] It should be noted that after the user dialog box 510 obtains the original demand of the user, the user dialog box 510, the intelligent Agent 520 and the LLM 530 can generate the final demand through multiple iterations.

[0166] Digital human knowledge base 521: includes knowledge related to digital humans as prior information, used to assist the LLM 530 in determining whether the user's demand is reasonable, and to ensure that the user presents accurate and reasonable digital human generation demand.

[0167] For example, the digital human knowledge base 521 includes the actions, appearance, skin color and other human features that a "reasonable digital human" should have, and the LLM 530 can determine whether the user's demand is reasonable according to the prior knowledge in the digital human knowledge base 521.

[0168] LLM 530: used to analyze the demand presented by the user through natural language. It should be understood that for the specific process of LLM analyzing natural language, reference can be made to the related description of the prior art, which will not be repeated here.

[0169] It should be understood that "digital human demand generation" is used to guide the user to generate accurate and reasonable digital human generation demand, the intelligent Agent 520 can interact with the user in both directions, and the LLM 530 can also be used to correct the user's digital human generation demand and finally generate a reasonable demand.

[0170] In one possible implementation, after the user dialog box 510 obtains the original demand of the user (demand #1), the demand #1 is passed to the intelligent Agent 520, and the intelligent Agent 520 uses the LLM 530 to understand the content of the demand #1. Specifically, the demand #1 is analyzed by the LLM 530, and the LLM 530 can further judge the rationality of the demand #1 based on the prior information of the digital human knowledge base 521 (i.e. whether the digital human to be generated by the demand #1 belongs to the category of the digital human knowledge base 521).

[0171] Optionally, when the LLM 530 judges that the demand #1 is an unreasonable demand, the demand #1 can be corrected according to prior information to obtain a demand #2, and the demand #2 is returned to the user through the intelligent Agent 520 and the user dialogue box 510. It should be noted that the above process can be repeated multiple times, and the number of interactions between the user dialogue box 510, the intelligent Agent 520, the digital human knowledge base 521 and the LLM 530 is not limited by the embodiments of the present application, as long as the final generated demand (demand #n) is a reasonable demand. That is, the user can also determine a new demand #3 according to the demand #2, so that the user dialogue box 510 can pass the demand #3 to the intelligent Agent 520, and the intelligent Agent 520 further understands the content of the demand #3 by using the LLM 530. The above scheme can ensure the rationality of the final generated demand, and solves the problem of uncontrollable generation results caused by unclear user-side demand.

[0172] Optionally, when the LLM 530 judges that the demand #1 is a reasonable demand, the judgment result can also be returned to the intelligent Agent 520, and the related content of the demand #1 is presented to the user through the user layer, so as to further confirm or update the demand #1.

[0173] Optionally, the LLM 530, the digital human knowledge base 521 and the intelligent Agent 520 can also present other features that the "reasonable digital human" can have to the user as prompt information, so as to enrich the final generated demand.

[0174] Step two: digital human demand decomposition and digital human generation

[0175] As shown in (c) of FIG. 5, the "digital human demand decomposition and digital human generation" step also involves the interaction between the user layer, the Agent layer and the model layer. The user layer can include the function interface 511, the Agent layer can include the intelligent Agent 520 and the function Agent 522, and the model layer can include the LLM 530 and the function model 531.

[0176] It should be understood that the "digital human demand decomposition and digital human generation" is used to decompose the generation demand of the digital human, and generate a corresponding digital human based on the decomposed generation demand. Specifically, the LLM 530 and the function Agent 522 are used for layer-by-layer decomposition of the demand, realizing demand decoupling, so as to generate a digital human with higher quality.

[0177] Intelligent Agent 520: can also be used to instruct the LLM 530 to decompose the final demand of the digital human generation generated in step one, to obtain at least one subtask, and guide the function Agent 522 to decompose the subtask based on the decomposed subtask.

[0178] LLM 530: is also configured to receive the instruction sent by the intelligent Agent 520, and return the final demand decomposed into a plurality of sub-tasks according to the instruction to the intelligent Agent 520.

[0179] The functional Agent 522 is implemented based on a multi-layer Agent architecture, and can be composed of at least one functional Agent module. Specifically, the functional Agent 522 classifies and layers the features of the digital human according to different dimensions (or fields), further decomposes the sub-tasks indicated by the intelligent Agent 520, determines the underlying functional Agent involved in the digital human generating the final demand, and thus calls the corresponding functional model 531.

[0180] It should be noted that the specific structure of the functional Agent 522 can be obtained by training the digital human knowledge base 521.

[0181] Optionally, the functional model 531 can have at least one layer of layer structure, and the number of layers of the functional model 531 is not limited in the embodiments of the present application.

[0182] For example, when the digital human knowledge base 521 includes the relevant information of a person running, a person dancing, a person speaking Chinese, a person wearing male clothes, and a person wearing female clothes, the structure of the functional Agent 522 obtained by training the digital human knowledge base 521 can have a two-layer structure, including a dressing Agent, an action Agent, and a voice Agent. The dressing Agent includes a male clothes Agent and a female clothes Agent; the action Agent includes a dance Agent and a running Agent; and the voice Agent includes a Chinese voice Agent.

[0183] The functional model 531 is configured to call the corresponding model to realize the generation of the digital human based on the indication of the functional Agent 522. The functional model 531 can include at least one functional model, which corresponds to the underlying functional Agent in the functional Agent 522.

[0184] For example, corresponding to the male clothes Agent, the female clothes Agent, the dance Agent, the running Agent, and the Chinese voice Agent mentioned above, the functional model 531 can include a male clothes model, a female clothes model, a dance model, a running model, and a Chinese voice model.

[0185] The functional interface 511 is configured to present the underlying functional Agent obtained by the functional Agent 522 decomposing the sub-tasks indicated by the intelligent Agent 520 to the user, and the user can further filter the underlying functional Agent to generate the final digital human.

[0186] In a possible implementation, after the intelligent Agent 520 obtains the final requirement (requirement #n) of the user, it sends an instruction to decompose the requirement #n to the LLM 530. After receiving the instruction, the LLM 530 decomposes the requirement #n into multiple sub-tasks and returns the decomposition result to the intelligent Agent 520. After receiving the sub-tasks decomposed from the requirement #n, the intelligent Agent 520 instructs the functional Agent 522 to further decompose the sub-tasks and determine the underlying functional Agent involved in the requirement #n. Compared with the existing scheme of directly calling the corresponding model based on the decomposition of the LLM, the further decomposition of the requirement by the functional Agent 531 and the determination of the underlying functional model 531 based on the decomposition result of the functional Agent 531 can improve the accuracy of model selection and significantly improve the generation quality of the digital person.

[0187] Optionally, the functional Agent 522 can present the underlying functional Agent related to the requirement #n to the user through the functional interface 511, and the user can further screen the underlying functional Agent to determine the underlying functional Agent involved in the finally generated digital person, thereby effectively improving the controllability of the user side on the generation result of the digital person.

[0188] It should be understood that after the functional Agent 522 determines the underlying functional Agent involved in the finally generated digital person, it calls the corresponding functional model 531 to generate the digital person and presents the final generation result to the user layer.

[0189] FIG. 6 is a digital person generation flowchart 600 suitable for the embodiments of the present application.

[0190] For ease of understanding, only the digital human generation system shown in FIG. 6 is taken as an example for description. The digital human knowledge base includes information of a person with white skin, a person with yellow skin, a person with black skin, a dancing person, a running person, a person speaking Chinese, a person wearing a short skirt, and a person wearing a dress, and the like, so that the functional Agents trained according to the digital human knowledge base include skin color Agents, dressing Agents, action Agents, and speech Agents. The skin color Agents include white skin Agents, yellow skin Agents, and black skin Agents; the dressing Agents include dress Agents and short skirt Agents; the action Agents include dancing Agents and running Agents; and the speech Agents include Chinese speech Agents. Correspondingly, the functional models include white skin models, yellow skin models, black skin models, dress models, short skirt models, dancing models, running models, and Chinese speech models. It should be understood that the content of the above-mentioned digital human knowledge base, the composition of the functional Agents, and the composition of the functional models are only for the purpose of more clearly describing the generation process of the digital human, and do not constitute a limitation on the embodiments of the present application. In the technical solutions of the present application, the digital human knowledge base can also contain more information, and correspondingly, the functional Agents and the functional models can also include more content.

[0191] As shown in FIG. 6, the original requirement input by the user is: "generate a digital human with orange skin dancing".

[0192] In the digital human requirement generation stage, since the prior information in the digital human knowledge base does not include the related content of orange skin, the LLM will judge that the original requirement is an "unreasonable" requirement based on the prior information in the digital human knowledge base, and return the judgment result to the intelligent Agent. The intelligent Agent calls the LLM to correct the original requirement, and then feeds back to the user through the user dialog box: "orange skin is not human skin, do you want to generate a digital human with yellow skin".

[0193] After the user updates the requirement (for example, generates a digital human with yellow skin), the intelligent Agent can further confirm with the user the content of "female dress or male dress", "dress or short skirt", and "whether to use Chinese speech" based on the information of the digital human knowledge base. After multiple interactions between the user and the digital human generation system, it is confirmed that the digital human generation requirement is: "please generate a girl with yellow skin, Chinese speech, dancing and wearing a short skirt".

[0194] In the digital human demand decomposition and digital human generation stage, through decomposition of the LLM, the digital human generation demand involves the skin color agent, the action agent, the voice agent and the dressing agent in the functional agent, and through decomposition of the functional agent, the digital human generation demand involves the yellow skin agent, the dancing agent, the Chinese voice agent and the skirt agent in the underlying functional agent. The functional agent can further confirm the demand of the user through the functional interface, that is, the user can screen the result of the decomposition of the functional agent, so as to obtain the final digital human generation demand (for example, the functional agent presents the decomposed agents including the yellow skin agent, the dancing agent, the Chinese voice agent and the skirt agent to the user through the functional interface, and the Chinese voice agent is removed after being screened by the user, so that the final digital human generation demand only involves the yellow skin agent, the dancing agent and the skirt agent). After confirming the final digital human generation demand, the functional agent calls the corresponding functional model (for example, the skirt model, the dancing model and the yellow skin model), generates the final digital human image, and presents the generation result to the user layer.

[0195] In order to facilitate understanding of the above-mentioned embodiments provided by the present application, the following points are explained.

[0196] (1) In the embodiments of the present application, "indication" can include direct indication, indirect indication, display indication and implicit indication. When describing that certain indication information is used to indicate A, it can be understood that the indication information carries A, directly indicates A, or indirectly indicates A.

[0197] In the present application, the information indicated by the indication information is referred to as the to-be-indicated information. In the specific implementation process, there are many ways to indicate the to-be-indicated information, for example but not limited to, the to-be-indicated information can be directly indicated, such as the to-be-indicated information itself or the index of the to-be-indicated information. The to-be-indicated information can also be indirectly indicated by indicating other information, where the other information and the to-be-indicated information have an association relationship. The to-be-indicated information can also be only indicated in part, and the other part of the to-be-indicated information is known or agreed in advance. For example, the indication of a specific information can also be realized by means of the arrangement order of each information agreed in advance (for example, the protocol stipulates), thereby reducing the indication overhead to a certain extent. In addition, the to-be-indicated information can be sent as a whole, or can be sent separately in multiple sub-information, and the sending period and / or sending time of these sub-information can be the same or different.

[0198] (2) In the present application, "sending" and "receiving" represent the direction of signal transmission. For example, "sending information to XX" can be understood as that the destination of the information is XX, which can include direct sending through the air interface, or indirect sending through the air interface by other units or modules. "Receiving information from YY" can be understood as that the source of the information is YY, which can include direct receiving from YY through the air interface, or indirect receiving from YY through the air interface by other units or modules. "Sending" can also be understood as the "output" of the chip interface, and "receiving" can also be understood as the "input" of the chip interface. In other words, sending and receiving can be carried out between devices, such as between network devices and terminal devices, or can be carried out within a device, such as between components, modules, chips, software modules or hardware modules within a device through a bus, wire or interface.

[0199] (3) In various embodiments of the present application, the terms and / or descriptions of different embodiments are consistent and can be referred to each other if there is no special description and logical conflict, and the technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationship.

[0200] (4) In the present application, "first", "second" are only for convenience of description, used to distinguish objects, and do not limit the scope of the embodiments of the present application. They are not used to describe the order or sequence of features. It should be understood that the objects thus described can be interchanged under appropriate circumstances, so as to be able to describe schemes other than the embodiments of the present application.

[0201] (5) In the present application, "predefined" can be realized by pre-saving corresponding codes, tables or other ways that can be used to indicate related information in the device, and the specific implementation manner is not limited in the present application.

[0202] (6) In the present application, "example", "for example", "exemplarily", "as (another) example" and the like are used to represent as an example, illustration or description. Any embodiment or design scheme described as "example" in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes.

[0203] (7) In the present application, "including", "containing", "having" and their variants mean "including but not limited to", unless otherwise specifically emphasized. "At least one" means one or more, and "multiple" means two or more.

[0204] (8) In the present application, "and / or", describes the association relationship of the associated objects, which means that there can be three kinds of relationships, for example, A and / or B, which can represent: A exists alone, A and B exist at the same time, B exists alone, where A and B can be singular or plural. The character " / " generally represents an "or" relationship between the associated objects before and after it. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or multiple items. For example, at least one of a, b and c can mean: a, or b, or c, or a and b, or a and c, or b and c, or a, b and c. Where a, b and c can be single or multiple.

[0205] (9) In various embodiments of the present application, the terms and / or descriptions of different embodiments are consistent and can be referred to each other if there is no special description and logical conflict, and the technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationship.

[0206] (10) Some optional features in various embodiments of the present application can not depend on other features in some scenarios, or can be combined with other features in some scenarios, without limitation.

[0207] The virtual object generation method provided by the embodiments of the present application is described in detail above in combination with FIGS. 4-6. The embodiments of the device of the present application will be described in detail below in combination with FIGS. 7-10. It should be understood that the description of the method embodiments corresponds to the description of the device embodiments, and therefore, the parts not described in detail can be referred to the foregoing method embodiments.

[0208] FIG. 7 is a schematic block diagram of a virtual object generation device 700 according to an embodiment of the present application.

[0209] The device 700 can be realized by software, hardware or a combination of both. The device 700 provided by the embodiments of the present application can implement the method flow shown in FIG. 4 of the embodiments of the present application, and the device 700 includes an interaction unit 710 and a processing unit 720.

[0210] The interaction unit 710 is configured to obtain a user demand.

[0211] The processing unit 720 includes an intelligent agent Agent and a functional agent.

[0212] The intelligent Agent is configured to perform the following operations: determining a generation demand according to the user demand, the generation demand being used to generate a virtual object; and decomposing the generation demand into at least one subtask, each subtask in the at least one subtask corresponding to a demand field, and each demand field including at least one demand function.

[0213] The function Agent is configured to determine a demand function corresponding to the virtual object in at least one demand field, and invoke a function model of the demand function corresponding to the virtual object in the at least one demand field to generate the virtual object.

[0214] Optionally, the processing unit further comprises a large language model (LLM) and a virtual object knowledge base. The determining the generated demand according to the user demand comprises: the intelligent Agent determines the generated demand according to the user demand by using the LLM and the virtual object knowledge base. The decomposing the generated demand into at least one subtask comprises: the intelligent Agent decomposes the generated demand into at least one subtask by using the LLM.

[0215] Optionally, the user demand comprises a first feature. The interaction unit is configured to output a correction prompt in a case where the virtual object knowledge base does not comprise the first feature, the correction prompt being configured to prompt modification of the first feature to a second feature, wherein the second feature is a feature comprised in the virtual object knowledge base; and acquire a confirmation message of the correction prompt. The intelligent Agent is configured to determine the generated demand according to the confirmation message of the correction prompt, the generated demand comprising the second feature.

[0216] Optionally, the interaction unit is configured to output a selection prompt, the selection prompt being configured to prompt selection of a third feature, wherein the third feature is a feature comprised in the virtual object knowledge base and not comprised in the user demand; and acquire a confirmation message of the selection prompt. The intelligent Agent is configured to determine the generated demand according to the confirmation message of the selection prompt, the generated demand comprising the third feature.

[0217] Optionally, the interaction unit is configured to output a confirmation prompt, the confirmation prompt being configured to prompt confirmation of the generated demand; and acquire a confirmation message of the confirmation prompt.

[0218] Optionally, the function Agent is configured to determine a demand function corresponding to the virtual object in each demand field. The interaction unit is configured to present a selection interface, the selection interface comprising the demand function corresponding to the virtual object in each demand field; and acquire the demand function corresponding to the virtual object in the at least one demand field selected through the selection interface.

[0219] Optionally, the interaction unit is further configured to present the virtual object.

[0220] The apparatus 700 here can be embodied in the form of functional modules. The term "unit" here can be implemented in the form of software and / or hardware, and is not specifically limited.

[0221] For example, the "unit" can be a software program, a hardware circuit, or a combination of both, which implements the above functions. The units in the apparatus 700 are described in detail below.

[0222] For example, the implementation of the interaction unit 710 in the apparatus 700 is described below. Similarly, the implementation of other units in the apparatus 700, such as the processing unit 720, can refer to the implementation of the interaction unit 710.

[0223] As an example of the interaction unit 710 as a software functional unit, the interaction unit 710 can include code running on a computing instance. The computing instance can include at least one of a physical host (computing device), a virtual machine, and a container. Further, the computing instance can be one or more. For example, the interaction unit 710 can include code running on multiple hosts / virtual machines / containers. It should be noted that the multiple hosts / virtual machines / containers used to run the code can be distributed in the same region, or in different regions. Further, the multiple hosts / virtual machines / containers used to run the code can be distributed in the same availability zone (AZ), or in different AZs, each AZ including a data center or multiple data centers with similar geographical locations. Generally, one region can include multiple AZs.

[0224] Similarly, the multiple hosts / virtual machines / containers used to run the code can be distributed in the same virtual private cloud (VPC), or in multiple VPCs. Generally, one VPC is set up in one region, and communication between two VPCs in the same region, or between VPCs in different regions, requires a communication gateway to be set up in each VPC to realize interconnection between VPCs.

[0225] As an example of a hardware functional unit, the interaction unit 710 can include at least one computing device, such as a server, etc. Alternatively, the interaction unit 710 can also be a device implemented by an application-specific integrated circuit (ASIC) or a programmable logic device (PLD), etc. The PLD can be implemented by a complex programmable logical device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof.

[0226] The plurality of computing devices included in the interaction unit 710 can be distributed in the same region or in different regions. The plurality of computing devices included in the interaction unit 710 can be distributed in the same AZ or in different AZs. Similarly, the plurality of computing devices included in the interaction unit 710 can be distributed in the same VPC or in multiple VPCs. The plurality of computing devices can be any combination of servers, ASICs, PLDs, CPLDs, FPGAs, and GALs, etc.

[0227] Therefore, the modules of the examples described in the embodiments of the present application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0228] It should be noted that the device 700 provided in the above embodiments is only used as an example to illustrate the division of the above functional units when performing the above method. In actual applications, the above functions can be completed by different functional units according to needs, i.e., the internal structure of the device 700 is divided into different functional units to complete all or part of the above-described functions. For example, the interaction unit 710 can be used to perform any step in the above method, and the processing unit 720 can be used to perform any step in the above method. The steps to be implemented by the interaction unit 710 and the processing unit 720 can be specified as needed, and the entire function of the device 700 can be implemented by the interaction unit 710 and the processing unit 720 respectively implementing different steps in the above method.

[0229] In addition, the apparatus and method embodiments provided by the above-described embodiments belong to the same concept, and the specific implementation process is described in the method embodiments above, which will not be described here.

[0230] The method provided by the embodiments of the present application can be executed by a computing device, which can also be referred to as a computer system. The computer system includes a hardware layer, an operating system layer running on the hardware layer, and an application layer running on the operating system layer. The hardware layer includes hardware such as a processing unit, a memory, and a memory control unit, and the functions and structures of the hardware will be described in detail later. The operating system is any one or more computer operating systems that implement business processing through processes, such as a Linux operating system, a Unix operating system, an Android operating system, an iOS operating system, or a windows operating system. The application layer includes application programs such as a browser, an address book, word processing software, and instant messaging software. In addition, the computer system can be a handheld device such as a smartphone or a terminal device such as a personal computer, and the present application is not particularly limited as long as the method provided by the embodiments of the present application can be executed. The execution subject of the method provided by the embodiments of the present application can be a computing device, or a functional module in the computing device that can call and execute a program.

[0231] Next, a computing device provided by the embodiments of the present application will be described in detail with reference to FIG. 8.

[0232] FIG. 8 is an architecture schematic diagram of a computing device 800 provided by the embodiments of the present application.

[0233] The computing device 800 can be a server or a computer or other device with computing capability. The computing device 800 shown in FIG. 8 includes at least one processor 810 and a memory 820.

[0234] It should be understood that the number of processors and memories in the computing device 800 is not limited by the present application.

[0235] The processor 810 executes instructions in the memory 820, so that the computing device 800 implements the method provided by the present application. Alternatively, the processor 810 executes instructions in the memory 820, so that the computing device 800 implements the functional modules provided by the present application, thereby implementing the method provided by the present application.

[0236] Optionally, the computing device 800 further includes a communication interface 830. The communication interface 830 uses a transceiver module such as but not limited to a network interface card and a transceiver to realize communication between the computing device 800 and other devices or communication networks.

[0237] Optionally, the computing device 800 also includes a system bus 840, wherein the processor 810, the memory 820 and the communication interface 830 are connected with the system bus 840 respectively. The processor 810 can access the memory 820 through the system bus 840, for example, the processor 810 can read and write data in the memory 820 or execute code in the memory 820 through the system bus 840. The system bus 840 is a peripheral component interconnect express (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The system bus 840 is divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, only one thick line is used in FIG. 8, but it does not mean that there is only one bus or only one type of bus.

[0238] In one possible implementation, the function of the processor 810 is mainly to interpret the instructions (or code) of the computer program and process the data in the computer software. Among them, the instructions of the computer program and the data in the computer software can be saved in the memory 820 or the cache 1316.

[0239] Optionally, the processor 810 can be an integrated circuit chip with a processing capability of signals. As an example but not limitation, the processor 810 is a general purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component. Among them, the general purpose processor is a microprocessor, etc. For example, the processor 810 is a central processing unit (CPU).

[0240] Optionally, each processor 810 includes at least one processing unit 812 and a memory control unit 814.

[0241] Optionally, processing units 812, also referred to as cores or core complexes, are the most important components of a processor. Processing units 812 are manufactured from single-crystal silicon using certain production processes, and all the calculations, command receiving, command storage, and data processing of a processor are performed by the core complexes. Processing units independently run program instructions and use the ability of parallel computing to speed up the running of programs. Various processing units have fixed logical structures, for example, a processing unit includes logical units such as a level one cache, a level two cache, an execution unit, an instruction stage unit, and a bus interface.

[0242] In one implementation, memory control units 814 are used to control the interaction of data between memory 820 and processing units 812. Specifically, memory control units 814 receive memory access requests from processing units 812 and control access to memory based on the memory access requests. By way of example and not limitation, a memory control unit is a memory management unit (MMU) or the like.

[0243] In one implementation, each memory control unit 814 addresses memory 820 through a system bus. An arbiter (not shown in FIG. 8) is configured in the system bus and is responsible for processing and coordinating the competing access of multiple processing units 812.

[0244] In one implementation, processing units 812 and memory control units 814 are communicatively connected through internal connections of a chip, such as address lines, so as to realize communication between processing units 812 and memory control units 814.

[0245] Optionally, each processor 810 further includes a cache 816, where the cache is a buffer for data exchange (referred to as cache). When a processing unit 812 wants to read data, it will first look for the required data in the cache, and if found, it will be directly executed, and if not found, it will be found from the memory. Since the running speed of the cache is much faster than that of the memory, the role of the cache is to help the processing unit 812 to run faster.

[0246] Memory 820 can provide a running space for a process in computing device 800, for example, the computer program (specifically, the code of the program) for generating the process is saved in memory 820. After the computer program is run by the processor to generate the process, the processor allocates a corresponding storage space for the process in memory 820. Further, the above-mentioned storage space further includes a text segment, an initialized data segment, a bit initialized data segment, a stack segment, a heap segment, and the like. Memory 820 saves the data generated during the running of the process in the above-mentioned storage space corresponding to the process, for example, intermediate data, or process data, and the like.

[0247] Optionally, memory (also referred to as storage) is used for storing data that is being accessed by processing unit 810. The data being used during processing and being transferred between memory and one or more hardware storage devices, such as a magnetic hard disk, optical disk, etc., for example. Processing unit 810 can access data stored in memory either by actually transferring the data from memory or by instructing processing unit 810 to retrieve data stored elsewhere (e.g., from one or more mass storage devices).

[0248] By way of example, and not limitation, memory 820 is volatile memory or nonvolatile memory, or can include both volatile and nonvolatile memory. By way of example, and not limitation, nonvolatile memory can be read-only memory (ROM), programmable ROM (PROM), erasable PROM (EPROM), electrically EPROM (EEPROM), or flash memory. Volatile memory is random access memory (RAM), which acts as external cache. By way of example and not limitation, many forms of RAM exist, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), Synchlink DRAM (SLDRAM), and direct rambus RAM (DR RAM). The system and method described herein can be stored on or transmitted across one or more of these forms of RAM, or nonvolatile memory.

[0249] The computing device 800 illustrated above is merely exemplary and the present application is not limited to this configuration. The computing device 800 of the embodiments of the present application includes various hardware in the computer system in the prior art, for example, the computing device 800 also includes other memories in addition to the memory 820, such as a disk memory, etc. Those skilled in the art should understand that the computing device 800 can also include other devices necessary for normal operation. Meanwhile, according to specific needs, those skilled in the art should understand that the above-mentioned computing device 800 can also include hardware devices for realizing other additional functions. In addition, those skilled in the art should understand that the above-mentioned computing device 800 can also only include devices necessary for realizing the embodiments of the present application, and does not necessarily include all the devices shown in FIG. 8.

[0250] Fig. 9 is a schematic diagram of an architecture of a computing device cluster according to an embodiment of the present application.

[0251] The computing device cluster comprises at least one computing device. The computing device can be a server. In some embodiments, the computing device can also be a desktop computer, a notebook computer, or a terminal device such as a smart phone.

[0252] As shown in Fig. 9, the computing device cluster comprises at least one computing device 800. The memory 820 in one or more computing devices 800 in the computing device cluster can store the same instructions for performing the above method.

[0253] In some possible implementation manners, the memory 820 in one or more computing devices 800 in the computing device cluster can also respectively store partial instructions for performing the above method. In other words, the combination of one or more computing devices 800 can collectively perform the instructions of the above method.

[0254] It should be noted that the memories 820 in different computing devices 800 in the computing device cluster can store different instructions, respectively used for performing part of the functions of the above apparatus. That is, the instructions stored in the memories 820 in different computing devices 800 can realize the functions of one or more modules in the above apparatus.

[0255] In some possible implementation manners, one or more computing devices in the computing device cluster can be connected through a network. The network can be a wide area network or a local area network, etc.

[0256] Fig. 10 is a schematic diagram of the connection between two computing devices 800A and 800B through a network according to an embodiment of the present application.

[0257] As shown in Fig. 10, two computing devices 800A and 800B are connected through a network. Specifically, the communication interface in each computing device is connected to the network.

[0258] It should be understood that the functions of the computing device 800A shown in Fig. 10 can also be completed by multiple computing devices 800. Similarly, the functions of the computing device 800B can also be completed by multiple computing devices 800.

[0259] The present application also provides a computer program product containing instructions, which can be software or program product containing instructions capable of running on a computing device or stored in any available medium. When it runs on the computing device, it enables the computing device to perform the above provided method, or enables the computing device to realize the functions of the above provided apparatus.

[0260] The application also provides a computer readable storage medium, which can be any available medium or data storage device including one or more available media that a computing device can store. The available media can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid state disk), etc. The computer readable storage medium includes instructions, which, when executed on a computing device, cause the computing device to perform the method provided above.

[0261] It should be understood that the size of the sequence number of each process described above in various embodiments of the application does not mean the order of execution, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the application.

[0262] Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the application.

[0263] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working process of the system, device and unit described above can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

[0264] In several embodiments provided in the application, it should be understood that the disclosed system, device and method can be realized by other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0265] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0266] In addition, each function unit in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit.

[0267] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the parts of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0268] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method of virtual object generation, the method comprising: The method comprises the following steps: obtaining a user demand; determining a generation demand according to the user demand, the generation demand being used to generate a virtual object; decomposing the generation demand into at least one subtask, each of the at least one subtask corresponding to a demand field, and each demand field comprising at least one demand function; determining a demand function corresponding to the virtual object under at least one demand field; calling a function model of the demand function corresponding to the virtual object under the at least one demand field to generate the virtual object.

2. The method of claim 1, wherein, The step of determining the generation demand according to the user demand comprises the following steps: determining the generation demand by using a large language model (LLM) and a virtual object knowledge base according to the user demand.

3. The method according to claim 1 or 2, characterized in that, The user demand comprises a first feature. The step of determining the generation demand according to the user demand comprises the following steps: in the case that the virtual object knowledge base does not comprise the first feature, outputting a correction prompt, the correction prompt being used to prompt to modify the first feature into a second feature, wherein the second feature is a feature comprised in the virtual object knowledge base; obtaining a confirmation message of the correction prompt; determining the generation demand according to the confirmation message of the correction prompt, the generation demand comprising the second feature.

4. The method according to any one of claims 1 to 3, characterized in that, The step of determining the generation demand according to the user demand comprises the following steps: outputting a selection prompt, the selection prompt being used to prompt to select a third feature, wherein the third feature is a feature comprised in the virtual object knowledge base and not comprised in the user demand; obtaining a confirmation message of the selection prompt; determining the generation demand according to the confirmation message of the selection prompt, the generation demand comprising the third feature.

5. The method according to any one of claims 1 to 4, characterized in that, Before the step of decomposing the generation demand into at least one subtask, the method further comprises the following steps: outputting a confirmation prompt, the confirmation prompt being used to prompt to confirm the generation demand; obtaining a confirmation message of the confirmation prompt.

6. The method according to any one of claims 1 to 5, characterized in that, The step of determining a demand function corresponding to the virtual object under at least one demand field comprises the following steps: determining a demand function corresponding to the virtual object under each demand field; presenting a selection interface, the selection interface comprising a demand function corresponding to the virtual object under each demand field; obtaining a demand function corresponding to the virtual object under the at least one demand field selected through the selection interface.

7. A virtual object generation apparatus characterized by comprising: The system comprises: an interaction unit, configured to obtain a user demand; a processing unit, the processing unit comprising an intelligent agent (Agent) and a function Agent; wherein the intelligent Agent is configured to perform the following operations: determining a generation demand according to the user demand, the generation demand being used to generate a virtual object; decomposing the generation demand into at least one subtask, each of the at least one subtask corresponding to a demand field, and each demand field comprising at least one demand function; the function Agent is configured to perform the following operations: determining a demand function corresponding to the virtual object under at least one demand field; calling a function model of the demand function corresponding to the virtual object under the at least one demand field to generate the virtual object.

8. The apparatus of claim 7, wherein, The processing unit further comprises a large language model (LLM) and a virtual object knowledge base; The generating requirement is determined according to the user requirement, and the determining includes: The intelligent agent determines the generating requirement according to the user requirement, by using the LLM and the virtual object knowledge base; The generating requirement is decomposed into at least one subtask, and the decomposing includes: The intelligent agent decomposes the generating requirement into at least one subtask by using the LLM.

9. The apparatus of claim 7 or 8, wherein, The user requirement includes a first feature; The interaction unit is configured to: In a case where the virtual object knowledge base does not include the first feature, output a correction prompt, the correction prompt being used to prompt to modify the first feature into a second feature, wherein the second feature is a feature included in the virtual object knowledge base; Obtain a confirmation message of the correction prompt; The intelligent agent is configured to: According to the confirmation message of the correction prompt, determine the generating requirement, and the generating requirement includes the second feature.

10. The apparatus of any one of claims 7 to 9, wherein, The interaction unit is configured to: Output a selection prompt, the selection prompt being used to prompt to select a third feature, wherein the third feature is a feature included in the virtual object knowledge base and not included in the user requirement; Obtain a confirmation message of the selection prompt; The intelligent agent is configured to: According to the confirmation message of the selection prompt, determine the generating requirement, and the generating requirement includes the third feature.

11. The apparatus of any one of claims 7 to 10, wherein, The interaction unit is configured to: Output a confirmation prompt, the confirmation prompt being used to prompt to confirm the generating requirement; Obtain a confirmation message of the confirmation prompt.

12. The apparatus of any one of claims 7-11, wherein, The functional agent is configured to: Determine a requirement function corresponding to the virtual object under each requirement field; The interaction unit is configured to: Present a selection interface, the selection interface including a requirement function corresponding to the virtual object under each requirement field; Obtain a requirement function corresponding to the virtual object under the at least one requirement field selected through the selection interface.

13. The apparatus of any one of claims 7-12, wherein, The interaction unit is further configured to: Present the virtual object.

14. A computer-readable storage medium, characterized in that, A computer program is stored thereon, and the computer program is executed by a computer to implement the method in any one of claims 1 to 6.

15. A computer program product, characterised in that, The computer program product includes computer program code, and when the computer program code is run on a computer, the method in any one of claims 1 to 6 is implemented.

16. A cluster of computing devices, characterized in that, At least one computing device is included, each computing device including a processor and a memory, and the processor of the at least one computing device is configured to execute instructions stored in the memory of the at least one computing device, so that the computing device cluster executes the method in any one of claims 1 to 6.

Images