Intelligent production and management method and system for cartoon game

Abstract

The invention relates to the technical field of intelligent production of cartoon games, in particular to an intelligent production and management method and system for cartoon games. The method comprises the following steps: acquiring animation game design data and animation game plot data; performing design element feature extraction according to the animation game design data so as to generate design element feature data, and performing plot emotion feature extraction and role positioning feature extraction according to the animation game plot data so as to generate plot emotion feature data and role positioning feature data; and constructing a role generation model according to the design element feature data so as to construct a cartoon game role generation model, and performing generation calculation on the cartoon game role generation model by utilizing the plot emotion feature data and the role positioning feature data so as to generate cartoon game role data. According to the method, the roles with various appearances and capabilities are generated by utilizing the design element features and the plot emotion features, so that the user experience is improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent production technology for animation and games, and in particular to an intelligent production and management method and system for animation and games. Background Technology

[0002] Anime and games refer to entertainment forms that combine animation and gaming elements. They typically present storylines and characters in animated form while incorporating the interactivity and gameplay mechanics of games. Anime and games can be played on various platforms, including computers, game consoles, and mobile devices. Intelligent production of anime and games refers to using intelligent technologies and algorithms to assist or enhance the production process. It involves using technologies such as machine learning, natural language processing, and computer vision to automate or enhance functions related to anime and game design, story generation, character generation, and character interaction.

[0003] However, in real-world applications, intelligent animation and game production often involves simply piling up existing materials for machine learning algorithms to generate content, resulting in relatively simplistic content for the generated animation and games. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes an intelligent production and management method and system for animation and games, thereby resolving at least one of the aforementioned technical issues.

[0005] This application provides a method for intelligent production and management of animation and games, including the following steps:

[0006] Step S1: Obtain anime / game design data and anime / game plot data;

[0007] Step S2: Extract design element features based on animation and game design data to generate design element feature data; and extract plot emotion features and character positioning features based on animation and game plot data to generate plot emotion feature data and character positioning feature data.

[0008] Step S3: Construct a character generation model based on the design element feature data, thereby constructing an animation and game character generation model. Then, use the plot emotion feature data and character positioning feature data to perform generation calculations on the animation and game character generation model, thereby generating animation and game character data.

[0009] Step S4: Utilize plot emotion feature data and character positioning feature data to generate character attributes from anime and game character data, thereby generating anime and game character attribute data for the anime and game intelligent production system to process character interactions; the character attribute generation is calculated using a character decision reward calculation formula, which is as follows:

[0010]

[0011] R represents the reward value for role decision, α represents the weighting coefficient of the random adjustment value for role decision, o represents the random adjustment value for role decision, δ represents the initial term, b represents the constant term, p represents the error adjustment term for anime / game characters, t represents the initial term for the decision sequence time, T represents the number of deployment steps in the decision sequence time, γ represents the decay factor, V() represents the estimated value of the value function, and s t+1 For the immediate reward at time step t+1, s t μ represents the immediate reward at time step t, and μ is a modifier for the role's decision reward value.

[0012] This invention utilizes feature extraction of design elements and character generation model construction to automatically generate design elements and characters for animation and games. This significantly reduces the workload of manual design, improves production efficiency, and allows for the generation of creative and diverse design elements and characters through algorithms. Through plot and emotional feature extraction and character generation in steps S2 and S3, animation and game characters with rich emotions and personalities can be generated based on plot and character positioning data. This helps enhance the game's emotional expression and character interaction experience, making players more immersive and engaged. Through character attribute generation and character decision reward calculation in step S4, attributes of animation and game characters can be generated based on plot and emotional feature data and character positioning feature data, and the decision-making of characters can be optimized through a decision reward calculation formula. This helps make the characters in the game more intelligent and adaptable, providing a more personalized and challenging gaming experience. Adopting intelligent production methods can significantly improve the production efficiency and consistency of animation and games. By automatically generating design elements, characters, and plots, the time and cost of manual production can be reduced, and the generated content can be kept consistent throughout the game, thereby improving the overall quality of the game and the user experience. Intelligent production methods can bring innovation and diversity to animation and game development. By automatically generating design elements and characters, the possibilities of different styles and creative ideas can be explored, thereby expanding the game's innovation and diversity and attracting a wider player base. It provides automated design generation, enhances narrative emotion and character interaction, generates character attributes, and optimizes decision-making, improving production efficiency and consistency, and promoting R&D innovation and diversity.

[0013] This invention constructs a formula for calculating role-based decision rewards. This formula fully considers the weighting coefficient α of the random adjustment value of the role-based decision, the random adjustment value o of the role-based decision, the initial term δ, the constant term b, the error adjustment term p of the anime / game character, the initial term t of the decision sequence time, the number of deployment steps T of the decision sequence time, the decay factor γ, the estimated value V() of the value function, and the immediate reward s at time step t+1. t+1 The instant reward s at time step t tThe text discusses the interactions between various elements, including the weighting coefficient α of the role decision random adjustment value (used to balance the impact of random adjustment on decision rewards and control the trade-off between exploration and utilization in decision-making), the role decision random adjustment value o (introducing randomness to increase decision diversity), the initial term δ (adjusting the offset of the initial reward value, which can adjust the initial state of the decision), the constant term b (adjusting the proportion of the reward value), the animation / game character error adjustment term p (representing the error of character behavior, which can be used to correct decision rewards), the decision sequence time initial term t (representing the start time of the decision sequence), the decision sequence time deployment step number T (representing the time length of the decision sequence), the decay factor γ (controlling the degree of reward decay, making the impact of the current reward on the decision decrease over time), the value function estimate V() (representing the value estimate of the current state, used to evaluate the quality of the state), and the immediate reward s at time step t+1. t+1 The instant reward at time step t+1 represents the reward the character receives in the next state, and the instant reward s at time step t represents the reward for the character. t This represents the reward a character receives in the current state. The correction term μ for the character's decision reward value is used to adjust and correct the decision reward. The formula considers multiple factors, such as random adjustment of decisions, decay factor, and value function estimation, so that the decision reward has a more comprehensive and detailed consideration, thereby improving the intelligence and adaptability of the character's decision-making.

[0014] Preferably, step S1 specifically includes:

[0015] Step S11: Obtain anime and game data, which includes anime and game image data and anime and game text data;

[0016] Step S12: Extract image design data and mark image design elements based on the anime and game image data to generate anime and game design data;

[0017] Step S13: Construct an anime / game text model based on the anime / game text data;

[0018] Step S14: Use the preset anime and game plot knowledge graph data to generate plot text for the anime and game text model, thereby generating anime and game plot data.

[0019] This invention acquires anime and game data and extracts and labels image and text data, making the game production process more efficient and convenient. By extracting and labeling image design elements, the design elements required in the game, such as character images and scene layouts, can be quickly generated. By constructing anime and game text models and utilizing plot knowledge graphs, the plot text in the game can be automatically generated, enriching the game's story and plot development, improving the efficiency and quality of game production, and enhancing the player's gaming experience.

[0020] Preferably, step S12 specifically includes:

[0021] Step S121: Extract image design data based on anime and game image data to generate anime and game image element data;

[0022] Step S122: Detect the anime and game image elements based on the data to generate anime and game image tag data;

[0023] Step S123: Extract design elements based on the anime and game image tag data and the anime and game image data to generate anime and game image design data;

[0024] Step S124: Obtain image design markup data, and use the image design markup data to mark image design elements in the animation and game image design data, thereby generating animation and game design data.

[0025] This invention improves the production efficiency and quality of animation and games by converting animation and game image data into image design element data, and by generating animation and game design data through design element extraction and image tagging. It also enhances the gaming experience for players. Traditional animation and game production processes require significant time and manpower, especially in image design, which necessitates manually drawing numerous image elements. This invention, through image design extraction and design element extraction methods, can automatically generate animation and game image design data, thereby significantly reducing production time and costs. Furthermore, by utilizing pre-defined animation and game plot knowledge graph data to generate plot text from the animation and game text model, and combining this with image design element tagging, more refined, vivid, and rich animation and game design data can be generated, thus improving the overall quality of game production.

[0026] Preferably, step S13 specifically includes:

[0027] Step S131: Perform text preprocessing and temporal text feature extraction based on the anime and game text data to obtain anime and game text preprocessing data and anime and game temporal text feature data;

[0028] Step S132: Perform word segmentation on the preprocessed anime and game text data to generate anime and game word segmentation data;

[0029] Step S133: Remove useless text from the segmented anime and game data to generate valid segmented data;

[0030] Step S134: Use the anime and game time-series text feature data to label and quantize the effective word segmentation data, thereby generating text vector data;

[0031] Step S135: Pre-train the text vector data to generate an anime / game text model.

[0032] This invention utilizes automated text preprocessing and word segmentation methods to quickly and accurately generate preprocessed text data, word segmentation data, and effective word segmentation data for anime and games, thereby significantly improving text processing efficiency. The plot of anime and games is a crucial factor in attracting players, and generating vivid, coherent, and engaging game plots is a challenging task. This invention, by utilizing temporal text feature data from anime and games and pre-training a text model, can generate coherent and attractive game plots, enhancing the generation capability and quality of game plots. Through preprocessing and feature extraction of anime and game text, information such as character positioning and emotional characteristics can be obtained, supporting personalized game plot generation and character interaction processing, providing a more personalized and richer gaming experience.

[0033] Preferably, step S14 specifically includes:

[0034] Step S141: Obtain anime and game plot knowledge graph data;

[0035] Step S142: Extract entity relationships based on the knowledge graph data of the anime and game plot to obtain entity relationship data;

[0036] Step S143: Generate a directed graph based on the entity relationship data, thereby generating entity relationship directed graph data;

[0037] Step S144: Use self-attention processing to process the directed graph data of entity relationships and the anime / game text model to generate script text, thereby generating anime / game plot data.

[0038] This invention utilizes plot knowledge graph data and entity relationship extraction to establish a structured representation of anime and game plots. Presenting plot information as a directed graph clearly shows the relationships and flow between entities, helping to maintain the logical coherence of the plot and avoiding abrupt or disjointed storylines. By combining the directed graph data of entity relationships with anime and game text models, a self-attention mechanism can be used to weight nodes and edges in the graph, thereby generating richer and more engaging plot text. The self-attention mechanism can adjust the weights of the generated text based on the relationships and importance between nodes, making the plot development more logical and attractive.

[0039] Preferably, the script text generation process is calculated using a script text error loss calculation formula, wherein the script text error loss calculation formula is specifically as follows:

[0040]

[0041] W represents the allowable value for the difference in text length generated by the target. p To generate text statistical relational data, s i The weighting coefficients for generating exact match probability values ​​of the text, c is the constant term for generating the text, and p i To generate a text exact match probability value, i takes the value 1, 2, 3...n, where n is the number of text exact match probability values ​​generated, and u is the correction coefficient for the allowable difference in length between the target and generated text.

[0042] This invention constructs a formula for calculating script text error loss, which fully considers the statistical relational data w of the generated text. p The weighting coefficient s of the probability value of generating an exact match for the text. i Generate text constant term c, generate text exact matching probability value p i The data includes the number of text matching probability values ​​(n) generated and their interrelationships, including the statistical relationship data (w) generated from the text. p Used to adjust the importance of statistical relational data in calculating error loss, a larger w p The value represents the weighting coefficient s of the probability value for generating an exact match of the generated text, with greater emphasis placed on the influence of statistical relational data. i The importance weight of each generated text exact match probability value is used to adjust the contribution of each matching probability value in calculating the error loss. The generated text constant term c is used to adjust the base of the logarithmic term of the generated text exact match probability value, affecting the scaling of the probability value and the result of calculating the error loss. The generated text exact match probability value p i This represents the probability value indicating the degree of matching between the generated text and the target text, used to measure the accuracy of the generated text. Each p... i Each value corresponds to a specific matching probability of the generated text. The correction coefficient u for the allowable difference in the length of the target generated text is used to adjust the allowable difference in the length of the target generated text.

[0043] Preferably, step S2 specifically includes:

[0044] Step S21: Extract design image features based on animation and game design data to generate design image feature data, which includes design image texture feature data, design image attribute feature data, and design image shape feature data.

[0045] Step S22: Perform symbol detection and labeling based on the design image feature data to generate design element feature data;

[0046] Step S23: Extract emotional features and character positioning features from the anime and game plot data to generate emotional feature data and character positioning feature data.

[0047] This invention extracts features such as texture, attributes, and shape from design images to accurately capture key design characteristics, thereby improving the quality and visual effects of animation and game design. Generating design image feature data helps designers better understand and analyze design elements, leading to improved and optimized designs. Extracting emotional features of the plot, such as joy / sorrow, tension / relaxation, allows for a more accurate expression of the emotional trajectory of the story, enhancing player resonance and engagement. Generating emotional plot feature data helps game developers create captivating story experiences and enhances the emotional appeal of games. Extracting character positioning features, such as protagonist and supporting characters, accurately grasps the character's position and role in the plot, helping game developers better shape and portray character images. Generating character positioning feature data helps game developers create distinctive characters, enhancing their appeal and emotional resonance. The extraction of design image feature data and the labeling of design element feature data effectively assist in the presentation and interpretation of the plot. Accurate detection and labeling of symbols and elements in design images helps ensure the smoothness and coherence of the plot, avoiding visual interference and disjointed plot twists. By designing image feature extraction, symbol detection and labeling, as well as extracting plot emotional features and character positioning features, this approach brings multifaceted benefits to the design and plot generation of animation and games. It not only improves design quality and the accuracy of plot expression but also enhances plot fluency and user experience, providing players with a richer and more engaging gaming experience.

[0048] Preferably, step S3 specifically includes:

[0049] Step S31: Perform feature preprocessing based on the design element feature data to generate design element feature preprocessing data;

[0050] Step S32: Perform neural network mapping calculation based on the preprocessed data of design element features to generate hidden layer data of design elements;

[0051] Step S33: Perform random data sampling based on the hidden layer data of the design elements to obtain random sampling data of the hidden layer of the design elements;

[0052] Step S34: Decode and reconstruct the hidden layer random sampling data of the design elements to build an animation and game character generation model;

[0053] Step S35: Use plot emotion feature data and character positioning feature data to perform error correction processing on the animation and game character generation model, thereby generating animation and game character data.

[0054] This invention employs feature preprocessing based on design element feature data, including data cleaning, dimensionality reduction, and standardization, to generate preprocessed design element feature data, ensuring data validity and accuracy. Next, neural network mapping calculations are performed using the preprocessed design element feature data to obtain hidden layer data. This neural network mapping method better uncovers the relationships between design elements, improving the accuracy of character generation. Following this, random data sampling is performed based on the hidden layer data. This random sampling method helps to select representative elements from a large number of design elements and improves the diversity of character generation. Then, decoding and reconstruction are performed based on the random sampling data to construct an animation / game character generation model. This decoding and reconstruction method better restores the original feature information of the design elements, thereby improving the accuracy and precision of character generation. Finally, error correction processing is performed on the animation / game character generation model using plot emotion feature data and character positioning feature data, resulting in animation / game character data. By using plot emotion feature data and character positioning feature data to perform error correction processing on the character generation model, the accuracy and conformity of character generation can be better controlled, improving the quality and reliability of character generation. This invention employs a comprehensive approach, integrating various data sources including anime / game plot data, design image feature data, plot emotional feature data, and character positioning feature data. By constructing hidden layer data of design elements and an anime / game character generation model, and combining techniques such as neural network mapping calculations and error correction processing, it achieves the goal of efficiently and accurately generating anime / game character data. This invention improves the accuracy, diversity, and quality of character generation, enriches the types of characters in anime / games and enhances the player experience, and has practical application value.

[0055] Preferably, step S4 specifically includes:

[0056] By performing temporal correlation on plot emotional feature data and character positioning feature data, temporal correlation feature data is generated.

[0057] Character attributes are predicted and generated based on time-series correlation feature data, thereby generating character attribute data for animation and game intelligent production systems to process character interactions.

[0058] This invention generates temporally correlated feature data by performing trigger condition temporal correlation on plot emotion feature data and character positioning feature data. The purpose of this step is to capture the temporal patterns and correlation rules between plot development and character behavior by analyzing the relationship between plot emotion and character positioning. The generation of temporally correlated feature data allows for more accurate prediction of character attributes and behaviors. In the step of predicting and generating character attributes based on the temporally correlated feature data, this invention utilizes the generated temporally correlated feature data, combined with a previously constructed animation and game character generation model, to predict and generate character attributes. The purpose of this step is to infer the attribute characteristics that a character may possess in different situations, such as personality, skills, and attitudes, based on the character's behavior, emotions, and plot development. Character attribute prediction and generation can provide rich character data for intelligent animation and game production systems, laying the foundation for character interaction processing and the establishment of relationships between characters. Through temporal correlation analysis of plot emotion and character positioning feature data and character attribute prediction and generation, the changes in character behavior and emotions within the plot can be captured more accurately, making the generated character data more realistic, rich, and coherent.

[0059] This application provides an intelligent production and management system for animation and games, the system comprising:

[0060] At least one processor; and,

[0061] A memory that is communicatively connected to the at least one processor;

[0062] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform an intelligent animation and game production and management method as described above.

[0063] The beneficial effects of this invention are as follows: By acquiring animation and game design data and animation and game plot data, this invention establishes a basic dataset. The design data includes image and text elements, while the plot data includes emotional and character positioning features. Through feature extraction of design elements from the animation and game design data, and feature extraction of plot emotion and character positioning features from the plot data, design element feature data, plot emotion feature data, and character positioning feature data are generated. These feature data capture the attributes of design elements in the animation and game, the changes in plot emotion, and the character's role positioning within the plot, providing basic features for subsequent steps. An animation and game character generation model is constructed based on the design element feature data, and the plot emotion feature data and character positioning feature data are used to perform generation calculations on this model, thereby generating animation and game character data. Through the construction and generation calculation of the character generation model, diverse and coherent animation and game character data can be generated based on features such as design elements, plot emotion, and character positioning. Character attributes are generated from the generated animation and game character data using the plot emotion feature data and character positioning feature data. This step analyzes the emotional nuances of the plot and the character's positioning characteristics to predict the character's attributes, such as personality, skills, and attitudes. This provides rich character attribute data for the intelligent animation and game production system, supporting the interaction between characters and the development of the plot. By extracting features from animation and game design and plot data, a character generation model is constructed to generate character data, and further, character attribute data is generated, enabling the automatic generation and personalized processing of animation and game characters. This series of steps makes animation and games richer, more realistic, and more coherent, enhancing their interactivity, emotional appeal, and playability. Attached Figure Description

[0064] Other features, objects, and advantages of this application will become more apparent from the following detailed description of the non-limiting embodiments, taken with reference to the accompanying drawings:

[0065] Figure 1 A flowchart illustrating the steps of an embodiment of an intelligent production and management method for animation and games is shown.

[0066] Figure 2 A flowchart illustrating the steps of an embodiment of anime / game design data and anime / game plot data acquisition method is shown.

[0067] Figure 3 A flowchart illustrating the steps of an embodiment of an animation and game design data generation method is shown.

[0068] Figure 4 A flowchart illustrating the steps of an embodiment of a method for constructing an animation game text model is shown.

[0069] Figure 5A flowchart illustrating the steps of an embodiment of a method for generating anime / game story data is shown.

[0070] Figure 6 A flowchart illustrating the steps of a method for generating plot emotion feature data and character positioning feature data according to an embodiment is shown.

[0071] Figure 7 A flowchart illustrating the steps of an embodiment of a method for generating anime / game character data is shown. Detailed Implementation

[0072] The technical method of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0073] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0074] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0075] Please see Figures 1 to 7 This application provides a method for intelligent production and management of animation and games, including the following steps:

[0076] Step S1: Obtain anime / game design data and anime / game plot data;

[0077] Specifically, anime and game design data can include character models, scene models, prop models, and related information such as texture maps, material information, and shape information. It also includes audio and video materials related to the game. Anime and game plot data can include the main storyline, side quests, and interactions between characters. Plot data can be in various formats, including text, audio, and video, and can also include the setting of story levels and mission objectives. Through plot data, information about the emotional state, behavioral preferences, and mission requirements of characters in the game can be obtained.

[0078] Step S2: Extract design element features based on animation and game design data to generate design element feature data; and extract plot emotion features and character positioning features based on animation and game plot data to generate plot emotion feature data and character positioning feature data.

[0079] Specifically, this involves analyzing and parsing character models, scene models, and prop models in game design data to extract their geometric shapes, textures, colors, and other visual features. Computer vision algorithms, such as edge detection, feature point extraction, and texture analysis, are used to extract key features from the design elements. For example, facial expression features of characters, lighting features of scenes, and shape features of props can be extracted. The extracted design element features are then appropriately represented and encoded for subsequent processing and analysis. Common representation methods include vector representation, matrix representation, and feature descriptors.

[0080] Specifically, for example, sentiment analysis can be performed on text within anime and game storyline data to identify the emotional tendency and intensity within the text. Sentiment analysis can be conducted using methods such as sentiment dictionaries and machine learning models. By analyzing character behavior and dialogue information within the storyline data, character localization and feature extraction can be performed. Information such as character identity, relationships between characters, and behavioral preferences can be identified. The extracted storyline sentiment features and character localization features are then appropriately represented and encoded for subsequent processing and analysis. Commonly used representation methods include vector representation, matrix representation, and feature descriptors.

[0081] Step S3: Construct a character generation model based on the design element feature data, thereby constructing an animation and game character generation model. Then, use the plot emotion feature data and character positioning feature data to perform generation calculations on the animation and game character generation model, thereby generating animation and game character data.

[0082] Specifically, for example, design element feature data can be used as input data, and anime / game character data as target data. Ensure data preparation and preprocessing are completed, including data cleaning and standardization. Based on specific needs and problems, select a suitable model to build the character generation model. Commonly used models include Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), and Recurrent Neural Networks (RNNs). The appropriate model selection should consider factors such as generation effect and training efficiency. Based on the selected model, design the model structure and parameter settings. For example, for GAN models, this includes building generator and discriminator networks; for VAE models, it includes building encoder and decoder networks. Simultaneously, determine the model's input and output layers based on the characteristics and dimensions of the design element feature data. Use design element feature data as input and anime / game character data as the target to train the constructed character generation model. Iteratively adjust the model parameters using optimization algorithms (such as gradient descent) to make the generated character data as close as possible to the target data. After training, use plot emotion feature data and character positioning feature data as input to perform generation calculations through the character generation model. The model will generate new anime and game character data based on the input feature data, which can be the character's appearance, behavior, and dialogue.

[0083] Step S4: Use plot emotion feature data and character positioning feature data to generate character attributes for anime and game character data, thereby generating anime and game character attribute data for the anime and game intelligent production system to process character interactions.

[0084] Specifically, this approach uses plot and emotional data, as well as character positioning data from anime and games, as state information for the environment. This state information can include the current plot progression, the intimacy between characters, and the emotional state of the characters. A set of possible actions for character attributes in the anime / game is defined, such as changing a character's personality traits, improving their skill level, or adjusting relationships between characters. Each action corresponds to a change in a character attribute. A reward function is defined for each action based on the game's design goals and strategies. The reward function can evaluate the effectiveness of the action based on changes in plot and emotional data and character positioning data. For example, if an action makes the relationship between a character and the protagonist closer, a positive reward can be given. A suitable reinforcement learning algorithm, such as Q-learning or Deep Q-Network (DQN), is selected to train an intelligent agent for character attribute generation. The agent's task is to select appropriate actions based on the current state information to maximize cumulative rewards. Training and optimization are performed using actual data from the anime / game. Through interaction with players or multiple experiments using simulators, the intelligent agent learns appropriate strategies and behaviors to generate character attribute data that aligns with the plot's emotional content and the character's positioning. The generated character attribute data for anime and games is applied to an intelligent anime and game production system for handling character interactions. The system can determine a character's dialogue, actions, and decisions based on this attribute data, enabling interactions with other characters or players.

[0085] The generation of character attributes is calculated using the character decision reward calculation formula, which is as follows:

[0086]

[0087] R represents the reward value for role decision, α represents the weighting coefficient of the random adjustment value for role decision, o represents the random adjustment value for role decision, δ represents the initial term, b represents the constant term, p represents the error adjustment term for anime / game characters, t represents the initial term for the decision sequence time, T represents the number of deployment steps in the decision sequence time, γ represents the decay factor, V() represents the estimated value of the value function, and s t+1 For the immediate reward at time step t+1, s t μ represents the immediate reward at time step t, and μ is a modifier for the role's decision reward value.

[0088] This invention utilizes feature extraction of design elements and character generation model construction to automatically generate design elements and characters for animation and games. This significantly reduces the workload of manual design, improves production efficiency, and allows for the generation of creative and diverse design elements and characters through algorithms. Through plot and emotional feature extraction and character generation in steps S2 and S3, animation and game characters with rich emotions and personalities can be generated based on plot and character positioning data. This helps enhance the game's emotional expression and character interaction experience, making players more immersive and engaged. Through character attribute generation and character decision reward calculation in step S4, attributes of animation and game characters can be generated based on plot and emotional feature data and character positioning feature data, and the decision-making of characters can be optimized through a decision reward calculation formula. This helps make the characters in the game more intelligent and adaptable, providing a more personalized and challenging gaming experience. Adopting intelligent production methods can significantly improve the production efficiency and consistency of animation and games. By automatically generating design elements, characters, and plots, the time and cost of manual production can be reduced, and the generated content can be kept consistent throughout the game, thereby improving the overall quality of the game and the user experience. Intelligent production methods can bring innovation and diversity to animation and game development. By automatically generating design elements and characters, the possibilities of different styles and creative ideas can be explored, thereby expanding the game's innovation and diversity and attracting a wider player base. It provides automated design generation, enhances narrative emotion and character interaction, generates character attributes, and optimizes decision-making, improving production efficiency and consistency, and promoting R&D innovation and diversity.

[0089] This invention constructs a formula for calculating role-based decision rewards. This formula fully considers the weighting coefficient α of the random adjustment value of the role-based decision, the random adjustment value o of the role-based decision, the initial term δ, the constant term b, the error adjustment term p of the anime / game character, the initial term t of the decision sequence time, the number of deployment steps T of the decision sequence time, the decay factor γ, the estimated value V() of the value function, and the immediate reward s at time step t+1. t+1 The instant reward s at time step t tThe text discusses the interactions between various elements, including the weighting coefficient α of the role decision random adjustment value (used to balance the impact of random adjustment on decision rewards and control the trade-off between exploration and utilization in decision-making), the role decision random adjustment value o (introducing randomness to increase decision diversity), the initial term δ (adjusting the offset of the initial reward value, which can adjust the initial state of the decision), the constant term b (adjusting the proportion of the reward value), the animation / game character error adjustment term p (representing the error of character behavior, which can be used to correct decision rewards), the decision sequence time initial term t (representing the start time of the decision sequence), the decision sequence time deployment step number T (representing the time length of the decision sequence), the decay factor γ (controlling the degree of reward decay, making the impact of the current reward on the decision decrease over time), the value function estimate V() (representing the value estimate of the current state, used to evaluate the quality of the state), and the immediate reward s at time step t+1. t+1 The instant reward at time step t+1 represents the reward the character receives in the next state, and the instant reward s at time step t represents the reward for the character. t This represents the reward a character receives in the current state. The correction term μ for the character's decision reward value is used to adjust and correct the decision reward. The formula considers multiple factors, such as random adjustment of decisions, decay factor, and value function estimation, so that the decision reward has a more comprehensive and detailed consideration, thereby improving the intelligence and adaptability of the character's decision-making.

[0090] Preferably, step S1 specifically includes:

[0091] Step S11: Obtain anime and game data, which includes anime and game image data and anime and game text data;

[0092] Specifically, for example, anime and game data can be obtained from an interface or a pre-set database, where the anime and game data includes anime and game image data as well as anime and game text data.

[0093] Step S12: Extract image design data and mark image design elements based on the anime and game image data to generate anime and game design data;

[0094] Specifically, this involves extracting design-related features, such as color palettes, texture styles, and shape elements, from anime and game image data. Image processing algorithms and pattern recognition techniques can be used to analyze and extract features from the images. The extracted image design elements are then labeled and classified; for example, different texture styles are labeled as texture element A, texture element B, etc., and specific shape elements are labeled as shape element X, shape element Y, etc.

[0095] Step S13: Construct an anime / game text model based on the anime / game text data;

[0096] Specifically, this involves preprocessing anime and game text data, including text cleaning, word segmentation, and stop word removal. Natural language processing techniques are then used to extract key information and features from the preprocessed text data. For example, features such as word frequency, word vectors, and sentiment can be extracted. Based on the extracted text features, an anime and game text model is constructed. Machine learning algorithms or deep learning models, such as recurrent neural networks (RNNs) or Transformer models, can be used.

[0097] Step S14: Use the preset anime and game plot knowledge graph data to generate plot text for the anime and game text model, thereby generating anime and game plot data.

[0098] Specifically, for example, a knowledge graph of anime / game plot can be pre-constructed, containing information such as character relationships, plot nodes, and plot development. This knowledge can be stored in the form of a graph, for example, using a graph database. Based on the anime / game text model and the plot knowledge graph, techniques such as generation algorithms and self-attention mechanisms are used to generate plot text with narrative coherence and emotional plausibility. The generated text can serve as plot development clues or dialogue content for the anime / game.

[0099] This invention acquires anime and game data and extracts and labels image and text data, making the game production process more efficient and convenient. By extracting and labeling image design elements, the design elements required in the game, such as character images and scene layouts, can be quickly generated. By constructing anime and game text models and utilizing plot knowledge graphs, the plot text in the game can be automatically generated, enriching the game's story and plot development, improving the efficiency and quality of game production, and enhancing the player's gaming experience.

[0100] Preferably, step S12 specifically includes:

[0101] Step S121: Extract image design data based on anime and game image data to generate anime and game image element data;

[0102] Specifically, computer vision techniques can be used to extract key features from images, such as color, texture, and shape. Image processing algorithms, feature extraction methods, or deep learning models, such as convolutional neural networks (CNNs), can be used to extract feature representations of the images. The extracted image features are then encoded, transforming them into a data format that can be processed by a computer. For example, vectorization methods can be used to convert image features into numerical vectors. Based on the extracted and encoded image features, anime and game image element data can be generated. This data can include color features, texture features, shape features, etc., used to describe the design elements of the image.

[0103] Step S122: Detect the anime and game image elements based on the data to generate anime and game image tag data;

[0104] Specifically, for example, a trained object detection model can be used to detect different elements in an image, such as people, objects, and scenes. Detection algorithms can be based on deep learning models, such as Faster R-CNN and YOLO. Based on the detection results, anime and game image tagging data is generated, recording information such as the location, category, and attributes of each image element.

[0105] Step S123: Extract design elements based on the anime and game image tag data and the anime and game image data to generate anime and game image design data;

[0106] Specifically, image segmentation techniques can be used to divide an image into different regions or pixels, with each region corresponding to a specific image element. Commonly used image segmentation methods include edge-based, region-based, and deep learning-based segmentation models.

[0107] Step S124: Obtain image design tag data and use it to tag image design elements in the animation / game image design data, thereby generating animation / game design data. The image design tag data is used to tag elements in the animation / game image design data. Based on the tag information in the design tag data, the corresponding design elements are tagged. For example, an image area tagged with tag A can be considered a texture element, generating animation / game design data containing image design element tags.

[0108] Specifically, for example, it can acquire predefined image design tag data, which contains labels and attribute information for different image design elements. For instance, label A could represent a texture element, label B could represent a shape element, and so on.

[0109] This invention improves the production efficiency and quality of animation and games by converting animation and game image data into image design element data, and by generating animation and game design data through design element extraction and image tagging. It also enhances the gaming experience for players. Traditional animation and game production processes require significant time and manpower, especially in image design, which necessitates manually drawing numerous image elements. This invention, through image design extraction and design element extraction methods, can automatically generate animation and game image design data, thereby significantly reducing production time and costs. Furthermore, by utilizing pre-defined animation and game plot knowledge graph data to generate plot text from the animation and game text model, and combining this with image design element tagging, more refined, vivid, and rich animation and game design data can be generated, thus improving the overall quality of game production.

[0110] Preferably, step S13 specifically includes:

[0111] Step S131: Perform text preprocessing and temporal text feature extraction based on the anime and game text data to obtain anime and game text preprocessing data and anime and game temporal text feature data;

[0112] Specifically, for example, in the text preprocessing stage, a series of processing operations are performed on the anime and game text data, including removing special characters, converting to lowercase, and removing stop words. At the same time, temporal text feature extraction is performed, which can capture the text's feature information by extracting word frequency, TF-IDF value, word vectors, etc.

[0113] Step S132: Perform word segmentation on the preprocessed anime and game text data to generate anime and game word segmentation data;

[0114] Specifically, for example, Chinese word segmentation tools (such as jieba segmentation) can be used to segment the preprocessed text data, breaking the text into individual word units.

[0115] Step S133: Remove useless text from the segmented anime and game data to generate valid segmented data;

[0116] Specifically, for example, based on domain-specific dictionaries or rules, the segmented data is screened and filtered to remove useless words or stop words and retain effective segmented words related to anime and games.

[0117] Step S134: Use the anime and game time-series text feature data to label and quantize the effective word segmentation data, thereby generating text vector data;

[0118] Specifically, for example, based on temporal text feature data, each effective word segment can be assigned a corresponding tag or label to express its importance or special meaning in the text. By converting the tagged effective words into vector representations, the text data can be converted into numerical form.

[0119] Step S135: Pre-train the text vector data to generate an anime / game text model.

[0120] Specifically, for example, using the generated text vector data, machine learning or deep learning methods can be used to pre-train the text. This can include using models such as autoencoders, recurrent neural networks (RNNs), and long short-term memory networks (LSTMs) for text modeling and feature learning, thereby generating anime and game text models to capture latent relationships and semantic information in the text data.

[0121] This invention utilizes automated text preprocessing and word segmentation methods to quickly and accurately generate preprocessed text data, word segmentation data, and effective word segmentation data for anime and games, thereby significantly improving text processing efficiency. The plot of anime and games is a crucial factor in attracting players, and generating vivid, coherent, and engaging game plots is a challenging task. This invention, by utilizing temporal text feature data from anime and games and pre-training a text model, can generate coherent and attractive game plots, enhancing the generation capability and quality of game plots. Through preprocessing and feature extraction of anime and game text, information such as character positioning and emotional characteristics can be obtained, supporting personalized game plot generation and character interaction processing, providing a more personalized and richer gaming experience.

[0122] Preferably, step S14 specifically includes:

[0123] Step S141: Obtain anime and game plot knowledge graph data;

[0124] Specifically, data related to the plot of anime and games can be obtained from reliable data sources or specially constructed knowledge graphs, including character information, scene information, and event relationships. The data is organized and constructed in the form of a graph.

[0125] Step S142: Extract entity relationships based on the knowledge graph data of the anime and game plot to obtain entity relationship data;

[0126] Specifically, for example, natural language processing and graph analysis methods can be used to parse and analyze knowledge graph data of animation and game plots, extracting entities (such as characters and scenes) and the relationships between them (such as interaction relationships and emotional relationships).

[0127] Step S143: Generate a directed graph based on the entity relationship data, thereby generating entity relationship directed graph data;

[0128] Specifically, for example, a directed graph can be constructed based on entity relationship data, where entities act as nodes and relationships connect nodes as edges. Each node can contain additional attribute information, such as character attributes or scene descriptions. The structure of the directed graph and the weights of the edges are designed according to the importance of the entity relationships.

[0129] Step S144: Use self-attention processing to process the directed graph data of entity relationships and the anime / game text model to generate script text, thereby generating anime / game plot data.

[0130] Specifically, for example, self-attention mechanisms (such as the Transformer model) can be used to process directed graph data of entity relationships, enabling the model to focus on important nodes and edges, thereby generating script text. Simultaneously, by combining anime and game text models, information from plot knowledge graphs and text features can be integrated to generate anime and game plot data that conforms to plot logic and emotional expression.

[0131] This invention utilizes plot knowledge graph data and entity relationship extraction to establish a structured representation of anime and game plots. Presenting plot information as a directed graph clearly shows the relationships and flow between entities, helping to maintain the logical coherence of the plot and avoiding abrupt or disjointed storylines. By combining the directed graph data of entity relationships with anime and game text models, a self-attention mechanism can be used to weight nodes and edges in the graph, thereby generating richer and more engaging plot text. The self-attention mechanism can adjust the weights of the generated text based on the relationships and importance between nodes, making the plot development more logical and attractive.

[0132] Preferably, the script text generation process is calculated using a script text error loss calculation formula, wherein the script text error loss calculation formula is specifically as follows:

[0133]

[0134] W represents the allowable value for the difference in text length generated by the target. p To generate text statistical relational data, s i The weighting coefficients for generating exact match probability values ​​of the text, c is the constant term for generating the text, and p i To generate a text exact match probability value, i takes the value 1, 2, 3...n, where n is the number of text exact match probability values ​​generated, and u is the correction coefficient for the allowable difference in length between the target and generated text.

[0135] This invention constructs a formula for calculating script text error loss, which fully considers the statistical relational data w of the generated text. p The weighting coefficient s of the probability value of generating an exact match for the text. i Generate text constant term c, generate text exact matching probability value p i The data includes the number of text matching probability values ​​(n) generated and their interrelationships, including the statistical relationship data (w) generated from the text. p Used to adjust the importance of statistical relational data in calculating error loss, a larger w p The value represents the weighting coefficient s of the probability value for generating an exact match of the generated text, with greater emphasis placed on the influence of statistical relational data. iThe importance weight of each generated text exact match probability value is used to adjust the contribution of each matching probability value in calculating the error loss. The generated text constant term c is used to adjust the base of the logarithmic term of the generated text exact match probability value, affecting the scaling of the probability value and the result of calculating the error loss. The generated text exact match probability value p i This represents the probability value indicating the degree of matching between the generated text and the target text, used to measure the accuracy of the generated text. Each p... i Each value corresponds to a specific matching probability of the generated text. The correction coefficient u for the allowable difference in the length of the target generated text is used to adjust the allowable difference in the length of the target generated text.

[0136] Preferably, step S2 specifically includes:

[0137] Step S21: Extract design image features based on animation and game design data to generate design image feature data, which includes design image texture feature data, design image attribute feature data, and design image shape feature data.

[0138] Specifically, for example, computer vision techniques can be used to extract features from images in animation and game design data. First, we extract texture features from the design images, such as color distribution and texture density, which can describe the texture style and feel of the image. Next, we extract attribute features from the design images, such as the character's clothing and scene attributes, describing the object attributes and environmental features of the image. Finally, we extract shape features from the design images, such as the character's outline and the shape of the scene.

[0139] Step S22: Perform symbol detection and labeling based on the design image feature data to generate design element feature data;

[0140] Specifically, for example, computer vision technology can be used to perform symbol detection and labeling on design image feature data. Object detection algorithms can be used to identify different design elements in the image, such as characters, props, and scenes. By detecting and labeling these design elements, feature data of the design elements can be obtained, including information on the element's position, size, and category.

[0141] Step S23: Extract emotional features and character positioning features from the anime and game plot data to generate emotional feature data and character positioning feature data.

[0142] Specifically, this includes extracting emotional features and character positioning features from anime and game plot data. Extracting emotional features includes analyzing the emotional trajectory and changes within the plot.

[0143] This invention extracts features such as texture, attributes, and shape from design images to accurately capture key design characteristics, thereby improving the quality and visual effects of animation and game design. Generating design image feature data helps designers better understand and analyze design elements, leading to improved and optimized designs. Extracting emotional features of the plot, such as joy / sorrow, tension / relaxation, allows for a more accurate expression of the emotional trajectory of the story, enhancing player resonance and engagement. Generating emotional plot feature data helps game developers create captivating story experiences and enhances the emotional appeal of games. Extracting character positioning features, such as protagonist and supporting characters, accurately grasps the character's position and role in the plot, helping game developers better shape and portray character images. Generating character positioning feature data helps game developers create distinctive characters, enhancing their appeal and emotional resonance. The extraction of design image feature data and the labeling of design element feature data effectively assist in the presentation and interpretation of the plot. Accurate detection and labeling of symbols and elements in design images helps ensure the smoothness and coherence of the plot, avoiding visual interference and disjointed plot twists. By designing image feature extraction, symbol detection and labeling, as well as extracting plot emotional features and character positioning features, this approach brings multifaceted benefits to the design and plot generation of animation and games. It not only improves design quality and the accuracy of plot expression but also enhances plot fluency and user experience, providing players with a richer and more engaging gaming experience.

[0144] Preferably, step S3 specifically includes:

[0145] Step S31: Perform feature preprocessing based on the design element feature data to generate design element feature preprocessing data;

[0146] Specifically, this could involve preprocessing the feature data of design elements to better suit subsequent neural network computations. Preprocessing could include operations such as feature normalization, scaling, and smoothing to ensure data consistency and comparability. For example, the positions of design elements could be normalized to map them to a uniform coordinate system; the sizes of elements could be scaled to have a similar range; and features could be smoothed to remove noise and outliers.

[0147] Step S32: Perform neural network mapping calculation based on the preprocessed data of design element features to generate hidden layer data of design elements;

[0148] Specifically, for example, a neural network model can be used to map and compute preprocessed data of design element features to generate hidden layer representations of the design elements. These hidden layer representations can be viewed as abstractions and encodings of the design element features, containing high-level semantic information and latent characteristics. Through the forward propagation process of the neural network, the design element features are mapped into the hidden layer space, yielding the hidden layer data of the design elements.

[0149] Step S33: Perform random data sampling based on the hidden layer data of the design elements to obtain random sampling data of the hidden layer of the design elements;

[0150] Specifically, for example, random sampling can be performed on the hidden layer data of design elements to obtain diverse representations of design elements. By randomly sampling the hidden layer data of design elements, new hidden layer vectors can be generated.

[0151] Step S34: Decode and reconstruct the hidden layer random sampling data of the design elements to build an animation and game character generation model;

[0152] Specifically, for example, a decoder model can be used to reconstruct and generate random sampling data from the hidden layers of design elements, thereby constructing an anime / game character generation model. The decoder model takes the hidden layer sampling data as input and decodes it into feature representations of the design elements through a reverse neural network computation process. By continuously adjusting the parameters of the decoder model, it can generate anime / game characters that meet the expected specifications.

[0153] Step S35: Use plot emotion feature data and character positioning feature data to perform error correction processing on the animation and game character generation model, thereby generating animation and game character data.

[0154] Specifically, for example, plot emotion feature data and character positioning feature data can be used to refine and adjust the character generation model for anime and games, making the generated characters more consistent with the emotional trajectory and character positioning of the plot. By analyzing plot emotion feature data, the emotional changes at different stages of the plot can be obtained, such as tension, sadness, and joy. Meanwhile, character positioning feature data can provide information about a character's role and relationships within the plot.

[0155] This invention employs feature preprocessing based on design element feature data, including data cleaning, dimensionality reduction, and standardization, to generate preprocessed design element feature data, ensuring data validity and accuracy. Next, neural network mapping calculations are performed using the preprocessed design element feature data to obtain hidden layer data. This neural network mapping method better uncovers the relationships between design elements, improving the accuracy of character generation. Following this, random data sampling is performed based on the hidden layer data. This random sampling method helps to select representative elements from a large number of design elements and improves the diversity of character generation. Then, decoding and reconstruction are performed based on the random sampling data to construct an animation / game character generation model. This decoding and reconstruction method better restores the original feature information of the design elements, thereby improving the accuracy and precision of character generation. Finally, error correction processing is performed on the animation / game character generation model using plot emotion feature data and character positioning feature data, resulting in animation / game character data. By using plot emotion feature data and character positioning feature data to perform error correction processing on the character generation model, the accuracy and conformity of character generation can be better controlled, improving the quality and reliability of character generation. This invention employs a comprehensive approach, integrating various data sources including anime / game plot data, design image feature data, plot emotional feature data, and character positioning feature data. By constructing hidden layer data of design elements and an anime / game character generation model, and combining techniques such as neural network mapping calculations and error correction processing, it achieves the goal of efficiently and accurately generating anime / game character data. This invention improves the accuracy, diversity, and quality of character generation, enriches the types of characters in anime / games and enhances the player experience, and has practical application value.

[0156] Preferably, step S4 specifically includes:

[0157] By performing temporal correlation on plot emotional feature data and character positioning feature data, temporal correlation feature data is generated.

[0158] Specifically, for example, plot emotion feature data and character positioning feature data can be used to establish temporal relationships between them. Temporal relationship feature data reflects the development and transformation of different events and emotions in the plot and is associated with the character's positioning. By analyzing the time-series information of plot emotion feature data and character positioning feature data, the triggering conditions and evolutionary patterns of specific events or emotions can be derived. The plot emotion feature data and character positioning feature data are then temporally sorted and labeled to ensure their continuity and consistency over time. By analyzing their temporal changes, characteristic patterns of some key events, emotional turning points, or character behaviors can be identified. Based on these characteristic patterns, temporal relationship feature data is constructed, including the triggering conditions, development trends, and temporal relationships of different events or emotions.

[0159] Character attributes are predicted and generated based on time-series correlation feature data, thereby generating character attribute data for animation and game intelligent production systems to process character interactions.

[0160] Specifically, for example, based on temporal correlation feature data, a pre-defined neural network model or reinforcement learning algorithm can be used to model the character attribute data and interactive environment of anime and games into a reinforcement learning environment. This involves defining a state space, including the character's current attribute state and scene information; defining an action space, including different behaviors or interaction options the character can take; defining a reward function to evaluate the quality of the character's actions in a specific state; and selecting a suitable reinforcement learning algorithm, such as Q-learning or Deep Q-Network (DQN). Using anime and game data as a training set, an agent interacts with the character in the game environment and learns the optimal strategy. The agent selects actions based on the current state and learns and optimizes based on the reward signals provided by the environment. After training, the trained reinforcement learning model is used to generate anime and game character attribute data based on the current game state. The agent selects the optimal action based on the current state and applies the generated character attribute data to game interaction processing.

[0161] This invention generates temporally correlated feature data by performing trigger condition temporal correlation on plot emotion feature data and character positioning feature data. The purpose of this step is to capture the temporal patterns and correlation rules between plot development and character behavior by analyzing the relationship between plot emotion and character positioning. The generation of temporally correlated feature data allows for more accurate prediction of character attributes and behaviors. In the step of predicting and generating character attributes based on the temporally correlated feature data, this invention utilizes the generated temporally correlated feature data, combined with a previously constructed animation and game character generation model, to predict and generate character attributes. The purpose of this step is to infer the attribute characteristics that a character may possess in different situations, such as personality, skills, and attitudes, based on the character's behavior, emotions, and plot development. Character attribute prediction and generation can provide rich character data for intelligent animation and game production systems, laying the foundation for character interaction processing and the establishment of relationships between characters. Through temporal correlation analysis of plot emotion and character positioning feature data and character attribute prediction and generation, the changes in character behavior and emotions within the plot can be captured more accurately, making the generated character data more realistic, rich, and coherent.

[0162] This application provides an intelligent production and management system for animation and games, the system comprising:

[0163] At least one processor; and,

[0164] A memory that is communicatively connected to the at least one processor;

[0165] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform an intelligent animation and game production and management method as described above.

[0166] The beneficial effects of this invention are as follows: By acquiring animation and game design data and animation and game plot data, this invention establishes a basic dataset. The design data includes image and text elements, while the plot data includes emotional and character positioning features. Through feature extraction of design elements from the animation and game design data, and feature extraction of plot emotion and character positioning features from the plot data, design element feature data, plot emotion feature data, and character positioning feature data are generated. These feature data capture the attributes of design elements in the animation and game, the changes in plot emotion, and the character's role positioning within the plot, providing basic features for subsequent steps. An animation and game character generation model is constructed based on the design element feature data, and the plot emotion feature data and character positioning feature data are used to perform generation calculations on this model, thereby generating animation and game character data. Through the construction and generation calculation of the character generation model, diverse and coherent animation and game character data can be generated based on features such as design elements, plot emotion, and character positioning. Character attributes are generated from the generated animation and game character data using the plot emotion feature data and character positioning feature data. This step analyzes the emotional nuances of the plot and the character's positioning characteristics to predict the character's attributes, such as personality, skills, and attitudes. This provides rich character attribute data for the intelligent animation and game production system, supporting the interaction between characters and the development of the plot. By extracting features from animation and game design and plot data, a character generation model is constructed to generate character data, and further, character attribute data is generated, enabling the automatic generation and personalized processing of animation and game characters. This series of steps makes animation and games richer, more realistic, and more coherent, enhancing their interactivity, emotional appeal, and playability.

[0167] Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended application documents rather than the foregoing description. Thus, it is intended that all variations falling within the meaning and scope of the equivalents of the application documents be incorporated into the invention.

[0168] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A method for intelligent production and management of animation and games, characterized in that, Includes the following steps: Step S1: Obtain anime / game design data and anime / game plot data; Step S2: Extract design element features based on animation and game design data to generate design element feature data; and extract plot emotion features and character positioning features based on animation and game plot data to generate plot emotion feature data and character positioning feature data. Step S3: Construct a character generation model based on the design element feature data, thereby building an animation and game character generation model. Then, use the plot emotion feature data and character positioning feature data to generate calculations for the animation and game character generation model, thereby generating animation and game character data. Step S4: Use plot emotion feature data and character positioning feature data to generate character attributes for anime and game character data, thereby generating anime and game character attribute data for the anime and game intelligent production system to process character interactions. The generation of character attributes is calculated using the character decision reward calculation formula, which is as follows: R represents the reward value for role decision, α represents the weighting coefficient of the random adjustment value for role decision, o represents the random adjustment value for role decision, δ represents the initial term, b represents the constant term, p represents the error adjustment term for anime / game characters, t represents the initial term for the decision sequence time, T represents the number of deployment steps in the decision sequence time, γ represents the decay factor, V() represents the estimated value of the value function, and s t+1 For the immediate reward at time step t+1, s t μ represents the immediate reward at time step t, and μ is a modifier for the role's decision reward value.

2. The method according to claim 1, characterized in that, Step S1 is as follows: Step S11: Obtain anime and game data, which includes anime and game image data and anime and game text data; Step S12: Extract image design data and mark image design elements based on the anime and game image data to generate anime and game design data; Step S13: Construct an anime / game text model based on the anime / game text data; Step S14: Use the preset anime and game plot knowledge graph data to generate plot text for the anime and game text model, thereby generating anime and game plot data.

3. The method according to claim 2, characterized in that, Step S12 is as follows: Image design is extracted from anime and game image data to generate anime and game image element data; Anime and game image element data is detected to generate anime and game image tag data; Design elements are extracted from anime and game image tag data and anime and game image data to generate anime and game image design data; The process involves acquiring image design tagging data and using this data to tag image design elements in animation and game image design data, thereby generating animation and game design data.

4. The method according to claim 2, characterized in that, Step S13 is as follows: Based on the text data of animation and games, text preprocessing and time-series text feature extraction are performed to obtain preprocessed text data of animation and games and time-series text feature data of animation and games. The preprocessed text data of anime and games is segmented into words to generate segmented anime and game word data; Useless text is removed from segmented anime and game data to generate effective segmented data; By using the temporal text feature data of animation and games, effective word segmentation data is labeled and quantized to generate text vector data; Pre-training is performed based on text vector data to generate anime and game text models.

5. The method according to claim 2, characterized in that, Step S14 is as follows: Obtain knowledge graph data of anime and game plots; Entity relationships are extracted from knowledge graph data of anime and game plots to obtain entity relationship data; A directed graph is generated based on the entity relationship data, thereby generating the entity relationship directed graph data; The self-attention processing method is used to process directed graph data of entity relationships and text models of animation and games to generate script text, thereby generating animation and game plot data.

6. The method according to claim 5, characterized in that, The script text generation process calculates the script text error loss using a formula, which is as follows: W represents the allowable value for the difference in text length generated by the target. p To generate text statistical relational data, s i The weighting coefficients for generating exact match probability values ​​of the text, c is the constant term for generating the text, and p i To generate a text exact match probability value, i takes the value 1, 2, 3...n, where n is the number of text exact match probability values ​​generated, and u is the correction coefficient for the allowable difference in length between the target and generated text.

7. The method according to claim 1, characterized in that, Step S2 is as follows: Based on animation and game design data, design image features are extracted to generate design image feature data, which includes design image texture feature data, design image attribute feature data, and design image shape feature data. Symbol detection and labeling are performed based on the design image feature data to generate design element feature data; Based on the plot data of anime and games, plot emotional features and character positioning features are extracted to generate plot emotional feature data and character positioning feature data.

8. The method according to claim 1, characterized in that, Step S3 is as follows: Feature preprocessing is performed based on the design element feature data to generate design element feature preprocessing data; Neural network mapping calculations are performed based on the preprocessed data of design element features to generate hidden layer data of design elements; Random data sampling is performed based on the hidden layer data of the design elements to obtain random sampling data of the hidden layer of the design elements; The design elements are randomly sampled from the hidden layer and then decoded and reconstructed to build an animation and game character generation model. Error correction processing is performed on the animation and game character generation model using plot emotion feature data and character positioning feature data, thereby generating animation and game character data.

9. The method according to claim 1, characterized in that, Step S4 is as follows: By performing temporal correlation on plot emotional feature data and character positioning feature data, temporal correlation feature data is generated. Character attributes are predicted and generated based on time-series correlation feature data, thereby generating character attribute data for animation and game intelligent production systems to process character interactions.

10. An intelligent production and management system for animation and games, characterized in that, The system includes: At least one processor; and, A memory that is communicatively connected to the at least one processor; The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform an intelligent animation and game production and management method as described in any one of claims 1 to 9.

Images