An animation script automatic generation method, system, storage medium and program product

Abstract

An animation script automatic generation method, system, storage medium and program product, in the method, a scene element set, a plot element set and a character element set in static text are extracted; an element association graph containing weights is generated; the content matching degrees of each element and each plot unit in the element association graph are calculated; a core plot unit is selected; the semantic vector cosine values between the core plot units are calculated as semantic association degrees, and the semantic association degrees are used as connection weights to construct a plot unit association network; the plot development main lines between the core plot units are determined; an extended core plot unit is determined; according to the weight distribution of the character elements in the element association graph, corresponding characters are assigned to the extended core plot unit, and character dialogue content is generated, to obtain a final core plot unit; and the final core plot unit is integrated according to a preset animation script format to generate an animation script text. The application improves the relevance of the generated script and the required result of the user.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of animation game development, and particularly relates to an animation script automatic generation method and system, a storage medium and a program product. BACKGROUND

[0002] With the rapid development of the animation industry, the demand for animation script creation is increasing, and the traditional manual writing of animation scripts not only consumes time and effort, but also is difficult to meet the large-scale and personalized creation demand. Especially in the era of short videos, a large number of creators need to quickly generate animation content, and the manual script writing method has become a bottleneck restricting the improvement of animation creation efficiency.

[0003] In the related art, a plurality of animation scene templates and character dialogue templates can be preset, a corresponding template is matched according to a theme keyword input by a user, and an animation script is automatically generated through a rule replacement method. This method improves the generation efficiency of the animation script to a certain extent and reduces the creation threshold.

[0004] However, this generation method based on fixed templates relies too much on preset rules and is difficult to accurately understand and extend the creative intention of the user. In particular, when the user provides scattered and abstract creative materials or inspiration fragments, it is difficult to integrate these creative elements into a coherent and expressive animation plot, resulting in a reduced relevance between the generated script and the user's desired result. SUMMARY

[0005] The application provides an animation script automatic generation method, system, storage medium and program product for improving the relevance between the generated script and the user's desired result.

[0006] In a first aspect, the application provides an animation script automatic generation method, which performs semantic analysis on static text description in creative material input by a user, extracts a scene element set, a plot element set and a character element set, and the creative material includes the static text description and a static keyword set;

[0007] According to the semantic similarity between each keyword in the static keyword set and each element in the scene element set, the plot element set and the character element set, each element is assigned a weight to obtain an element association graph containing weights;

[0008] A preset animation script structure template is divided into a plurality of plot units, and the content matching degree of each element in the element association graph and each plot unit is calculated;

[0009] According to the content matching degree, each plot unit is sorted in descending order of importance, and the plot units ranked before a preset number in the descending order of importance are selected as core plot units;

[0010] A semantic vector included angle cosine value between the core plot units is calculated as a semantic correlation degree, and the semantic correlation degree is used as a connection weight to construct a plot unit correlation network;

[0011] A maximum weight path search algorithm is performed in the plot unit correlation network to obtain plot development main lines between the core plot units;

[0012] A preset narrative rule library is used to determine a time sequence relationship of the core plot units based on the plot development main lines, and to supplement scene descriptions and expand plot details of the core plot units to obtain expanded core plot units;

[0013] According to a weight distribution of a role element in the element correlation graph, a corresponding role is assigned to the expanded core plot unit, and role dialogue content is generated to obtain a final core plot unit;

[0014] The final core plot unit is integrated according to a preset animation script format to generate an animation script text.

[0015] By using the above technical solutions, the creative materials are subjected to semantic analysis to extract key elements, and a weighted element correlation graph is constructed based on the semantic similarity between the elements and keywords, so that the creative intention input by the user is accurately quantified. The content matching degree is used as an evaluation index to screen the core plot units, and the semantic correlation degree is used to construct a plot unit correlation network and perform a maximum weight path search, which not only ensures that the generated script content is highly consistent with the user's creativity, but also makes the plot development context more clear and coherent. The scene and plot are supplemented and expanded by the preset narrative rule library, and the dialogue content is assigned to the role weight, so that the generated animation script not only retains the creative core, but also has rich detail performance and complete dramatic structure, reducing the problems of loose script structure, harsh plot connection, and thin character portrayal, and fully restoring the user's creative intention, improving the artistic and completeness of the generated script, and improving the quality and practical value of the automatic generation of the animation script.

[0016] In combination with some embodiments of the first aspect, in some embodiments, the static text description in the creative material input by the user is subjected to semantic analysis to extract a scene element set, a plot element set and a role element set, specifically including:

[0017] The static text description is divided into a plurality of text segments;

[0018] The plurality of text segments are subjected to part-of-speech tagging and dependency syntax analysis to identify scene description sentences, plot description sentences and role description sentences in the text segments;

[0019] The scene description sentences are clustered based on preset semantic feature indicators to obtain the scene element set;

[0020] classify the plot description sentences based on a preset plot type model to obtain a plot element set;

[0021] extract features of the role description sentences based on a preset role feature dimension to obtain a role element set.

[0022] By adopting the above technical solutions, the part-of-speech tagging and dependency syntax analysis are performed on the static text description, and different types of description sentences are accurately identified. The preset semantic feature indicators, plot type model and role feature dimension are used to respectively analyze the scene, plot and role description sentences, so that the various elements extracted from the creative material have clear semantic attributes and structured expression forms. The accuracy and completeness of element extraction are improved, a standardized data basis is provided for subsequent weight calculation and plot unit generation, the semantic ambiguity and information loss problems are reduced, and the accuracy of creative material analysis is improved.

[0023] In some embodiments of the first aspect, in some embodiments, according to the semantic similarity between each keyword in the static keyword set and each element in the scene element set, the plot element set and the role element set, the weight of each element is assigned to obtain an element association graph containing weights, specifically including:

[0024] A keyword-element bipartite graph is constructed, and the nodes of the keyword-element bipartite graph include keyword nodes in the static keyword set and element nodes in each element set;

[0025] The word vector cosine similarity between each keyword node and each element node is calculated as an edge weight;

[0026] The importance score of each element node in the keyword-element bipartite graph is determined;

[0027] The importance score is used as the weight value of each element to construct an element association graph.

[0028] By adopting the above technical solutions, the keyword-element bipartite graph is constructed, the explicit association relationship between the keywords and various elements is established, and the word vector cosine similarity is used as a quantitative evaluation index, considering the direct semantic association between the keywords and the elements. The core degree of the elements in the overall creative is also reflected through the node importance score. The element association graph constructed in this way has a reasonable weight distribution, accurately reflecting the importance and mutual relationship of each element in the creative expression. The problem of subjective and arbitrary weight allocation is reduced, so that the subsequent plot unit generation process can better grasp the creative focus and maintain the primary and secondary distinction of the content.

[0029] In some embodiments of the first aspect, before calculating the semantic vector cosine value between the core plot units as the semantic association degree, the method further includes:

[0030] construct a multi-dimensional emotion feature space of the core plot unit, the multi-dimensional emotion feature space including tension, joy, and sadness;

[0031] extract an emotion vector based on text content of the core plot unit, and map the emotion vector into the multi-dimensional emotion feature space;

[0032] calculate an emotion trajectory curve of each core plot unit in the multi-dimensional emotion feature space;

[0033] adjust the emotion trajectory curve according to a preset emotion fluctuation pattern.

[0034] By adopting the technical solution, the multi-dimensional emotion feature space is introduced to perform emotion modeling on the core plot unit, the emotion vector is extracted, and the emotion trajectory curve is calculated, so that the accurate depiction of the plot emotion change is realized. The emotion trajectory curve is adjusted based on the preset emotion fluctuation pattern, so that the generated plot is more consistent with the artistic rules in emotion expression. The subsequent generated plot development main line is improved in logic continuity, and the rationality of the fluctuation change in the emotion level is improved.

[0035] In combination with some embodiments of the first aspect, in some embodiments, the multi-dimensional emotion feature space of the core plot unit is constructed, and specifically includes:

[0036] obtain emotion annotation data in a preset animation script corpus;

[0037] perform dimension reduction processing on the emotion annotation data, and extract tension, joy, and sadness;

[0038] calculate an independence coefficient between the tension, the joy, and the sadness;

[0039] construct a quantitative scoring standard based on the independence coefficient between the tension, the joy, and the sadness;

[0040] apply the quantitative scoring standard to the tension, the joy, and the sadness, and construct the multi-dimensional emotion feature space.

[0041] By adopting the technical solution, the emotion annotation data in the preset animation script corpus is obtained and dimension reduction processing is performed, three basic dimensions of tension, cheerfulness and sadness are extracted, the independence coefficients between the three dimensions are calculated, and a quantitative scoring standard is constructed accordingly, and finally a multi-dimensional emotion feature space is constructed. The uncertainty caused by subjective experience judgment is reduced. The orthogonality between the emotion dimensions is ensured through the calculation of the independence coefficients, so that each dimension can independently reflect the emotion characteristics of the plot in one aspect, and the information redundancy between the dimensions is reduced. The quantitative scoring standard constructed based on the large-scale corpus data has strong statistical reliability and universality, can accurately capture the emotion characteristic differences of different types of plot units, and makes the emotion analysis result comparable and repeatable, thereby improving the scientificity and accuracy of the emotion design of the animation script.

[0042] In combination with some embodiments of the first aspect, in some embodiments, after adjusting the emotion trajectory curve according to the preset emotion fluctuation pattern, the method further comprises:

[0043] Based on the emotion trajectory curve, significant feature points of emotion changes are extracted, and the emotion span and duration between adjacent significant feature points are calculated to obtain an emotion fluctuation feature sequence;

[0044] Based on the emotion fluctuation feature sequence, a neural oscillator network is constructed, the frequency of each oscillator node in the neural oscillator network corresponds to the duration, and the amplitude corresponds to the emotion span;

[0045] The neural oscillator network is used to model the emotion rhythm to generate a rhythm feature vector;

[0046] The rhythm change of the emotion trajectory curve is determined according to the golden section ratio and the rhythm feature vector.

[0047] By adopting the technical solution, the emotion span and duration between the significant feature points of emotion changes are respectively mapped to the amplitude and frequency parameters of the oscillator nodes by using the neural oscillator network to model the emotion rhythm, so that the complex emotion fluctuation can be expressed and analyzed by a simple oscillation model. The neural oscillator network has good dynamic characteristics and can simulate the continuity and periodicity of the emotion fluctuation, reducing the limitations of the traditional discrete sampling method in describing continuous emotion changes. The rhythm feature vector is generated by quantitatively modeling the emotion rhythm, so that the control of the emotion rhythm is more accurate and flexible, and the accuracy and fluency of the animation script in the emotion rhythm control are improved.

[0048] In combination with some embodiments of the first aspect, in some embodiments, the rhythm change of the emotion trajectory curve is determined according to the golden section ratio and the rhythm feature vector, specifically comprising:

[0049] The total time length of the emotional trajectory curve is divided into multiple time segments according to the golden section ratio;

[0050] An energy distribution of the rhythm feature vector in each time segment is calculated, and a rhythm adjustment function is constructed based on the energy distribution, the rhythm adjustment function being used to amplify or suppress the emotional fluctuation intensity in each time segment;

[0051] The rhythm adjustment function is applied to the emotional trajectory curve to obtain an optimized emotional rhythm change.

[0052] By adopting the technical solution, the total time length of the emotional trajectory curve is divided according to the golden section ratio, and the rhythm adjustment function is constructed based on the energy distribution of the rhythm feature vector, so that the emotional fluctuation intensity in each time segment is accurately controlled. The golden section ratio is a classical artistic composition rule, which can create natural and harmonious rhythm in the time dimension. By analyzing the energy distribution of the rhythm feature vector in each time segment, the rhythm adjustment function can amplify or suppress the emotional fluctuation intensity in different segments, so that the overall emotional curve not only conforms to the aesthetic rule but also is rich in changes, improves the artistic nature of the emotional rhythm, and enhances the narrative tension and viewing experience of the animation script.

[0053] In a second aspect, the embodiments of the present application provide an animation script automatic generation system, which comprises one or more processors and a memory; the memory is coupled with the one or more processors, and is used to store computer program codes, the computer program codes comprising computer instructions, and the one or more processors invoke the computer instructions to enable the system to perform the method described in the first aspect and any possible implementation manner of the first aspect.

[0054] In a third aspect, the embodiments of the present application provide a computer readable storage medium, which comprises instructions, and when the instructions run on a system, enable the system to perform the method described in the first aspect and any possible implementation manner of the first aspect.

[0055] In a fourth aspect, the embodiments of the present application provide a computer program product, which is characterized by enabling the system to perform the method described in any possible implementation manner of the first aspect when the computer program product runs on the system.

[0056] The one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0057] 1. The application provides a method for automatically generating an animation script, which performs semantic analysis on creative materials to extract key elements, and constructs a weighted element association graph based on the semantic similarity between the elements and keywords, so that the user's creative intention can be accurately quantified. The content matching degree is used as an evaluation index to screen the core plot units, and then the semantic association degree is used to construct a plot unit association network and perform maximum weight path search, which not only ensures that the generated script content is highly consistent with the user's creativity, but also makes the plot development more clear and coherent. The scene and plot are supplemented and expanded by the pre-set narrative rule library, and the dialogue content is combined with the role weight distribution, so that the generated animation script has rich detail performance and complete dramatic structure while maintaining the creative core, reducing the problems of loose script structure, harsh plot connection and thin character portrayal, and fully restoring the user's creative intention, improving the artistic and integrity of the generated script, and improving the quality and practical value of the automatic generation of animation scripts.

[0058] 2. The application provides a method for automatically generating an animation script, which introduces a multi-dimensional emotional feature space to model the core plot units, extracts emotional vectors and calculates emotional trajectory curves to accurately depict the emotional changes of the plot. Based on the pre-set emotional fluctuation mode, the emotional trajectory curve is adjusted, so that the generated plot is more consistent with the artistic rules in emotional expression. The logical coherence of the subsequent generated plot development main line is improved, and the rationality of the fluctuation change in the emotional level is improved.

[0059] 3. The application provides a method for automatically generating an animation script, which uses a neural oscillator network to model the emotional rhythm, maps the emotional span and duration between the significant feature points of emotional changes to the amplitude and frequency parameters of the oscillator nodes respectively, so that complex emotional fluctuations can be expressed and analyzed by a simple oscillation model. The neural oscillator network has good dynamic characteristics and can simulate the continuity and periodicity of emotional fluctuations, reducing the limitations of traditional discrete sampling methods in describing continuous emotional changes. By quantitatively modeling the emotional rhythm to generate rhythm feature vectors, the control of emotional rhythm is more accurate and flexible, improving the accuracy and fluency of the animation script in emotional rhythm control. BRIEF DESCRIPTION OF DRAWINGS

[0060] Figure 1 is a flowchart of a method for automatically generating an animation script in an embodiment of the application.

[0061] Figure 2 is another flowchart of a method for automatically generating an animation script in an embodiment of the application.

[0062] Figure 3is a schematic structural diagram of an entity device of an animation script automatic generation system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0063] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments and are not intended to be limiting of the present application. As used in the specification and the appended claims of the present application, the singular forms "a," "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "and / or," as used in the present application, refers to any or all possible combinations of one or more of the associated listed items.

[0064] Hereinafter, the terms "first" and "second" are only for the purpose of description and cannot be understood as implying or suggesting relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first" and "second" can explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, the meaning of "a plurality of" is two or more, unless otherwise specified.

[0065] Hereinafter, an embodiment is used and combined with Figure 1 An animation script automatic generation method in an embodiment of the present application is described.

[0066] Please refer to Figure 1 is a flowchart of an animation script automatic generation method in an embodiment of the present application.

[0067] S101, semantic analysis is performed on static text description in creative material input by a user, and scene element set, plot element set and character element set are extracted;

[0068] The system performs semantic analysis on static text description in creative material input by a user, and extracts scene element set, plot element set and character element set, and the creative material includes static text description and static keyword set. Specifically, the static text description is divided into a plurality of text segments;

[0069] The plurality of text segments are subjected to part-of-speech tagging and dependency syntax analysis, and scene description sentences, plot description sentences and character description sentences in the text segments are identified;

[0070] The scene description sentences are clustered based on preset semantic feature indicators, and the scene element set is obtained;

[0071] The plot description sentences are classified based on a preset plot type model, and the plot element set is obtained;

[0072] The role element set is obtained by performing feature extraction on the role description sentence based on a preset role feature dimension.

[0073] In this step, the system processes and analyzes the creative material input by the user. The creative material includes two parts: static text description and static keyword set. The system mainly performs semantic analysis on the static text description, aiming to extract key information related to the animation script from the text, i.e., scene element set, plot element set, and role element set. These three element sets provide basic data for subsequent generation of animation scripts.

[0074] Specifically, the system can implement this step using the following technical solutions: First, the static text description is divided into several text segments to facilitate subsequent semantic analysis. Then, part-of-speech tagging and dependency syntax analysis are performed on each text segment to identify scene description sentences, plot description sentences, and role description sentences. Next, the system can cluster the scene description sentences based on preset semantic feature indicators to obtain the scene element set; classify the plot description sentences based on a preset plot type model to obtain the plot element set; and perform feature extraction on the role description sentences based on a preset role feature dimension to obtain the role element set.

[0075] S102, according to the semantic similarity between each keyword in the static keyword set and each element in the scene element set, plot element set, and role element set, weight assignment is performed on each element to obtain an element association graph containing weights;

[0076] The system assigns weights to each element based on the semantic similarity between each keyword in the static keyword set and each element in the scene element set, plot element set, and role element set to obtain an element association graph containing weights. Specifically, a keyword-element bipartite graph is constructed, and the nodes of the keyword-element bipartite graph include keyword nodes in the static keyword set and element nodes in each element set;

[0077] The word vector cosine similarity between each keyword node and each element node is calculated as the edge weight;

[0078] The importance score of each element node in the keyword-element bipartite graph is determined;

[0079] The importance score is used as the weight value of each element to construct an element association graph.

[0080] In this step, the system uses the static keyword set to evaluate the importance of the extracted scene elements, plot elements, and role elements, and constructs an element association graph. The static keyword set reflects the user's focus and key points on the content of the animation script. By calculating the semantic similarity between keywords and elements, the importance of the elements can be quantitatively evaluated.

[0081] Specifically, the system can implement this step using the following technical solution: First, construct a keyword element bipartite graph, where nodes include keyword nodes in the static keyword set and element nodes in each element set, and the edge weights are the cosine similarity of word vectors between keyword nodes and element nodes. Then, calculate the importance score of each element node in the bipartite graph, using it as the element's weight value. Finally, construct an element association graph based on the element weight values ​​to reflect the importance relationships between elements.

[0082] In practical applications, relying solely on word vector similarity to assess element importance may have limitations, failing to fully consider the element's role and impact within the overall storyline. To address this issue, the system can incorporate story structuring and element relationship reasoning techniques. By structurally parsing the text description, a storyline graph is constructed, and element relationship reasoning is performed on this graph to obtain more comprehensive and accurate element importance assessment results. For example, the system can associate core plot nodes and key plot developments in the storyline graph with element nodes, calculating the importance of element nodes in plot progression; by analyzing the causal and logical relationships between element nodes in the storyline graph, the importance of element nodes in the story's causal chain is inferred. In this way, based on static keyword matching, the system can fully utilize story structure and plot logic to obtain a more reasonable and accurate element weight allocation, providing a more reliable basis for the generation of subsequent animation scripts.

[0083] S103. Divide the preset animation script structure template into several plot units, and calculate the content matching degree between each element in the element association diagram and each plot unit.

[0084] In this step, the system uses a pre-defined animation script structure template to divide and arrange the overall structure of the animation script. The animation script structure template defines the basic components and progression path of a typical animated story, such as opening, development, climax, and ending. The system divides the structure template into several plot units, each corresponding to a paragraph or scene in the animation script. Then, the system calculates the degree of matching between the scene elements, plot elements, and character elements in the element association diagram and the content of each plot unit, providing a basis for subsequently selecting core plot units.

[0085] Specifically, the system can implement this step by the following technical solution: first, the preset animation script structure template is divided according to fixed rules and formats, and a plurality of plot units are obtained. Each plot unit contains a certain plot and scene content, and is consistent with the development context of the whole story. Then, the system calculates the content matching degree of each element and each plot unit by using the semantic features of each element in the element association graph, and obtains the content matching degree of each element and each plot unit. The higher the content matching degree, the stronger the relevance of the element and the plot unit, and the greater the possibility of appearing in the plot unit.

[0086] S104, according to the content matching degree, the importance of each plot unit is sorted in descending order, and the plot units ranked in the top of the preset number are selected as the core plot units;

[0087] In this step, the system evaluates and sorts the importance of all plot units according to the content matching degree calculated in the previous step. The higher the content matching degree of a plot unit, the higher its importance, and the more critical role it plays in the animation script. The system sorts the plot units in descending order of importance according to the content matching degree, and then selects a preset number of plot units ranked at the top as the core plot units of the animation script. These core plot units constitute the main line of the animation script and play a decisive role in the development and advancement of the plot.

[0088] Specifically, the system can implement this step by the following technical solution: first, according to the content matching degree of each plot unit and each element in the element association graph, the comprehensive importance score of each plot unit is calculated. The importance score can be the weighted average of the content matching degrees of all related elements, and the weight can be determined according to the importance of the element in the element association graph. Then, the system sorts all plot units in descending order according to the importance score to obtain an importance ranking list. Finally, the system selects a preset number (for example, the top 10) of plot units from the ranking list as the core plot units.

[0089] S105, calculate the semantic vector cosine value between the core plot units as the semantic correlation degree, and construct a plot unit association network with the semantic correlation degree as the connection weight;

[0090] In this step, the system analyzes the semantic association relationship between the core plot units selected in the previous step, and constructs a plot unit association network. The semantic association relationship between plot units reflects the coherence of the plot and the causality of the plot, which is crucial to ensure the logical rationality and smoothness of the story. The system calculates the semantic correlation degree between the core plot units, and constructs a weighted plot unit association network by taking the semantic correlation degree as the weight of the connection between plot units.

[0091] Specifically, the system can implement this step by the following technical solution: first, for each core plot unit, extract its semantic feature vector of the text content. The semantic feature vector can be obtained using word vectors, sentence vectors and other natural language processing techniques, reflecting the position and direction of the plot unit in the semantic space. Then, for each pair of core plot units, calculate the cosine value between their semantic feature vectors as the semantic correlation between the two plot units. The larger the cosine value, the more similar the two plot units are in semantics, and the stronger the plot correlation between them. Finally, the system constructs a plot unit correlation network with core plot units as nodes and semantic correlation as edge weights.

[0092] S106, performing a maximum weight path search algorithm in the plot unit correlation network to obtain plot development main lines between core plot units;

[0093] In this step, the system uses the plot unit correlation network constructed in the previous step to find a path that connects all core plot units and has the maximum weight, as the plot development main line of the animation script, through the maximum weight path search algorithm. The plot development main line determines the plot direction and story context of the animation script, and is the skeleton and outline of the entire animation story. Along the plot development main line, the core plot units are unfolded in a certain order, forming a complete and coherent animation narrative structure.

[0094] Specifically, the system can implement this step by the following technical solution: first, consider the plot unit correlation network as a weighted directed graph, where nodes represent core plot units and edge weights represent the correlation between adjacent plot units. Then, the system performs a maximum weight path search algorithm, such as Dijkstra's algorithm or Bellman-Ford's algorithm, on this graph to find an optimal path from the starting node (usually the plot unit at the beginning of the story) to the ending node (usually the plot unit at the end of the story) with the maximum weight on the path. This optimal path is the plot development main line of the animation script, which reflects the best arrangement and combination of core plot units in the plot progression.

[0095] S107, determining the chronological relationship of each core plot unit based on the plot development main line, and supplementing scene descriptions and expanding plot details of each core plot unit according to the preset narrative rule library, to obtain expanded core plot units;

[0096] In this step, the system determines the time and logical sequence of the core plot units according to the plot development main line obtained in the previous step, and enriches and perfects the content of each core plot unit to obtain more complete and vivid extended core plot units. The extended core plot units not only contain the original core plot content, but also supplement more scene descriptions and plot details, making the animation script more full, three-dimensional, and attractive.

[0097] Specifically, the system can implement this step by using the following technical solutions: first, according to the arrangement order of the core plot units in the plot development main line, determine their time sequence relationship in the animation script, and form an ordered plot unit sequence. Then, for each core plot unit, the system matches the corresponding scene description template and plot detail template from the preset narrative rule library, and expands the content of the core plot unit. The narrative rule library contains a large number of scene description materials and plot detail materials, covering various common animation narrative scenes and segments. The system can retrieve the most relevant description templates and detail templates according to the semantic features and keywords of the core plot unit, and fuse them with the core plot unit to generate extended core plot units with richer content, more specific description, and more vivid details.

[0098] In practical applications, the content expansion method based on fixed templates may have difficulty in generating innovative and diverse animation script content, and may easily produce homogenization and stereotyped problems. In order to improve the originality and attractiveness of the animation script, the system can introduce deep learning-based natural language generation technology, train a large-scale animation script corpus, learn the language style and creative ideas of animation narration, and realize flexible, diverse, and creative plot unit expansion. For example, the system can use a sequence-to-sequence neural network model to automatically generate scene descriptions and plot details that match the key information of the core plot unit; for different types and styles of animation scripts, the system can train multiple specialized generation models to capture the narrative characteristics and creative techniques of different genres and themes; in addition, the system can use reinforcement learning algorithms to constantly optimize and adjust the parameters of the generation model according to user feedback and evaluation, so that it can generate animation script content that is closer to user needs and aesthetic preferences. In this way, the system can break through the traditional template expansion method, create content-rich, diverse, creative, and attractive extended core plot units through intelligent natural language generation technology, and provide users with higher quality and personalized animation script creation services.

[0099] S108, according to the weight distribution of the role elements in the element association graph, assigning corresponding roles to the extended core plot units and generating role dialogue content to obtain final core plot units;

[0100] In this step, the system assigns appropriate roles to each extended core plot unit based on the role element information in the element association graph and generates corresponding dialogue content for these roles, ultimately obtaining a complete and full core plot unit. Characters are one of the core elements in an animation story, and the setting and dialogue of characters are key means to shape the image of characters and drive the development of the plot. By reasonably assigning roles to the plot unit and generating character dialogue, the animation script can be more readable, immersive, and artistically compelling.

[0101] Specifically, the system can implement this step by using the following technical solutions: First, according to the weight values of the role element nodes in the element association graph, determine the importance and frequency of appearance of each character in the animation story. The higher the weight value of the character, the more important the role and function in the animation script. Then, the system traverses each extended core plot unit, matches several most relevant characters according to its plot content and scene description, and assigns them in order from high to low according to the weight of the characters. For each character appearing, the system automatically generates dialogue content that meets the role identity and plot requirements according to its character attributes, emotional state, and situational factors, combined with preset dialogue generation rules and templates. Character dialogue not only includes the text content of the lines, but also includes performance instructions such as tone, emotion, and action. Finally, the system integrates the generated character dialogue information with the extended core plot unit to obtain a final core plot unit containing complete plot description and character interaction.

[0102] S109, integrating the final core plot unit according to the preset animation script format to generate an animation script text.

[0103] In this step, the system arranges all the final core plot units obtained in the previous step in chronological order, and integrates and standardizes them according to the preset animation script format, finally outputting a complete and standardized animation script text file. Animation script format is a set of text organization and presentation standards recognized in the field of animation creation, and using a unified animation script format can make the script text more professional and standardized, and also facilitate the subsequent shot design, animation production, and other links.

[0104] Specifically, the system can implement this step by adopting the following technical solution: first, arrange all the final core plot units in sequence according to the chronological order determined by the plot development main line to form a complete animated story sequence. Then, the system performs structured processing and rich text marking on the content of each plot unit according to the preset animated script format template. For example, in the script format, each scene usually includes scene title, scene number, scene description, character dialogue, voice-over commentary, and shot instruction, etc. The system needs to identify and extract these elements from the text content of the plot unit and arrange them according to the standard format. At the same time, the system also needs to perform grammar checking, word polishing, and layout formatting on the script text to ensure that the language quality and form of the script meet professional standards. Finally, output the integrated animated script content as a structured text file to complete the automatic generation of the animated script.

[0105] In the above embodiment, the creative material is subjected to semantic analysis to extract key elements, and a weighted element association graph is constructed based on the semantic similarity between the elements and keywords, so that the user's creative intention is accurately quantified. The content matching degree is used as an evaluation index to screen the core plot units, and the semantic association degree is used to construct a plot unit association network and perform maximum weight path search, which not only ensures that the generated script content is highly consistent with the user's creativity, but also makes the plot development context more clear and coherent. By using the preset narrative rule library to supplement and expand the scenes and plots, and combining the role weight allocation with the dialogue content, the generated animated script not only maintains the creative core but also has rich detail performance and complete dramatic structure, reducing problems such as loose script structure, harsh plot transitions, and thin character portrayal, and can fully restore the user's creative intention, improve the artistic and completeness of the generated script, and improve the quality and practical value of the automatic generation of animated scripts.

[0106] To further optimize the emotional expression effect of the generated script and solve the problems of monotonous emotional rhythm and harsh emotional transition in traditional automatic generation methods, the present application also realizes accurate modeling and optimization control of script emotional changes by constructing a multi-dimensional emotional feature space and applying a neural oscillator network. The following describes the implementation of the present application in combination with Figure 2 Another animated script automatic generation method in the embodiment of the present application is described as follows:

[0107] Please refer to Figure 2 Another flowchart of the animated script automatic generation method in the embodiment of the present application is shown in FIG. 6.

[0108] S201, a multi-dimensional emotional feature space of the core plot unit is constructed, and the multi-dimensional emotional feature space includes tension, cheerfulness, and sadness;

[0109] The system constructs a multi-dimensional emotion feature space for the core plot unit, which includes tension, joy, and sadness. Specifically, emotion annotation data in the preset animation script corpus is obtained;

[0110] The emotion annotation data is processed by dimension reduction to extract tension, joy, and sadness;

[0111] The independence coefficients between tension, joy, and sadness are calculated;

[0112] A quantitative scoring standard is constructed based on the independence coefficients between tension, joy, and sadness;

[0113] The quantitative scoring standard is applied to tension, joy, and sadness to construct a multi-dimensional emotion feature space.

[0114] In this step, the system constructs a multi-dimensional emotion feature space by analyzing and processing the emotion annotation data in the preset animation script corpus. The feature space contains three dimensions of tension, joy, and sadness, which are used to quantify and describe the emotional changes in the core plot unit. By extracting key features from the emotion annotation data and performing dimension reduction, the system can effectively capture and represent the differences and relationships between different emotional states.

[0115] To construct the multi-dimensional emotion feature space, the system first obtains the emotion annotation data in the preset animation script corpus and performs preprocessing and feature extraction. Then, by applying dimension reduction algorithms such as principal component analysis or independent component analysis, the extracted features are reduced to obtain three key emotional dimensions of tension, joy, and sadness. Next, the system calculates the independence coefficients between these three dimensions to measure their correlation and independence. Based on the independence coefficients, the system constructs a quantitative scoring standard that maps tension, joy, and sadness into a normalized numerical range, forming a multi-dimensional emotion feature space.

[0116] In the process of constructing the multi-dimensional emotion feature space, the system may face technical challenges in emotion dimension selection and quantitative scoring standard design. To solve these problems, the system can introduce more emotional dimensions such as surprise, disgust, etc. to capture more rich and delicate emotional changes. At the same time, the system can use machine learning algorithms such as support vector machines or neural networks to automatically learn and optimize the quantitative scoring standard, improving the representation ability and accuracy of the emotion feature space. Through continuous iteration and optimization, the system can construct a more perfect and effective multi-dimensional emotion feature space, providing a solid foundation for subsequent emotion analysis and script generation.

[0117] S202, extract an emotion vector based on the text content of the core plot unit, and map the emotion vector into the multi-dimensional emotion feature space;

[0118] The system extracts emotion vectors from the text content of the core plot units and maps these emotion vectors into a multi-dimensional emotion feature space.

[0119] In this step, the system analyzes the text content of the core plot units, extracts emotion vectors that represent the emotional characteristics of these units, and maps them to the previously constructed multi-dimensional emotion feature space. An emotion vector is a mathematical representation used to quantify and describe the emotional information contained in the text. By mapping emotion vectors to the multi-dimensional emotion feature space, the system can establish a direct link between the text content and the corresponding emotional state, facilitating subsequent emotion analysis and optimization.

[0120] To extract sentiment vectors, the system can employ various natural language processing techniques, such as sentiment dictionaries, sentiment dependency analysis, and deep learning. First, the system can construct a sentiment dictionary containing a rich vocabulary of sentiment terms and use this dictionary to annotate the text content with sentiment. Then, through sentiment dependency analysis, the system can identify sentiment keywords and their dependencies in the text, thereby inferring the overall sentiment tendency of the text. Furthermore, the system can apply deep learning models, such as recurrent neural networks or attention mechanisms, to automatically learn the sentiment representation of the text, capturing contextual information and semantic connections. Finally, the system maps the extracted sentiment vectors into a multi-dimensional sentiment feature space, obtaining a quantitative representation of the text in dimensions such as tension, joy, and sadness.

[0121] In the process of extracting and mapping sentiment vectors, the system may face issues of sentiment ambiguity and context dependence. To address these issues, the system can introduce sentiment disambiguation techniques, analyzing the sentiment tendencies of words in different contexts to eliminate ambiguity and improve the accuracy of sentiment vectors. Simultaneously, the system can employ context-aware sentiment analysis methods, considering semantic information before and after the text to capture dynamic changes in sentiment. Furthermore, the system can utilize transfer learning and domain adaptation techniques to apply sentiment analysis models trained in other domains to the sentiment analysis of animation scripts, improving the generalization ability and robustness of sentiment vector extraction.

[0122] S203. Calculate the emotional trajectory curve of each core plot unit in the multidimensional emotional feature space;

[0123] In this step, the system calculates the emotional trajectory curves of the core plot units based on their mapping positions in a multi-dimensional emotional feature space. An emotional trajectory curve is a mathematical representation describing the process of emotional change, reflecting the dynamic trends of plot units in dimensions such as tension, joy, and sadness. By analyzing the emotional trajectory curves, the system can grasp the emotional direction of the entire story, providing a basis for subsequent emotional optimization and pacing adjustments.

[0124] To calculate the emotional trajectory curve, the system first needs to determine the position coordinates of the core plot units in the multi-dimensional emotional feature space. This can be achieved by mapping the emotional vector of each plot unit to the corresponding coordinate point. Then, the system connects these coordinate points in the order of the plot units to form a continuous curve, which is the emotional trajectory curve. The system can use curve fitting algorithms such as spline interpolation or Bezier curve to smooth and optimize the emotional trajectory curve, making it more natural and coherent.

[0125] S204, adjust the emotional trajectory curve according to the preset emotional fluctuation pattern.

[0126] In this step, the system adjusts and optimizes the previously calculated emotional trajectory curve according to the preset emotional fluctuation pattern. The emotional fluctuation pattern is a template that describes the ideal emotional change rule, which reflects the general rule and rhythm arrangement of the plot development. By matching and adjusting the emotional trajectory curve with the preset emotional fluctuation pattern, the system can make the generated animation script more reasonable, lively and engaging in emotional expression.

[0127] To adjust the emotional trajectory curve, the system first needs to build a preset emotional fluctuation pattern library, which contains various typical emotional change patterns such as gradual rise, fluctuation, climax and so on. Then, the system matches the emotional trajectory curve with the patterns in the pattern library to find the most similar pattern. Next, the system deforms and adjusts the emotional trajectory curve according to the matched pattern, making it more consistent with the ideal emotional fluctuation rule. The adjustment process can be achieved through various mathematical transformations and optimization algorithms, such as scaling, translation, rotation, etc.

[0128] In the process of adjusting the emotional trajectory curve, the system may face the problems of inaccurate pattern matching and unreasonable adjustment amplitude. To solve these problems, the system can use a multi-scale pattern matching strategy to match at different time scales, capturing local and global features of emotional changes. At the same time, the system can introduce an adaptive adjustment strategy to dynamically adjust the adjustment amplitude and method according to the characteristics of the emotional trajectory curve and the plot requirements, avoiding excessive changes to the original emotional trend. In addition, the system can use reinforcement learning algorithms to learn and optimize the adjustment strategy of the emotional trajectory curve through continuous trial and feedback, making the generated animation script more natural, smooth and attractive in emotional expression.

[0129] After that, the system extracts the salient feature points of emotional changes based on the emotional trajectory curve, and calculates the emotional span and duration between adjacent salient feature points to obtain the emotional fluctuation feature sequence;

[0130] A neural oscillator network is constructed based on the emotion fluctuation feature sequence, the frequency of each oscillator node in the neural oscillator network corresponds to the duration, and the amplitude corresponds to the emotion span;

[0131] The neural oscillator network is used to model the emotion rhythm, and a rhythm feature vector is generated;

[0132] The rhythm change of the emotion trajectory curve is determined according to the golden section ratio and the rhythm feature vector, specifically: the total duration of the emotion trajectory curve is divided into multiple time segments according to the golden section ratio;

[0133] The energy distribution of the rhythm feature vector in each time segment is calculated, and a rhythm adjustment function is constructed based on the energy distribution, the rhythm adjustment function is used to amplify or suppress the emotion fluctuation intensity in each time segment;

[0134] The rhythm adjustment function is applied to the emotion trajectory curve to obtain an optimized emotion rhythm change.

[0135] In the above embodiment, a multi-dimensional emotion feature space is introduced to model the emotions of the core plot unit, the emotion vector is extracted and the emotion trajectory curve is calculated to accurately depict the emotional changes of the plot. The emotion trajectory curve is adjusted based on the preset emotion fluctuation mode, so that the generated plot is more consistent with the artistic rules in emotional expression. The logical coherence of the subsequent generated plot development main line is improved, and the rationality of the fluctuation change in the emotional level is improved.

[0136] The system in the embodiment of the present application will be described from the perspective of hardware processing. Please refer to Figure 3 An entity device structure diagram of an animation script automatic generation system provided by the embodiment of the present application.

[0137] It should be noted that, Figure 3 The structure of the system shown is only an example, and should not bring any limitation to the function and use range of the embodiment of the present application.

[0138] As Figure 3As shown, the system includes a Central Processing Unit (CPU) 301 which can perform various appropriate actions and processes, such as executing the methods in the above embodiments, according to programs stored in a Read-Only Memory (ROM) 302 or loaded from a storage section 308 into a Random Access Memory (RAM) 303. In the RAM 303, various programs and data required for operation of the system are also stored. The CPU 301, the ROM 302, and the RAM 303 are connected to each other through a bus 304. An Input / Output (I / O) interface 305 is also connected to the bus 304.

[0139] Connected to the I / O interface 305 are an input section 306 including a camera, an infrared sensor, and the like; an output section 307 including a Liquid Crystal Display (LCD), a speaker, and the like; the storage section 308 including a hard disk, and the like; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card, a modem, and the like. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as necessary. A removable media 311 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, and the like is attached to the drive 310 as necessary, so that a computer program read therefrom is installed into the storage section 308 as necessary.

[0140] In particular, according to embodiments of the present application, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, embodiments of the present application include a computer program product comprising a computer program carried on a computer readable medium, the computer program containing a computer program for executing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network by the communication section 309, and / or installed from the removable media 311. When the computer program is executed by the Central Processing Unit (CPU) 301, various functions defined in the present application are performed.

[0141] It should be noted that the computer-readable medium in the embodiments of the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the two. The computer-readable storage medium may, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any combination of the above. More specific examples of the computer-readable storage medium can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (Compact Disc Read-Only Memory, CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or apparatus. In the present application, the computer-readable signal medium can include a data signal carried in a baseband or as a part of a carrier wave, which carries computer-readable computer programs. Such a propagated data signal can take many forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above.

[0142] The flowcharts and block diagrams in the drawings illustrate the possible implementation architectures, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. Each block in the flowcharts or block diagrams can represent a module, a program segment, or a part of code containing one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur in different orders than that shown in the drawings. For example, two blocks that are shown in succession can actually be executed substantially in parallel, and sometimes in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams or flowcharts, and the combination of blocks in the block diagrams or flowcharts, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0143] As another aspect, the present application also provides a computer readable storage medium, which can be included in the system described in the above embodiments, or can exist independently without being assembled into the system. The above storage medium carries one or more computer programs, which, when executed by a processor of a system, enable the system to implement the method provided in the above embodiments.

[0144] The above embodiments are only used to illustrate the technical solutions of the present application, but not limit the present application; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still make modifications to the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0145] In the above embodiments, according to the context, the term "when" can be interpreted as meaning "if" or "after" or "in response to determining" or "in response to detecting". Similarly, according to the context, the phrase "upon determining" or "if detecting (the stated condition or event)" can be interpreted as meaning "if determining" or "in response to determining" or "upon detecting (the stated condition or event)" or "in response to detecting (the stated condition or event)".

[0146] In the above embodiments, all or some of the embodiments can be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or some of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or some of the processes or functions described in the embodiments of the present application are generated. The computer can be a general purpose computer, a special purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer readable storage medium, or transmitted from one computer readable storage medium to another computer readable storage medium, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through wired (such as coaxial cable, optical fiber, digital subscriber line) or wireless (such as infrared, wireless, microwave, etc.) mode. The computer readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. integrated with one or more available media. The available media can be magnetic media (for example, floppy disk, hard disk, magnetic tape), optical media (for example, DVD), or semiconductor media (for example, solid state disk) and the like.

[0147] Those skilled in the art can understand that all or part of the processes in the above-mentioned method embodiments can be implemented by a computer program instructing relevant hardware to complete, the program can be stored in a computer readable storage medium, and the program can include the processes of the above-mentioned method embodiments when executed. The aforementioned storage medium includes ROM or random storage memory RAM, magnetic disc or optical disc and various storage code medium.

Claims

1. An animation script automatic generation method characterized by, The method comprises the following steps: performing semantic analysis on static text description in creative material input by a user to extract a scene element set, a plot element set and a character element set, the creative material comprising the static text description and a static keyword set; assigning weights to each element in the scene element set, the plot element set and the character element set according to semantic similarity between each keyword in the static keyword set and each element, to obtain an element association graph containing weights; dividing a preset animation script structure template into a plurality of plot units and calculating content matching degrees between each element in the element association graph and each plot unit; ranking the plot units in descending order of importance according to the content matching degrees and selecting plot units ranking before a preset number as core plot units; constructing a multi-dimensional emotional feature space of the core plot units, the multi-dimensional emotional feature space comprising tension, joy and sadness; extracting an emotional vector based on text content of the core plot units and mapping the emotional vector into the multi-dimensional emotional feature space; calculating emotional trajectory curves of the core plot units in the multi-dimensional emotional feature space; adjusting the emotional trajectory curves according to a preset emotional fluctuation pattern; extracting salient feature points of emotional changes based on the emotional trajectory curves, calculating emotional spans and duration between adjacent salient feature points to obtain an emotional fluctuation feature sequence; constructing a neural oscillator network based on the emotional fluctuation feature sequence, a frequency of each oscillator node in the neural oscillator network corresponding to the duration and an amplitude corresponding to the emotional span; modeling emotional rhythm by using the neural oscillator network to generate a rhythm feature vector; determining rhythm changes of the emotional trajectory curves according to a golden section ratio and the rhythm feature vector; calculating a semantic correlation degree as a cosine value of a semantic vector between the core plot units and constructing a plot unit association network by taking the semantic correlation degree as a connection weight; performing a maximum weight path search algorithm in the plot unit association network to obtain plot development main lines between the core plot units; determining a time sequence relationship of the core plot units based on the plot development main lines and supplementing scene descriptions and expanding plot details of the core plot units according to a preset narration rule library to obtain expanded core plot units; allocating corresponding characters to the expanded core plot units and generating character dialogue content according to a weight distribution of character elements in the element association graph to obtain final core plot units; integrating the final core plot units according to a preset animation script format to generate an animation script text.

2. The method of claim 1, wherein, The semantic analysis on the static text description in the creative material input by the user specifically comprises: segmenting the static text description into a plurality of text segments; performing part-of-speech tagging and dependency syntax analysis on a plurality of the text segments, and identifying scene description sentences, plot description sentences, and character description sentences in each of the text segments; performing clustering on the scene description sentences based on preset semantic feature indicators, to obtain a scene element set; performing classification on the plot description sentences based on a preset plot type model, to obtain a plot element set; performing feature extraction on the character description sentences based on a preset character feature dimension, to obtain a character element set.

3. The method of claim 1, wherein, The weight assignment of each element based on the semantic similarity between each keyword in the static keyword set and each element in the scene element set, the plot element set, and the character element set, to obtain an element association graph containing weights, specifically includes: constructing a keyword-element bipartite graph, wherein the nodes of the keyword-element bipartite graph include keyword nodes in the static keyword set and element nodes in each element set; calculating the word vector cosine similarity between each keyword node and each element node as an edge weight, and determining the importance score of each element node in the keyword-element bipartite graph; using the importance score as the weight value of each element, and constructing an element association graph.

4. The method of claim 1, wherein, The construction of the multi-dimensional emotional feature space of the core plot unit specifically includes: obtaining emotional annotation data in a preset animation script corpus; performing dimensionality reduction processing on the emotional annotation data, to extract tension, cheerfulness, and sadness; calculating the independence coefficients between the tension, the cheerfulness, and the sadness; constructing a quantitative scoring standard based on the independence coefficients between the tension, the cheerfulness, and the sadness; applying the quantitative scoring standard to the tension, the cheerfulness, and the sadness, to construct a multi-dimensional emotional feature space.

5. The method of claim 1, wherein, The determination of the rhythm change of the emotional trajectory curve according to the golden section ratio and the rhythm feature vector specifically includes: dividing the total duration of the emotional trajectory curve into a plurality of time segments according to the golden section ratio; calculating the energy distribution of the rhythm feature vector in each time segment, and constructing a rhythm adjustment function based on the energy distribution, the rhythm adjustment function being used to amplify or suppress the emotional fluctuation intensity in each time segment; applying the rhythm adjustment function to the emotional trajectory curve, to obtain an optimized emotional rhythm change.

6. An animation script automatic generation system characterized by comprising: The system includes: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code, the computer program code including computer instructions, the one or more processors invoking the computer instructions to cause the system to perform the method of any one of claims 1-5.

7. A computer-readable storage medium comprising instructions, wherein: When the instructions run on the system, the system is caused to perform the method of any one of claims 1-5.

8. A computer program product, characterised in that, When the computer program product runs on the system, the system is caused to perform the method of any one of claims 1-5.

Images