A script content-driven multi-dimensional visual asset generation and management method

Abstract

The present application relates to the technical field of industrial big data, and more particularly to a script content-driven multi-dimensional visual asset generation and management method, which performs full-dimension semantic deconstruction and weight mapping processing on a script to obtain a semantic unit set and a comprehensive semantic weight set, performs narrative rhythm quantization and generation parameter adaptation processing to obtain a visual generation correction parameter set, and performs hierarchical asset matching and standardized management processing to obtain a target multi-dimensional visual asset; the present application realizes full-link driving of script content on visual asset generation and management, solves the core problem of disconnection between visual assets and script narrative logic in the prior art, and improves asset generation adaptability and management efficiency.

Description

Technical Field

[0001] This invention relates to the field of industrial big data technology, and in particular to a script-content-driven method for generating and managing multi-dimensional visual assets. Background Technology

[0002] In industrial-grade scenarios such as film and animation, and digital content production, script content is the core basis for visual asset generation. Existing technologies mostly extract keywords from the explicit text of the script and then generate visual assets through fixed parameter models. In practical applications, existing solutions can only achieve a superficial match between text content and visual images, failing to capture the inherent narrative logic and implicit expressive constraints of the script. The generated visual assets are severely disconnected from the script's core narrative needs, such as narrative rhythm, emotional delivery, and foreshadowing, failing to meet the standardization and high adaptability requirements of industrial-grade digital content production. Furthermore, existing solutions lack a unified standardized management system for generated visual assets, hindering cross-project asset reuse and resulting in high computational and labor costs for digital content production. Therefore, a technological solution capable of addressing these issues is urgently needed. Summary of the Invention

[0003] The purpose of this invention is to address the shortcomings of existing technologies by proposing a script-content-driven method for generating and managing multi-dimensional visual assets, comprising: S1: Obtain the script content to be processed, and perform full-dimensional semantic deconstruction and weight mapping processing on the script content to obtain a set of semantic units and a comprehensive semantic weight set bound to the set of semantic units. S2: Based on the semantic unit set and the comprehensive semantic weight set, perform narrative rhythm quantization and generation parameter adaptation processing to obtain a visual generation correction parameter set. S3: Based on the comprehensive semantic weight set and the visual generation correction parameter set, hierarchical asset matching and standardized management processing are performed to obtain the target multidimensional visual asset.

[0004] Preferably, the script full-dimensional semantic deconstruction and weight mapping processing includes extracting explicit semantic dimension information and implicit semantic dimension information contained in the script content to be processed, dividing the script content to be processed into a set of semantic units based on the explicit and implicit semantic dimension information, and performing normalization calculation on the explicit and implicit semantic dimension information to obtain a comprehensive semantic weight set bound to the set of semantic units.

[0005] Preferably, the comprehensive semantic weight set is calculated using the script's full-dimensional semantic weight calculation formula, which is: ; in, The comprehensive semantic weight bound to the i-th semantic unit in the set of semantic units. The coefficients are used to assign explicit semantic weights. This represents the total number of explicit semantic dimensions. The weight coefficients bound to the m-th explicit semantic dimension. Let be the normalized feature value bound to the i-th semantic unit in the m-th explicit semantic dimension in the semantic unit set. This represents the total number of implicit semantic dimensions. The weight coefficients bound to the nth implicit semantic dimension. It is the normalized feature value bound to the i-th semantic unit in the n-th implicit semantic dimension in the semantic unit set.

[0006] Preferably, the narrative rhythm quantification and generation parameter adaptation process includes extracting narrative rhythm dimension information contained in the semantic unit set, normalizing the narrative rhythm dimension information to obtain the narrative rhythm intensity set bound to the semantic unit set, and combining the comprehensive semantic weight set and the narrative rhythm intensity set to calculate the visual generation correction parameter set.

[0007] Preferably, the hierarchical asset matching and standardized management process includes constructing a hierarchical visual asset feature library, performing feature matching from the hierarchical visual asset feature library based on a comprehensive semantic weight set and a visual generation correction parameter set to obtain a matching asset set, calculating the asset reusability set contained in the matching asset set, filtering based on the asset reusability set to obtain a reusable asset set, performing incremental generation processing on the content not covered by the reusable asset set to obtain an incrementally generated asset set, binding the reusable asset set and the incrementally generated asset set to obtain a target multidimensional visual asset, and standardizing and encoding the target multidimensional visual asset and storing it in the hierarchical visual asset feature library.

[0008] Preferably, the explicit semantic dimension information includes character dimension information, scene dimension information, prop dimension information, action dimension information, and dialogue dimension information, and the implicit semantic dimension information includes narrative rhythm dimension information, emotion transmission dimension information, foreshadowing constraint dimension information, and subtext dimension information. The step of dividing the script content to be processed into a semantic unit set based on the explicit and implicit semantic dimension information includes processing the script content to be processed into sentences and blocks according to the script narrative logic, and binding the processed content with the explicit and implicit semantic dimension information to obtain the semantic unit set.

[0009] Preferably, the narrative rhythm dimension information includes conflict density information, scene switching frequency information, dialogue density information, and emotional fluctuation amplitude information. The normalization calculation of the narrative rhythm dimension information to obtain the narrative rhythm intensity set bound to the semantic unit set includes normalizing a single narrative rhythm dimension information and weighting the normalized multi-dimensional information to obtain the narrative rhythm intensity set bound to the semantic unit set.

[0010] Preferably, the incremental generation process includes setting visual asset generation parameters based on a comprehensive semantic weight set and a visual generation correction parameter set, performing generation operations based on the visual asset generation parameters to obtain initial incremental generated assets, performing consistency verification on the initial incremental generated assets, and including the initial incremental generated assets that pass the consistency verification as the content of the incremental generated asset set. The hierarchical visual asset feature library includes a core creative layer asset library, a narrative framework layer asset library, and an element detail layer asset library. The step of obtaining a matching asset set by feature matching from the hierarchical visual asset feature library includes performing feature matching from the core creative layer asset library, the narrative framework layer asset library, and the element detail layer asset library respectively.

[0011] Preferably, the standardization encoding and storage of the target multidimensional visual asset in the hierarchical visual asset feature library includes extracting semantic feature information, generation parameter information and narrative feature information bound to the target multidimensional visual asset; generating a standardized management code based on the semantic feature information, generation parameter information and narrative feature information; binding the standardized management code with the target multidimensional visual asset; storing the bound target multidimensional visual asset and standardized management code in the hierarchical visual asset feature library; and iteratively updating the standardized management code based on asset reuse data.

[0012] Preferably, obtaining the script content to be processed includes performing text cleaning processing on the script content to be processed, removing invalid format characters and redundant content contained in the script content to be processed, and performing format unification processing on the script content to be processed after text cleaning.

[0013] Technical effects: This invention solves the core problem of the disconnect between visual assets and script narrative logic in existing technologies through a collaborative solution of full-dimensional semantic deconstruction and weight mapping of scripts, quantification of narrative rhythm and adaptation of generation parameters, and hierarchical asset matching and standardized management. It realizes the full-link drive of script content on the generation and management of visual assets, and at the same time builds a standardized asset reuse and management system, reducing the cost of industrial-grade digital content production and improving the narrative adaptability and production efficiency of visual assets. Attached Figure Description

[0014] Figure 1A schematic diagram of the overall process for generating and managing multi-dimensional visual assets driven by script content; Figure 2 A schematic diagram of the script's full-dimensional semantic deconstruction and weight mapping process; Figure 3 A schematic diagram of the process for quantifying narrative rhythm and adapting visual generation parameters; Figure 4 A schematic diagram of the overall process for hierarchical visual asset matching and standardized management; Figure 5 A schematic diagram of the incremental generation and hierarchical matching process for the visual asset feature library; Figure 6 A schematic diagram of the standardized coding and storage iteration process for target multidimensional visual assets. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0016] Technical problems with existing technologies: Existing solutions only extract keywords based on the explicit text content of the script, which cannot capture the inherent narrative logic and implicit expressive constraints of the script. The generated visual assets are out of touch with the script's narrative needs and cannot meet the requirements of industrial-grade production.

[0017] Based on this, please refer to Figures 1-6 The core scheme of the script content-driven multidimensional visual asset generation and management method provided in this embodiment includes obtaining script content to be processed, performing full-dimensional semantic deconstruction and weight mapping processing on the script content to be processed to obtain a set of semantic units and a comprehensive semantic weight set bound to the set of semantic units, performing narrative rhythm quantification and generation parameter adaptation processing based on the set of semantic units and the comprehensive semantic weight set to obtain a set of visual generation correction parameters, and performing hierarchical asset matching and standardized management processing based on the comprehensive semantic weight set and the visual generation correction parameter set to obtain the target multidimensional visual asset.

[0018] It is worth mentioning that the method in this embodiment is implemented based on an industrial big data processing platform. This platform is equipped with a multi-core processor and a distributed storage module, supporting batch processing of script data and standardized storage of visual assets. The overall processing flow follows the temporal logic of industrial-grade digital content production, forming a closed-loop processing system from script content acquisition to the output of target multi-dimensional visual assets. The process of acquiring the script content to be processed relies on a standardized script data interface, which is compatible with various mainstream script file formats such as TXT, DOCX, and XML. After the script content to be processed is input into the industrial big data processing platform, it is first temporarily stored in the data caching module to provide data support for subsequent full-dimensional semantic deconstruction and weight mapping processing. The full-dimensional semantic deconstruction and weight mapping processing of the script is the foundation of the entire method. Its processing results are directly used as input data for narrative rhythm quantification and generation parameter adaptation processing. The two are transmitted in real time through the internal data bus of the industrial big data processing platform to ensure the temporality and accuracy of data processing. The set of visual generation correction parameters output from the narrative rhythm quantification and generation parameter adaptation processing, along with the comprehensive semantic weight set, is fed into the hierarchical asset matching and standardized management module. This module serves as the core quantitative basis for asset matching and incremental generation. The hierarchical asset matching and standardized management module establishes a bidirectional data connection with a pre-set hierarchical visual asset feature library. It retrieves reusable assets from the feature library and stores newly generated target multi-dimensional visual assets in the feature library after standardized encoding, thus achieving full lifecycle management of visual assets. This solution, through end-to-end script content-driven processing, solves the problem that existing technologies can only achieve surface-level text matching and cannot adapt to narrative logic. It achieves deep adaptation between visual assets and the core requirements of the script, constructing a complete generation and management closed loop.

[0019] Technical problems with existing technologies: Existing solutions only cover explicit elements in the semantic extraction of script content, ignoring implicit narrative constraints, resulting in incomplete semantic decomposition and failing to provide a comprehensive basis for visual generation.

[0020] Based on this, the script full-dimensional semantic deconstruction and weight mapping processing includes extracting explicit and implicit semantic dimension information contained in the script content to be processed, dividing the script content to be processed into a set of semantic units based on the explicit and implicit semantic dimension information, and normalizing the explicit and implicit semantic dimension information to obtain a comprehensive semantic weight set bound to the set of semantic units.

[0021] It is worth mentioning that the extraction of explicit and implicit semantic dimension information is achieved based on a pre-trained natural language processing model. This model has been fine-tuned for the linguistic features and narrative structure of the script text, enabling it to accurately identify explicit elements and implicit expressions within the script. When extracting explicit semantic dimension information, the model uses entity recognition, keyword extraction, and syntactic analysis to identify textual content related to characters, scenes, props, actions, and lines from the script content, and annotates these types of content with features to form structured explicit semantic dimension information. When extracting implicit semantic dimension information, the model uses deep semantic analysis, contextual reasoning, and narrative logic mining to extract implicit information related to narrative rhythm, emotional transmission, foreshadowing constraints, and subtext from the script content. Simultaneously, it verifies the implicit information within the context of the script to ensure the accuracy of the implicit semantic dimension information. When dividing the script content based on explicit and implicit semantic dimensions, the script is divided into multiple independent but interconnected text fragments according to its narrative logic, using scenes, plot points, and dialogue units as basic criteria. Each text fragment is bound to corresponding explicit and implicit semantic dimensions, forming a set of semantic units. Each semantic unit is assigned a unique identifier for easy subsequent quantification and data management. Before normalizing the explicit and implicit semantic dimensions, feature quantification is performed, converting textual semantic information into numerical feature values. An extreme value normalization algorithm is then used to map all feature values ​​to the same numerical range, eliminating dimensional differences between different feature values ​​and laying the foundation for subsequent comprehensive semantic weight calculation. The resulting comprehensive semantic weight set, bound to the set of semantic units, is stored as a numerical matrix on the industrial big data processing platform for easy retrieval and calculation in subsequent stages. This solution, through explicit and implicit dual-dimensional semantic extraction, solves the problem of single-dimensional semantic decomposition in existing technologies, providing comprehensive semantic basis for visual generation and ensuring complete semantic coverage.

[0022] Technical problems with existing technologies: Existing solutions lack a clear way to quantify the influence weight of different semantic units in the script, resulting in a mismatch between the focus of visual generation and the core content of the script.

[0023] Based on this, the comprehensive semantic weight set is calculated using the script's full-dimensional semantic weight calculation formula, which is as follows: ; The core design basis of this formula is the dual requirements for the generation of script visual assets. Visual assets must not only meet the presentation requirements of the script's visual elements, but also conform to the script's inherent narrative logic requirements. Existing technologies only quantify semantic information in a single dimension, which cannot cover both types of requirements at the same time. This formula, however, achieves a comprehensive quantification of the influence weight of script semantic units through the weighted fusion calculation of explicit and implicit semantics. This is also the core creative point that distinguishes this formula from existing quantification methods. By incorporating the basic element constraints of explicit semantics and the narrative logic constraints of implicit semantics into the same quantification system, the calculated comprehensive semantic weight can truly reflect the actual degree of influence of each semantic unit on the generation of visual assets.

[0024] The following provides a detailed explanation of each part of the formula. This is the comprehensive semantic weight bound to the i-th semantic unit in the semantic unit set. This parameter is a dimensionless normalized parameter with a fixed value range of 0 to 1. A higher value indicates a greater influence of the semantic unit on visual asset generation. It is the core foundational parameter of the entire technical solution, providing the core basis for subsequent generation parameter adaptation and asset matching. Its value directly determines the priority of the corresponding semantic unit in the visual asset generation process. The value will be significantly higher than that of non-core semantic units, ensuring that the focus of visual generation remains consistent with the core content of the script. This is the explicit semantic weighting coefficient. This parameter is a dimensionless normalized parameter, ranging from 0 to 1. It is used to allocate the proportion of explicit and implicit semantics in the comprehensive weighting calculation. It can be dynamically adjusted according to the script type. In script types that emphasize visual presentation, A value closer to 1 can be chosen to strengthen the impact of explicit semantics; in script genres that emphasize narrative expression, A value closer to 0 can be used to strengthen the influence of implicit semantics. By dynamically adjusting this coefficient, the formula can be adapted to the quantitative needs of different types of scripts, thereby improving the universality of the formula. The total number of explicit semantic dimensions represents the number of dimensions into which the explicit semantics of the script are broken down. In this embodiment, the total number of explicit semantic dimensions is fixed at 5, corresponding to five dimensions: character, scene, prop, action, and dialogue. This value is set based on the core explicit elements of the script text and is the optimal value after extensive script text analysis and industrial-grade production verification, which can fully cover the core explicit requirements of the script's visual presentation. This is the weight coefficient bound to the m-th explicit semantic dimension. This parameter is a dimensionless normalized parameter, ranging from 0 to 1. The sum of the weight coefficients for all explicit semantic dimensions is 1. It is used to adjust the degree of influence of different explicit semantic dimensions on the overall weight and can be adjusted according to the creative focus of the script. For example, in a character-driven script, the character dimension corresponds to... The value will be appropriately increased; in scene-driven scripts, the scene dimension corresponds to... The value will be appropriately increased. By differentiating the settings of this coefficient, the calculation of the comprehensive semantic weight will be more in line with the creative characteristics of the script. This is the normalized feature value bound to the i-th semantic unit in the m-th explicit semantic dimension within the semantic unit set. This parameter is a dimensionless normalization parameter, ranging from 0 to 1, representing the feature strength of the semantic unit in the corresponding explicit dimension. It is obtained by extracting semantic features from the script text and then normalizing them to ensure that the feature values ​​of all dimensions are of the same computational magnitude, satisfying the principle of homogeneity of dimensions. Its value is determined by the richness and importance of the explicit semantic information in the corresponding semantic unit. For example, if a semantic unit contains key actions of a core character, its corresponding action dimension... The value will approach 1. The total number of implicit semantic dimensions represents the number of dimensions into which the implicit semantics of the script are broken down. In this embodiment, the total number of implicit semantic dimensions is fixed at 4, corresponding to the four dimensions of narrative rhythm, emotional transmission, foreshadowing constraints, and subtext. This value is set based on the core implicit narrative elements of the script text and can fully cover the core implicit needs of the script's narrative logic expression. This is the weight coefficient bound to the nth implicit semantic dimension. This parameter is a dimensionless normalized parameter, ranging from 0 to 1. The sum of the weight coefficients of all implicit semantic dimensions is 1. It is used to adjust the degree of influence of different implicit semantic dimensions on the overall weight. It can be adjusted according to the narrative characteristics of the script. For example, in a suspense script, the foreshadowing constraint dimension corresponds to... The value will be appropriately increased; in emotional scripts, the dimension corresponding to emotion transmission will be increased. The value will be increased appropriately. This is the normalized feature value bound to the i-th semantic unit in the n-th implicit semantic dimension within the semantic unit set. This parameter is a dimensionless normalization parameter, ranging from 0 to 1, representing the feature strength of the semantic unit in the corresponding implicit dimension. It is obtained through deep semantic analysis of the script text followed by normalization processing, maintaining the same dimension as the explicit semantic feature value to ensure dimensional consistency in formula addition and subtraction operations, fully conforming to the principle of dimensional homogeneity. Its value is determined by the expression intensity and correlation degree of the implicit semantic information in the corresponding semantic unit. For example, if a semantic unit contains key narrative foreshadowing, its corresponding foreshadowing constraint dimension... The value will approach 1.

[0025] The logical derivation of this formula is as follows: First, the core objective of generating script visual assets is to simultaneously satisfy both the script's visual element requirements and narrative logic requirements. Visual element requirements correspond to the explicit semantic dimension, while narrative logic requirements correspond to the implicit semantic dimension. Therefore, the comprehensive weight needs to be composed of both parts. This is the core logical premise of the formula construction, breaking through the limitations of existing technologies that only construct quantitative formulas from a single dimension. Second, for the explicit semantic part, elements of different dimensions have different degrees of influence on visual generation, making direct summation impossible. Therefore, it is necessary to use dimension weight coefficients to weight and sum the normalized feature values ​​of different dimensions to obtain the comprehensive influence value of the explicit semantics. This process is achieved through… This ensures that the impact of different explicit semantic dimensions can be reasonably quantified according to their importance. The implicit semantic component also exhibits differences in the degree of influence across different dimensions. Therefore, it is necessary to use dimension weight coefficients to perform a weighted summation of the normalized feature values ​​of different dimensions to obtain the comprehensive impact value of the implicit semantics. This process involves... Implementation. Coefficients are assigned through explicit semantic weights. With complementarity coefficient The combined influence values ​​of the explicit and implicit components are weighted and summed to obtain the final comprehensive semantic weight. This ensures that the weight calculation simultaneously covers both visual elements and narrative logic, providing accurate quantitative basis for subsequent visual generation. The formula is implemented in the industrial big data processing platform through parallel computing. By substituting the numerical matrix of the semantic unit set into the formula, the platform can simultaneously calculate the comprehensive semantic weight of all semantic units, significantly improving computational efficiency and meeting the needs of industrial-grade batch processing.

[0026] The existing technology has the following technical problems: The visual generation parameters of existing solutions are fixed and cannot adapt to the changing narrative rhythm of different parts of the script. This results in a mismatch between the generated visual assets and the script's narrative rhythm, failing to convey the script's core emotions and narrative tension. Therefore, the narrative rhythm quantification and generation parameter adaptation process includes extracting narrative rhythm dimension information contained in the semantic unit set, normalizing the narrative rhythm dimension information to obtain the narrative rhythm intensity set bound to the semantic unit set, and combining the comprehensive semantic weight set and the narrative rhythm intensity set to calculate the visual generation correction parameter set.

[0027] It is worth mentioning that the extraction of narrative rhythm dimension information is based on a set of semantic units and relies on a script narrative feature analysis algorithm. This algorithm can extract core information related to narrative rhythm from the text content of each semantic unit, forming structured narrative rhythm dimension information. When normalizing the narrative rhythm dimension information, various types of narrative rhythm information are first converted into numerical feature values. Then, an extreme value normalization algorithm is used to map the feature values ​​to a numerical range of 0 to 1, eliminating the dimensional differences between different types of narrative rhythm information, resulting in a narrative rhythm intensity set. Each semantic unit corresponds to a unique narrative rhythm intensity value, which can accurately reflect the magnitude of the narrative tension of the corresponding semantic unit. When calculating the visual generation correction parameter set by combining the comprehensive semantic weight set and the narrative rhythm intensity set, the numerical matrices of the two sets are calculated collaboratively. This allows the visual generation correction parameters to simultaneously carry the importance of semantic units and narrative rhythm features. The calculated visual generation correction parameter set is stored in the form of a numerical matrix. Each value directly corresponds to the specific parameter adjustment basis for visual asset generation. For example, high-value visual generation correction parameters correspond to visual generation requirements of high detail precision and strong visual tension, while low-value parameters correspond to visual generation requirements of simplification and weak visual tension. This solution addresses the problem of fixed generation parameters and inability to adapt to changing narrative rhythm in existing technologies by quantifying narrative rhythm and dynamically adapting generation parameters, ensuring a high degree of match between the generated visual assets and the script's narrative rhythm. In this implementation, the set of visual generation correction parameters is calculated using a narrative rhythm-driven generation parameter correction formula. This formula is designed to ensure that visual generation parameters not only match the comprehensive weight and rhythmic intensity of the current semantic unit but also consider the narrative coherence of related semantic units, avoiding disjointed visual assets. Existing technologies only consider the local features of a single semantic unit, neglecting the overall narrative of the script. This formula, however, introduces window analysis of related semantic units, combining local features with the overall narrative logic. This is the core innovation of the formula, allowing the generated visual generation correction parameters to simultaneously meet both local narrative needs and overall coherence requirements. The specific form of the formula is: ; The following is a detailed explanation of the formula, in which... This is the visual generation correction parameter bound to the i-th semantic unit in the semantic unit set. This parameter is a dimensionless normalized parameter with a value range of 0 to 1. It is the core basis for adjusting the visual asset generation parameters. The higher the value, the higher the detail precision and stronger visual tension required for the visual asset corresponding to the semantic unit in order to match the narrative needs of the script. Its value directly determines the parameter settings of the corresponding semantic unit in the visual generation process, which can make the presentation effect of the visual asset highly consistent with the narrative rhythm and semantic importance of the script. The comprehensive semantic weight bound to the i-th semantic unit in the semantic unit set is completely consistent with the output parameters of the aforementioned script full-dimensional semantic weight calculation formula. The dimension is dimensionless to ensure the consistency of the dimensions in the formula multiplication operation. It is used as the basic coefficient of the formula so that the magnitude of the visual generation correction parameter is directly linked to the importance of the semantic unit, ensuring that the core semantic units of the script can be highlighted. This parameter, a dimensionless normalized parameter ranging from 0 to 1, is used to assign a weighting coefficient to the local rhythm of the current semantic unit and the coherent rhythm of the preceding and following semantic units in the parameter calculation. This balances the needs of local narrative with the overall narrative coherence, particularly important in script segments that emphasize plot conflict. A value closer to 1 can be used to strengthen the impact of local narrative rhythm; in script segments that emphasize narrative transitions, A value closer to 0 can be used to enhance the overall narrative coherence. By dynamically adjusting this coefficient, the formula can be adapted to the parameter calculation needs of different narrative segments in the script. The narrative rhythm intensity is the value bound to the i-th semantic unit in the set of semantic units. This parameter is a dimensionless normalized parameter, ranging from 0 to 1, representing the narrative tension intensity of this semantic unit. It is obtained through weighted calculation of the narrative rhythm dimension information, and its value is determined by factors such as conflict density, scene switching frequency, dialogue density, and emotional fluctuation amplitude within the semantic unit. The stronger the narrative tension of a semantic unit, the stronger its narrative rhythm intensity. The closer the value is to 1. This is the window length for related semantic units, representing the number of related semantic units that need to be considered before and after the current semantic unit. It is used to ensure narrative coherence. This value can be dynamically adjusted according to the pace of the script's narrative. In scripts with a faster narrative pace... Smaller values ​​can be used, focusing on nearby semantic units; in scripts with a slower narrative pace, A larger value can be chosen to take into account more related semantic units. In this embodiment, The value range is from 1 to 5, which is the optimal value range after industrial-grade production verification. The comprehensive semantic weights bound to the associated semantic units. The narrative rhythm intensity, bound to the associated semantic units, is a dimensionless normalized parameter, maintaining the same dimensions as the aforementioned parameters. It corresponds to the comprehensive semantic weight and narrative rhythm intensity of the k-th associated semantic unit within the window, respectively, and is the core parameter for calculating overall narrative coherence. In the formula... This is the weighted average rhythmic influence value within the preceding and following windows. This value is a dimensionless parameter and is related to the local rhythmic intensity. Maintaining the same dimensions ensures consistency in the dimensions of addition and subtraction operations, fully complying with the principle of homogeneity of dimensions. This value can accurately reflect the overall rhythmic characteristics of the narrative segment in which the current semantic unit is located, allowing the visual generation correction parameters to take into account the overall narrative logic and avoid the problem of disjointed visual asset presentation.

[0028] The logical derivation of this formula is as follows: The core determinant of visual generation parameters is the comprehensive semantic weight of the current semantic unit. The higher the weight, the greater the impact on the generation parameters. Therefore, the comprehensive semantic weight... As the basic coefficients of the formula, the overall magnitude of the visual generation correction parameters is matched with the importance of the semantic units. The generation parameters need to match the local narrative rhythm of the current semantic unit and the narrative coherence of the overall script at the same time.

[0029] Existing technologies only consider local narrative rhythm, resulting in a lack of overall coherence in the presentation of visual assets. This formula incorporates local rhythm and overall coherent rhythm into the same calculation system, with local rhythm determined by the narrative rhythm intensity of the current semantic unit. Decision, passed This achieves the quantification of local rhythm; the overall coherent rhythm is determined by the weighted average rhythm influence value within the preceding and following related windows, through... To quantify the overall coherent rhythm, local rhythm weighting coefficients are assigned. A weighted summation is performed to obtain a comprehensive rhythmic influence factor, ensuring a reasonable balance between local narrative needs and overall narrative coherence. The comprehensive semantic weights are then applied. Multiplying this by the overall rhythm influence factor yields the final visual generation correction parameter. This ensures that the generated parameters simultaneously match semantic weights, local rhythm, and overall coherence, providing a precise dynamic adjustment basis for visual asset generation. This formula is implemented in an industrial big data processing platform using a time-series calculation method, sequentially substituting semantic units into the formula for calculation, while also supporting window length... Local rhythm weight allocation coefficient Its dynamic adjustment can quickly adapt to the narrative characteristics of different scripts and meet the needs of industrial-grade real-time processing.

[0030] The existing technology has the following technical problems: Current solutions employ a single-script, full-scale generation model, lacking a standardized asset matching and reuse system. This results in high computing and labor costs, and the generated assets lack unified management standards, hindering cross-project asset reuse. Therefore, the hierarchical asset matching and standardized management process includes constructing a hierarchical visual asset feature library; performing feature matching from the hierarchical visual asset feature library based on a comprehensive semantic weight set and a visual generation correction parameter set to obtain a matching asset set; calculating the asset reusability set contained in the matching asset set; filtering based on the asset reusability set to obtain a reusable asset set; performing incremental generation processing on content not covered by the reusable asset set to obtain an incrementally generated asset set; binding the reusable asset set and the incrementally generated asset set to obtain the target multi-dimensional visual asset; and standardizing and encoding the target multi-dimensional visual asset and storing it in the hierarchical visual asset feature library.

[0031] It is worth mentioning that the hierarchical visual asset feature library is built based on industrial big data distributed storage technology. It is stored in layers according to the core creative layer, narrative framework layer, and element detail layer, with each layer establishing an independent feature index system to facilitate rapid asset retrieval and matching. All visual assets stored in the feature library are accompanied by corresponding semantic features, generation parameters, narrative features, and other structured information, providing data support for asset matching. When performing feature matching from the hierarchical visual asset feature library based on the comprehensive semantic weight set and visual generation correction parameter set, a cosine similarity algorithm is used. The comprehensive semantic weight set and visual generation correction parameter set are converted into feature vectors, and similarity is calculated between these vectors and the feature vectors of visual assets in the feature library. Visual assets with similarity reaching a preset threshold are included in the matching asset set. When calculating the asset reusability set included in the matching asset set, the feature matching results are used as a basis, combined with hierarchical reuse rules for quantitative calculation, to obtain the reusability value of each matched asset. When selecting reusable assets based on the asset reusability set, matching assets with reusability values ​​reaching a preset threshold are considered reusable. This reusability threshold can be dynamically adjusted according to the script's creative requirements. For incremental generation of content not covered by the reusable asset set, visual asset generation parameters are set based on a comprehensive semantic weight set and a visual generation correction parameter set. Incremental assets are generated using a visual generation model and undergo consistency verification before being included in the incremental generation asset set. When binding the reusable asset set and the incremental generation asset set, the two types of assets are integrated according to the script's narrative logic, ensuring consistency in style, precision, and narrative expression after integration, ultimately resulting in the target multidimensional visual asset. During the standardized coding of the target multidimensional visual asset, core information such as semantic features, generation parameters, and narrative features are extracted. A unique standardized management code is generated according to industrial-grade visual asset coding rules. This code is then bound to the target multidimensional visual asset and stored in a hierarchical visual asset feature library for standardized management and cross-project reuse.

[0032] This solution addresses the issues of low asset reuse rates and chaotic management in existing technologies through hierarchical asset matching and standardized management, significantly reducing the cost of industrial-grade digital content production while achieving full lifecycle management of assets. In this implementation phase, the asset reusability set is calculated using a hierarchical asset reusability calculation formula. This formula is based on the premise that asset reusability is jointly determined by semantic feature matching and generation parameter adaptation. Different reuse weights are assigned according to hierarchical features to ensure that only assets that meet the core needs of the script are reused, avoiding a disconnect between reused assets and script requirements. Existing technologies only match assets based on a single feature, neglecting hierarchical feature differences and generation parameter adaptation, resulting in insufficient accuracy and rationality in asset reuse. This formula, through hierarchical weight allocation and two-dimensional matching quantification, achieves precise calculation of asset reusability. This is the core innovation of the formula, enabling asset reuse to both reduce production costs and ensure the core creative expression of the script. The specific form of the formula is: ; The following is a detailed explanation of the formula, in which... This parameter represents the reusability of the asset associated with the i-th semantic unit in the semantic unit set. It is a dimensionless normalized parameter with a value ranging from 0 to 1. The higher the value, the stronger the reusability of the asset. When the value reaches a preset threshold, the asset can be reused. Its value directly determines whether the corresponding asset can be included in the reusable asset set. It is the core quantitative basis for asset reuse screening and can accurately determine the degree of matching between the asset and the semantic requirements and generation parameter requirements of the current script. A coefficient is assigned to the semantic feature matching weights. This parameter is a dimensionless normalized parameter, ranging from 0 to 1, and is used to allocate the proportion of semantic feature matching degree and generated parameter fit degree in the reusability calculation. In scripts that emphasize semantic expression, A value closer to 1 can be chosen to strengthen the impact of semantic feature matching; in scripts that emphasize visual presentation parameters, A value closer to 0 can be used to enhance the impact of the generated parameter adaptability. By dynamically adjusting this coefficient, the formula can be adapted to the asset reuse needs of different scripts. The comprehensive semantic weight bound to the i-th semantic unit in the set of semantic units. The maximum value of the comprehensive semantic weight set is the comprehensive semantic weight. Both are dimensionless parameters. The normalized weight ratio obtained by dividing them is also a dimensionless parameter to ensure dimensional consistency. This normalized weight ratio can incorporate the importance of semantic units into the calculation of asset reusability, making the asset matching requirements of core semantic units more stringent and ensuring the presentation effect of the core content of the script. The total number of hierarchical semantic feature dimensions is fixed at 3 in this embodiment, corresponding to the three dimensions of core creative layer, narrative framework layer, and element detail layer. This value is set based on the hierarchical structure of the hierarchical visual asset feature library, which can fully cover the core feature level of visual assets, while taking into account the rationality of asset reuse and the protection of the core creative ideas of the script. The reuse weight coefficient is bound to the t-th level feature. This parameter is a dimensionless normalized parameter, with a value ranging from 0 to 1. The sum of the weight coefficients of all levels is 1. The weight coefficient of the element detail layer is the highest, and the weight coefficient of the core creative layer is the lowest. This setting is an important creative design of this formula, ensuring that only element detail assets are reused at a high proportion, while core creative layer and narrative framework layer assets are reused at a low proportion. This can give full play to the cost advantage of asset reuse, and avoid the reuse of assets from affecting the core creative expression of the script. This coefficient can be dynamically adjusted according to the originality requirements of the script. This parameter represents the feature matching degree between the i-th semantic unit in the semantic unit set at the t-th level and the matched asset. It is a dimensionless normalized parameter, ranging from 0 to 1, representing the degree of feature matching between the asset and the current semantic unit at the corresponding level. It is calculated using the cosine similarity algorithm; a higher matching degree indicates a better match. The closer the value is to 1, the more directly it reflects the feature fit between the matched asset and the current semantic unit at the corresponding level. The visual generation correction parameters are bound to the i-th semantic unit in the set of semantic units. To match the visual generation correction parameters bound to the asset, both are dimensionless parameters. The absolute value of the difference is a dimensionless parameter. The generation parameter fit obtained by subtracting this difference from 1 is a dimensionless parameter, which maintains the same dimension as the semantic feature matching part, ensuring the consistency of the dimension of addition and subtraction operations. It fully conforms to the principle of homogeneity of dimensions. The generation parameter fit can accurately reflect the degree of fit between the generation parameters of the matching asset and the generation parameter requirements of the current semantic unit. The smaller the difference, the higher the fit and the stronger the reusability of the asset.

[0033] The logical derivation of this formula is as follows: First, the core criterion for asset reusability is the degree of matching between the asset and the semantic requirements of the current script. This degree of matching is composed of two parts: semantic feature matching degree and generation parameter adaptation degree. Existing technologies only consider matching in a single dimension, leading to insufficient accuracy in asset reuse. Therefore, this formula incorporates the matching results of both dimensions into the same calculation system; this is the core logic of the formula construction. Second, the semantic feature matching degree needs to be weighted according to hierarchical features, prioritizing the reuse of element detail-level assets to avoid affecting the core creative ideas of the script. Therefore, it first... We calculate the weighted sum of feature matching degrees at different levels to obtain a hierarchical semantic feature comprehensive matching degree, and then combine it with the normalized weight ratio of the current semantic unit. The importance of semantic units is incorporated into the calculation to obtain the final semantic feature matching value, which is then used to... To ensure semantic feature matching, the calculation of semantic feature matching takes into account both hierarchical feature differences and the semantic focus of the script. Then, the adaptability of generated parameters is a crucial factor for asset reusability. Even with high semantic feature matching, effective reuse is impossible if the generated parameters don't match. Therefore, the adaptability of generated parameters is calculated by the difference between the current semantic unit and the generated correction parameters of the matching asset. The smaller the difference, the higher the adaptability. The quantification of the adaptation of generation parameters is achieved, and then the complementary coefficients are used. The weighted values ​​are then used to obtain the final generated parameter adaptation values. Finally, the semantic feature matching values ​​and the generated parameter adaptation values ​​are weighted and summed to obtain the final asset reusability. This provides precise quantitative criteria for asset reuse screening, ensuring that the selected reusable assets not only meet the semantic and generation parameter requirements of the script, but also balance cost control and core creative protection. The formula is implemented in conjunction with the feature library retrieval system within the industrial big data processing platform. When a matching asset is retrieved from the feature library, the system automatically extracts relevant parameters and substitutes them into the formula to calculate the asset's reusability in real time, meeting the needs of rapid matching and screening at the industrial level.

[0034] Technical problems with existing technologies: Existing solutions lack clear boundaries and content definitions for the semantic dimension of scripts, resulting in insufficient completeness and consistency of semantic extraction, and failing to stably support subsequent weight calculation and generation adaptation.

[0035] Based on this, explicit semantic dimension information includes character dimension information, scene dimension information, prop dimension information, action dimension information, and dialogue dimension information, while implicit semantic dimension information includes narrative rhythm dimension information, emotion transmission dimension information, foreshadowing constraint dimension information, and subtext dimension information. Based on explicit and implicit semantic dimension information, the script content to be processed is divided into semantic unit sets. This includes processing the script content to be processed into sentences and blocks according to the script's narrative logic, and binding the processed content with explicit and implicit semantic dimension information to obtain the semantic unit set.

[0036] It is worth mentioning that the five categories of explicit semantic information are all defined based on the core visual elements of the script text. The character dimension information covers the core information such as the identity, personality, appearance, and dialogue characteristics of all characters in the script. The scene dimension information covers the core information such as the spatiotemporal background, environmental characteristics, and screen layout of all scenes in the script. The prop dimension information covers the core information such as the appearance, functional attributes, and usage scenarios of all props in the script. The action dimension information covers the core information such as the body movements, facial expressions, and behavioral actions of all characters in the script. The dialogue dimension information covers the core information such as the dialogue content, tone, and dialogue logic of all characters in the script. The clear definition of the five categories of information provides a unified standard for explicit semantic extraction, ensuring the completeness and consistency of the extraction results. The four categories of implicit semantic information are all defined based on the core narrative elements of the script text. Narrative rhythm information covers core information such as conflict density, scene switching frequency, and dialogue density in each paragraph; emotional transmission information covers core information such as the emotional expression, changes, and rendering of each character; foreshadowing constraints information covers core information such as the setting, foreshadowing, and echoing of various foreshadowing elements; and subtext information covers core information such as the implied meaning, implied meaning, and contextual expression of various lines in the script. The clear definition of these four categories provides a unified standard for implicit semantic extraction, overcoming the lack of clear definition in existing technologies. When processing the script content by dividing it into sentences and blocks according to the script's narrative logic, the basic units are scenes, acts, plot points, and dialogue units. The division process relies on the script's format and textual features; for example, scene identifiers and scene transition identifiers are used as the basis for scene division, and dialogue identifiers and character switching identifiers are used as the basis for dialogue unit division, ensuring that the divided text fragments conform to the script's narrative logic. When binding the segmented and block-processed content with explicit and implicit semantic dimensions, a feature annotation and identifier encoding binding method is adopted. Each segmented text fragment is assigned a unique identifier code, and the extracted explicit and implicit semantic dimensions are bound to this identifier code to form structured semantic units. All semantic units are integrated to form a semantic unit set, where each semantic unit possesses complete explicit and implicit semantic information, providing stable and unified basic data for subsequent weight calculation and adaptation. This scheme, through clear semantic dimension definition and semantic unit division, solves the problem of insufficient consistency in semantic extraction in existing technologies, ensuring the completeness and stability of semantic extraction and providing unified basic data for subsequent processing.

[0037] Technical problems with existing technologies: Existing solutions lack clear dimensional definitions and calculation methods for quantifying narrative rhythm, resulting in rhythm quantification results that cannot accurately reflect the narrative tension of the script and cannot provide an effective basis for parameter adaptation.

[0038] Based on this, the narrative rhythm dimension information includes conflict density information, scene switching frequency information, dialogue density information, and emotional fluctuation amplitude information. The narrative rhythm dimension information is normalized to obtain the narrative rhythm intensity set bound to the semantic unit set. This includes normalizing a single narrative rhythm dimension information and weighting the normalized multi-dimensional information to obtain the narrative rhythm intensity set bound to the semantic unit set.

[0039] It is worth mentioning that the four categories of narrative rhythm information are all defined based on the core influencing factors of the script's narrative tension. Conflict density information refers to the frequency and intensity of plot conflicts within a unit of semantics; scene switching frequency information refers to the frequency and speed of scene switching within a unit of semantics; dialogue density information refers to the number of words, sentences, and dialogue frequency within a unit of semantics; and emotional fluctuation amplitude information refers to the amplitude and intensity of emotional changes and expressions of characters within a unit of semantics. The clear definition of these four categories provides a unified standard for quantifying narrative rhythm, ensuring that the quantification results accurately reflect the script's narrative tension. When normalizing a single dimension of narrative rhythm information, the values ​​of each type of information are first quantified. For example, conflict density information is converted into a weighted value of the frequency and intensity of conflicts, and scene switching frequency information is converted into a ratio of the number of scene switching times to the time interval. Then, an extreme value normalization algorithm is used to map each quantified value to a numerical range of 0 to 1, eliminating the dimensional differences between different dimensions of narrative rhythm information and ensuring that all types of information can be calculated collaboratively. When performing weighted calculations on the normalized multi-dimensional information, a corresponding weight coefficient is assigned to each type of normalized narrative rhythm information. The value of the weight coefficient is set based on the degree of influence of each type of information on narrative tension. For example, the weight coefficient for conflict density information is the highest, followed by the weight coefficient for dialogue density information, and the sum of all weight coefficients is 1. The narrative rhythm intensity value of each semantic unit is obtained by weighted summation. The narrative rhythm intensity values ​​of all semantic units are integrated to form a narrative rhythm intensity set. Each value in this set can accurately reflect the narrative tension intensity of the corresponding semantic unit, providing an effective and accurate quantitative basis for parameter adaptation. This scheme solves the problem of inaccurate rhythm quantification in existing technologies by clearly defining the narrative rhythm dimensions and using quantitative calculation methods, ensuring that the rhythm quantification results can accurately reflect the narrative tension of the script and provide an accurate basis for parameter adaptation.

[0040] Technical problems with existing technologies: There is no consistency verification between incrementally generated assets and reused assets in existing solutions, which leads to inconsistent styles and accuracy of integrated visual assets. At the same time, the asset feature library is not hierarchically divided, resulting in insufficient efficiency and accuracy of asset matching.

[0041] Based on this, the incremental generation process includes setting visual asset generation parameters based on a comprehensive semantic weight set and a visual generation correction parameter set, performing generation operations based on the visual asset generation parameters to obtain initial incremental generated assets, performing consistency verification on the initial incremental generated assets, and including the initial incremental generated assets that pass the consistency verification as the content of the incremental generated asset set. The hierarchical visual asset feature library includes a core creative layer asset library, a narrative framework layer asset library, and an element detail layer asset library. The matching asset set obtained by feature matching from the hierarchical visual asset feature library includes performing feature matching from the core creative layer asset library, the narrative framework layer asset library, and the element detail layer asset library respectively.

[0042] It's worth noting that when setting visual asset generation parameters based on the comprehensive semantic weight set and the visual generation correction parameter set, the numerical parameters of both sets are mapped to various adjustable parameters of the visual generation model. For example, high-value comprehensive semantic weights and visual generation correction parameters are mapped to generation parameters with high detail accuracy, high resolution, and strong color contrast, while low-value parameters are mapped to simplified, low-resolution, and weak color contrast generation parameters. This ensures that the generation parameters are highly consistent with the semantic needs and narrative rhythm of the script. When performing generation operations based on the visual asset generation parameters, a pre-trained visual generation model is used. This model has been fine-tuned for the needs of industrial-grade digital content production, supporting the generation of visual assets in various styles and with varying levels of precision. The initial incremental generated assets are stored in standardized image and video formats. When performing consistency checks on the initial incrementally generated assets, the checks are performed from four dimensions: style, precision, color, and narrative expression. The initial incrementally generated assets are compared with the assets in the reusable asset set. If the similarity between the two in the four dimensions reaches a preset threshold, the consistency check is passed. Initial incrementally generated assets that fail the check are regenerated by readjusting the generation parameters until they pass the check, ensuring that the incrementally generated assets and reusable assets are consistent in presentation. The hierarchical visual asset feature library is divided into three layers based on the creative attributes and reusability value of visual assets. The core creative layer asset library stores assets related to the core creative design of the script, with the lowest reusability value. The narrative framework layer asset library stores assets related to the narrative framework design of the script, with medium reusability value. The element detail layer asset library stores assets related to the element detail design of the script, with the highest reusability value. Each layer of the asset library establishes an independent feature index and retrieval system, which greatly improves the efficiency of asset matching. When performing feature matching to obtain a set of matched assets from the three-layer asset library, a hierarchical matching and filtering approach is adopted. Matching is first performed from the core creative layer asset library, followed by matching from the narrative framework layer and the element detail layer asset libraries. Each layer of matching uses a corresponding feature matching algorithm and similarity threshold to ensure the accuracy and reasonableness of the matched assets. The assets obtained from each layer are then integrated to form the set of matched assets. This solution addresses the problems of inconsistent asset styles and low matching efficiency in existing technologies through incremental generation consistency verification and hierarchical asset feature library division, ensuring the consistency of the integrated visual assets while improving the efficiency and accuracy of asset matching.

[0043] Technical problems with existing technologies: The visual assets generated by existing solutions lack a unified standardized coding system, which makes it extremely difficult to retrieve, reuse and manage the assets, and makes it impossible to achieve full lifecycle management and iterative optimization of the assets.

[0044] Based on this, the standardization encoding and storage of target multidimensional visual assets in a hierarchical visual asset feature library includes extracting semantic feature information, generation parameter information, and narrative feature information bound to the target multidimensional visual assets; generating standardized management codes based on the semantic feature information, generation parameter information, and narrative feature information; binding the standardized management codes with the target multidimensional visual assets; storing the bound target multidimensional visual assets and standardized management codes in the hierarchical visual asset feature library; and iteratively updating the standardized management codes based on asset reuse data.

[0045] It is worth mentioning that when extracting semantic feature information, generation parameter information, and narrative feature information bound to the target multidimensional visual asset, structured information corresponding to the target multidimensional visual asset is retrieved from the database of the industrial big data processing platform. Semantic feature information includes the semantic units corresponding to the asset, explicit and implicit semantic dimensions, etc. Generation parameter information includes various parameter settings during asset generation, visual generation correction parameters, etc. Narrative feature information includes the narrative rhythm intensity and narrative tension corresponding to the asset. These three types of information comprehensively cover the core attributes of the target multidimensional visual asset, providing complete data support for the generation of standardized management codes. When generating standardized management codes based on these three types of information, industrial-grade visual asset coding rules are adopted. The codes consist of a fixed number of digits and letters, with different segments of the code corresponding to different feature information. For example, the first segment corresponds to semantic feature information, the middle segment corresponds to generation parameter information, and the last segment corresponds to narrative feature information. This ensures that the code can uniquely identify the core attributes of the target multidimensional visual asset, while facilitating rapid asset retrieval and classification management. When binding standardized management codes with target multidimensional visual assets, a metadata binding method is used. The code, as the core metadata of the target multidimensional visual asset, is embedded into the asset's file information, ensuring a tight binding between the code and the asset and preventing loss due to asset copying or transmission. When storing the bound target multidimensional visual assets and standardized management codes in a hierarchical visual asset feature library, the assets are assigned to the corresponding hierarchical asset libraries based on their core attributes. Simultaneously, the standardized management code is incorporated into the feature library's index system, serving as the core basis for asset retrieval, thus achieving standardized asset storage and rapid retrieval.

[0046] During the iterative update of standardized management codes based on asset reuse data, it is necessary to collect reuse data such as the number of times assets are reused, reuse scenarios, and matching similarity in real time. This data is then used as additional information to update the extended segments of the standardized management codes, allowing the codes to reflect the reuse status of assets in real time. At the same time, the indexing system of the feature library is optimized based on the reuse data to improve the efficiency and accuracy of asset matching, thereby achieving full lifecycle management and iterative optimization of visual assets.

[0047] This solution addresses the challenges of existing technologies in asset management and the inability to iteratively optimize assets through standardized coding and iterative update mechanisms that integrate multi-dimensional information. It enables full lifecycle management and continuous optimization of assets, thereby further improving asset reuse rates.

[0048] Technical problems with existing technologies: The script content to be processed has problems such as inconsistent format and a lot of redundant content. Directly performing semantic extraction will lead to insufficient accuracy of semantic feature extraction, which will affect the accuracy of subsequent processing.

[0049] Based on this, obtaining the script content to be processed includes performing text cleaning on the script content to be processed, removing invalid format characters and redundant content contained in the script content to be processed, and performing format unification processing on the script content to be processed after text cleaning.

[0050] It's worth noting that the text cleaning process for the script content relies on a text cleaning algorithm. This algorithm can accurately identify various invalid formatting characters in the script text, including spaces, line breaks, special symbols, and formatting marks. It can also identify redundant content in the script text, including repeated lines, invalid annotations, and formatting instructions. Identified invalid formatting characters and redundant content are deleted in batches, ensuring that the cleaned script text retains only the core content information. When performing format standardization on the cleaned script content, it is standardized according to the standard format of industrial-grade script texts. This includes standardizing the layout format of lines, character identifiers, scene identifiers, and plot nodes, ensuring that the formatted script text has standardized and consistent structural features, facilitating subsequent semantic extraction and sentence / block segmentation. Text cleaning and format unification are both implemented in the preprocessing module of the industrial big data processing platform. This module supports batch processing of script files in various formats, and its processing efficiency meets the batch processing needs of industrial-grade digital content production. The processed script content is stored in the platform's cache module, providing high-quality basic text data for subsequent full-dimensional semantic deconstruction and weight mapping of the script. This solution solves the problems of chaotic formatting and redundant content in the original script content through script content preprocessing, improves the accuracy of subsequent semantic feature extraction, and provides reliable basic data for the stable implementation of the entire technical solution.

[0051] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for generating and managing multi-dimensional visual assets driven by script content, characterized in that, include: S1: Obtain the script content to be processed, and perform full-dimensional semantic deconstruction and weight mapping processing on the script content to obtain a set of semantic units and a comprehensive semantic weight set bound to the set of semantic units. S2: Based on the semantic unit set and the comprehensive semantic weight set, perform narrative rhythm quantization and generation parameter adaptation processing to obtain a visual generation correction parameter set. S3: Based on the comprehensive semantic weight set and the visual generation correction parameter set, hierarchical asset matching and standardized management processing are performed to obtain the target multidimensional visual asset.

2. The method for generating and managing multi-dimensional visual assets driven by script content according to claim 1, characterized in that, The script full-dimensional semantic deconstruction and weight mapping process includes extracting explicit and implicit semantic dimension information contained in the script content to be processed, dividing the script content to be processed into a set of semantic units based on the explicit and implicit semantic dimension information, and performing normalization calculation on the explicit and implicit semantic dimension information to obtain a comprehensive semantic weight set bound to the set of semantic units.

3. The method for generating and managing multi-dimensional visual assets driven by script content according to claim 2, characterized in that, The comprehensive semantic weight set is calculated using the script's full-dimensional semantic weight calculation formula, which is as follows: ； in, The comprehensive semantic weight bound to the i-th semantic unit in the set of semantic units. The coefficients are used to assign explicit semantic weights. This represents the total number of explicit semantic dimensions. The weight coefficients bound to the m-th explicit semantic dimension. Let be the normalized feature value bound to the i-th semantic unit in the m-th explicit semantic dimension in the semantic unit set. This represents the total number of implicit semantic dimensions. The weight coefficients bound to the nth implicit semantic dimension. It is the normalized feature value bound to the i-th semantic unit in the n-th implicit semantic dimension in the set of semantic units.

4. The method for generating and managing multi-dimensional visual assets driven by script content according to claim 1, characterized in that, The narrative rhythm quantification and generation parameter adaptation process includes extracting narrative rhythm dimension information contained in the semantic unit set, normalizing the narrative rhythm dimension information to obtain the narrative rhythm intensity set bound to the semantic unit set, and combining the comprehensive semantic weight set and the narrative rhythm intensity set to calculate the visual generation correction parameter set.

5. The method for generating and managing multi-dimensional visual assets driven by script content according to claim 1, characterized in that, The hierarchical asset matching and standardized management process includes constructing a hierarchical visual asset feature library, performing feature matching from the hierarchical visual asset feature library based on a comprehensive semantic weight set and a visual generation correction parameter set to obtain a matching asset set, calculating the asset reusability set contained in the matching asset set, filtering based on the asset reusability set to obtain a reusable asset set, performing incremental generation processing on the content not covered by the reusable asset set to obtain an incrementally generated asset set, binding the reusable asset set and the incrementally generated asset set to obtain the target multidimensional visual asset, and standardizing and encoding the target multidimensional visual asset and storing it in the hierarchical visual asset feature library.

6. The method for generating and managing multi-dimensional visual assets driven by script content according to claim 2, characterized in that, The explicit semantic dimension information includes character dimension information, scene dimension information, prop dimension information, action dimension information, and dialogue dimension information. The implicit semantic dimension information includes narrative rhythm dimension information, emotion transmission dimension information, foreshadowing constraint dimension information, and subtext dimension information. The process of dividing the script content to be processed into a semantic unit set based on explicit and implicit semantic dimension information includes processing the script content to be processed into sentences and blocks according to the script's narrative logic, and binding the processed content with explicit and implicit semantic dimension information to obtain the semantic unit set.

7. The method for generating and managing multi-dimensional visual assets driven by script content according to claim 4, characterized in that, The narrative rhythm dimension information includes conflict density information, scene switching frequency information, dialogue density information, and emotional fluctuation amplitude information. The normalization calculation of the narrative rhythm dimension information to obtain the narrative rhythm intensity set bound to the semantic unit set includes normalizing a single narrative rhythm dimension information and weighting the normalized multi-dimensional information to obtain the narrative rhythm intensity set bound to the semantic unit set.

8. The method for generating and managing multi-dimensional visual assets driven by script content according to claim 5, characterized in that, The incremental generation process includes setting visual asset generation parameters based on a comprehensive semantic weight set and a visual generation correction parameter set, performing generation operations based on the visual asset generation parameters to obtain initial incremental generated assets, performing consistency verification on the initial incremental generated assets, and including the initial incremental generated assets that pass the consistency verification as the content of the incremental generated asset set. The hierarchical visual asset feature library includes a core creative layer asset library, a narrative framework layer asset library, and an element detail layer asset library. The step of obtaining a matching asset set by feature matching from the hierarchical visual asset feature library includes performing feature matching from the core creative layer asset library, the narrative framework layer asset library, and the element detail layer asset library respectively.

9. A method for generating and managing multi-dimensional visual assets driven by script content according to claim 5, characterized in that, The process of standardizing and encoding the target multidimensional visual assets and storing them in a hierarchical visual asset feature library includes extracting semantic feature information, generation parameter information, and narrative feature information bound to the target multidimensional visual assets; generating standardized management codes based on the semantic feature information, generation parameter information, and narrative feature information; binding the standardized management codes to the target multidimensional visual assets; storing the bound target multidimensional visual assets and standardized management codes in the hierarchical visual asset feature library; and iteratively updating the standardized management codes based on asset reuse data.

10. A method for generating and managing multi-dimensional visual assets driven by script content according to claim 1, characterized in that, The process of obtaining the script content to be processed includes performing text cleaning on the script content to be processed, removing invalid format characters and redundant content, and performing format unification processing on the script content to be processed after text cleaning.

Images