Multidimensional feature fusion-based short animation plot emotional driving plot generation method

Abstract

The application relates to the technical field of artificial intelligence content generation, and discloses a multi-dimensional feature fusion cartoon short story emotional driving plot generation method. The application constructs five independent quantitative indexes, namely an emotional tendency coefficient, a character relationship tension coefficient, an emotional fluctuation coefficient, a scene emotional matching coefficient and a conflict accumulation coefficient, quantifies the overall character emotional tone of the plot, the interpersonal opposition degree between characters, the adjacent plot node emotional jump amplitude, the scene atmosphere and the character emotional fit degree in a continuous plot window, and then obtains a comprehensive evaluation coefficient through weight weighted summation operation, uniformly integrates multi-dimensional narrative features, and further calculates an emotional transfer distance for representing the deviation of the narrative rhythm and dynamically controlling the subsequent plot development based on the emotional transfer distance.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence content generation technology, specifically a method for generating short animated stories driven by multi-dimensional feature fusion. Background Technology

[0002] With the rapid development of industrialized automatic generation technology for animated short dramas and micro-dramas, automatic plot continuation solutions based on large models have become a core technical means for mass content production. Most existing plot generation models rely solely on textual semantics to complete plot continuation, and can only judge the plot direction from the level of dialogue and plot text. They lack quantitative modeling of multi-dimensional narrative features such as character emotions, character relationships, scene atmosphere, and plot conflicts, resulting in many unavoidable technical defects in the actual generation process.

[0003] First, existing technologies cannot quantify and distinguish multi-dimensional emotional narrative features, resulting in fragmented emotional elements. Traditional generation models rely solely on the semantics of single sentences to determine emotional trends, failing to establish standardized quantitative indicators to represent five core features: overall character emotional bias, degree of interpersonal conflict between characters, amplitude of emotional fluctuations in adjacent plots, matching degree between scene atmosphere and character emotions, and cumulative conflict in consecutive paragraphs. This makes it impossible to simultaneously manage multiple dimensions such as character emotions, interpersonal conflicts, scene tone, and dramatic conflict, easily leading to narrative breakdowns such as inconsistencies in character portrayal, abrupt emotional shifts, inconsistencies between scene atmosphere and character emotions, and fluctuating intensity of plot conflicts. Consequently, the generated content exhibits poor logical self-consistency.

[0004] Secondly, the lack of a comprehensive quantitative evaluation mechanism integrating multiple indicators means that there is no unified numerical basis for plot control. The current solution can only roughly define the plot style through text prompts, and there is no quantifiable comprehensive evaluation standard. It cannot simultaneously weigh the impact of multiple narrative elements such as character emotions, character relationships, scene adaptation, and conflict accumulation on the plot quality, and cannot accurately determine whether the current plot segment conforms to the preset narrative rhythm. When the plot continues or plot twists are triggered, it can only rely on random generation by the model, making it difficult to stably control the emotional and conflict rhythm of the short drama's development. The content style and dramatic tension of the mass-produced content fluctuate greatly, resulting in insufficient controllability. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for generating short animated stories driven by emotional appeal through multi-dimensional feature fusion. This method has advantages such as quantifying narrative emotional indicators from multiple dimensions, weighted comprehensive evaluation of the story, precise control of emotions and conflict rhythm, prevention of character derailment, and improvement of the consistency and controllability of generated story logic, thus solving the aforementioned technical problems.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for generating short animated stories driven by emotional resonance through multi-dimensional feature fusion, comprising the following steps:

[0007] S1: Collect multi-dimensional data on characters in the anime storyline;

[0008] S2: Based on the collected multi-dimensional data, multi-dimensional evaluation coefficients are calculated, including emotional tendency coefficient, role relationship tension coefficient, emotional fluctuation coefficient, scene emotional matching coefficient, and conflict accumulation coefficient;

[0009] S3: The comprehensive evaluation coefficient is obtained by weighted summation of the evaluation coefficients from multiple dimensions;

[0010] S4: Emotional transfer distance is calculated based on a comprehensive evaluation coefficient;

[0011] S5: Predict the direction of the subsequent plot based on the distance of emotional transfer.

[0012] As a preferred technical solution of the present invention, the S1 multi-dimensional data includes the character's emotional state value, the sequence of favorability between characters, the current scene atmosphere value, and the plot tension of the current narrative node;

[0013] The expression for the emotional state value is: ,in, Represents a set of emotional states. Indicates the total number of characters. , and They represent the first The, the The and the first The emotional state value of each character.

[0014] As a preferred embodiment of the present invention, the expression for the affinity sequence between characters is as follows:

[0015] ;

[0016] in, This indicates the sequence of affection levels between characters; Indicates the first The first role and the first The favorability rating of each character; Indicates the first The first role and the first The favorability rating of each character; Indicates the first The first role and the first The favorability rating of each character.

[0017] As a preferred embodiment of the present invention, the expression for the S2 sentiment tendency coefficient is as follows:

[0018] ;

[0019] in, Indicates the sentiment tendency coefficient; Indicates the total number of characters; Indicates the first The emotional state value of each character. Positive values ​​represent positive emotions, while negative values ​​represent negative emotions; This indicates a summation operation.

[0020] As a preferred embodiment of the present invention, the expression for the S2 role relationship tension coefficient is as follows:

[0021] ;

[0022] in, This indicates the tension level in the relationship between the characters. Indicates the length of the affinity sequence between characters; Indicates the first The first role and the first The favorability rating of each character A positive value indicates a friendly relationship between the two characters, while a negative value indicates a hostile relationship. This indicates a summation operation.

[0023] As a preferred embodiment of the present invention, the expression for the S2 emotional fluctuation coefficient is as follows:

[0024] ;

[0025] in, Indicates the emotional fluctuation coefficient; This represents the emotional state value of the character at the current node. This represents the emotional state value of the character in the previous node; It represents Euclidean distance.

[0026] As a preferred embodiment of the present invention, the expression for the S2 scene emotion matching coefficient is as follows:

[0027] ;

[0028] in, Indicates the scene sentiment matching coefficient; Indicates the total number of characters; Indicates the current scene atmosphere value; Indicates the first The emotional state value of each character; This represents the summation operation; It represents the absolute value.

[0029] As a preferred embodiment of the present invention, the expression for the S2 conflict accumulation coefficient is as follows:

[0030] ;

[0031] in, Indicates the cumulative conflict coefficient; Indicates recent The number of times a node conflict event occurs; Indicates the magnification factor; Indicates the length of the statistical window; This indicates taking the minimum value.

[0032] As a preferred embodiment of the present invention, the expression for the S3 comprehensive evaluation coefficient is as follows:

[0033] ;

[0034] in, This represents the comprehensive evaluation coefficient; Indicates the first The weights corresponding to each evaluation coefficient; Indicates the first Each evaluation coefficient This is the sentiment index. This represents the tension coefficient in the relationship between the characters. The emotional fluctuation coefficient. The scene emotion matching coefficient, This is the cumulative conflict coefficient; This indicates a summation operation.

[0035] As a preferred embodiment of the present invention, the expression for the S4 emotion transfer distance is as follows:

[0036] ;

[0037] in, Indicates the distance of emotional transference; This represents the comprehensive evaluation coefficient; Indicates the target threshold; Represents absolute value;

[0038] The S5 method determines the subsequent plot direction based on the distance of emotional transfer, specifically: when the distance of emotional transfer... If the current plot development is deemed reasonable, the narrative pace will be maintained, and no special events will be triggered; if the first threshold is less than the threshold, the plot will be considered reasonable. If the value is less than the second threshold, then it is determined that the plot direction needs to be adjusted. < If so, a collaboration event is triggered. ≥ This will trigger a conflict event.

[0039] Compared with existing technologies, this invention provides a method for generating short animated stories driven by emotion through multi-dimensional feature fusion, which has the following beneficial effects:

[0040] 1. This invention constructs five quantitative indicators: emotional tendency, tension in character relationships, emotional fluctuations, scene emotional matching, and conflict accumulation. It can comprehensively quantify narrative dimensions such as character emotions, interpersonal conflicts, scene atmosphere, and conflict accumulation, thereby solving the defects of traditional solutions in the fragmentation of emotional characteristics and the inability to control them synchronously. It also avoids narrative problems such as character collapse, scene incongruity, and emotional jumpiness.

[0041] 2. This invention obtains a comprehensive evaluation coefficient by weighted integration of five indicators, and introduces emotional transfer distance to quantify the deviation between the current plot and the target narrative, forming a standardized numerical control basis. This can accurately constrain the rhythm of short dramas, greatly improve the controllability of plot generation, and produce content with stable and unified style and dramatic tension in batches. Attached Figure Description

[0042] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] Please see Figure 1 A method for generating short animated stories driven by emotion through multi-dimensional feature fusion includes the following steps:

[0045] S1: Collect multi-dimensional data on characters in the anime storyline;

[0046] Multi-dimensional data includes the character's emotional state value, the sequence of favorability between characters, the atmosphere value of the current scene, and the plot tension of the current narrative node;

[0047] The expression for the emotional state value is: ,in, Represents a set of emotional states. Indicates the total number of characters. , and They represent the first The, the The and the first The emotional state value of each character;

[0048] The expression for the affinity sequence between characters is as follows:

[0049] ;

[0050] in, This indicates the sequence of affection levels between characters; Indicates the first The first role and the first The favorability rating of each character; Indicates the first The first role and the first The favorability rating of each character; Indicates the first The first role and the first The favorability rating of each character;

[0051] S2: Based on the collected multi-dimensional data, multi-dimensional evaluation coefficients are calculated, including emotional tendency coefficient, role relationship tension coefficient, emotional fluctuation coefficient, scene emotional matching coefficient, and conflict accumulation coefficient;

[0052] The expression for the sentiment index is as follows:

[0053] ;

[0054] in, Indicates the sentiment tendency coefficient; Indicates the total number of characters; Indicates the first The emotional state value of each character. Positive values ​​represent positive emotions, while negative values ​​represent negative emotions; This represents the summation operation. It reflects the overall emotional bias of a single plot segment. >0:00 The overall plot leans towards optimism and positive emotions. The overall tone of the plot at time 0 is depressing and negative. When the time is approximately 0, the plot does not have a clear single emotional tendency, which makes it easier to adjust the plot generation direction according to the preset emotional rhythm.

[0055] The expression for the tension coefficient in the character relationship is as follows:

[0056] ;

[0057] in, This indicates the tension level in the relationship between the characters. Indicates the length of the affinity sequence between characters; Indicates the first The first role and the first The favorability rating of each character A positive value indicates a friendly relationship between the two characters, while a negative value indicates a hostile relationship. Indicates the summation operation; when When >0, the characters in the story generally get along harmoniously and there is little conflict; when When <0, there are generally conflicting characters within the plot, resulting in stronger dramatic conflict; when When the value is approximately 0, the friendly and hostile relationships between characters are balanced.

[0058] The expression for the emotional fluctuation coefficient is as follows:

[0059] ;

[0060] in, Indicates the emotional fluctuation coefficient; This represents the emotional state value of the character at the current node. This represents the emotional state value of the character in the previous node; This represents the Euclidean distance. The larger the value of this coefficient, the stronger the overall emotional contrast between the characters and the greater the emotional jump between two adjacent plot segments; the smaller the value, the smoother and more coherent the emotional transition in the plot.

[0061] The expression for the scene sentiment matching coefficient is as follows:

[0062] ;

[0063] in, Indicates the scene sentiment matching coefficient; Indicates the total number of characters; Indicates the current scene atmosphere value; Indicates the first The emotional state value of each character; This represents the summation operation; Represents absolute value. The closer the value is to 1, the more unified the scene atmosphere is with the emotional state of all the characters, and the stronger the narrative coordination. The closer the value is to 0, the greater the difference between the scene atmosphere and the character's emotions, and the poor the match.

[0064] The expression for the conflict accumulation coefficient is as follows:

[0065] ;

[0066] in, Indicates the cumulative conflict coefficient; Indicates recent The number of times a node conflict event occurs; Indicates the magnification factor; Indicates the length of the statistical window; This indicates taking the minimum value. The higher the value, the more frequent the conflicts and the greater the accumulation of contradictions in the recent plot; the lower the value, the more leisurely the plot and the lack of dramatic conflict.

[0067] S3: The comprehensive evaluation coefficient is obtained by weighted summation of the evaluation coefficients from multiple dimensions;

[0068] The expression for the comprehensive evaluation coefficient is as follows:

[0069] ;

[0070] in, This represents the comprehensive evaluation coefficient; Indicates the first The weights corresponding to each evaluation coefficient; Indicates the first Each evaluation coefficient This is the sentiment index. This represents the tension coefficient in the relationship between the characters. The emotional fluctuation coefficient. The scene emotion matching coefficient, This is the cumulative conflict coefficient; This represents a summation operation. This coefficient can uniformly summarize five core narrative features: character emotions, interpersonal conflicts, emotional shifts, scene adaptation, and conflict accumulation. The value of the comprehensive evaluation coefficient can be used to determine whether the current plot segment conforms to the preset narrative rhythm. This serves as a quantitative basis for plot transitions and plot continuation control, solving the shortcomings of existing plot generation schemes that lack a unified comprehensive evaluation standard and cannot coordinate the control of multi-dimensional emotional features.

[0071] S4: The emotional transfer distance is calculated based on the comprehensive evaluation coefficient, and its expression is as follows:

[0072] ;

[0073] in, Indicates the distance of emotional transference; This represents the comprehensive evaluation coefficient; Indicates the target threshold; Represents absolute value; The larger the value, the greater the deviation between the current overall emotional and conflict state of the plot and the expected narrative rhythm; the closer the value is to 0, the better the plot state is in line with the preset goal.

[0074] S5: Predicting subsequent plot developments based on emotional transfer distance, specifically: when the emotional transfer distance... If the current plot development is deemed reasonable, the narrative pace will be maintained, and no special events will be triggered; if the first threshold is less than the threshold, the plot will be considered reasonable. If the value is less than the second threshold, then it is determined that the plot direction needs to be adjusted. < This triggers collaborative events, such as role reconciliation and mutual assistance, thus shifting emotions in a positive direction. ≥ This can trigger conflict events, such as role disputes and disagreements, reducing the shift of emotions towards a negative direction.

[0075] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for generating short animated stories driven by emotional engagement through multi-dimensional feature fusion, characterized by: Includes the following steps: S1: Collect multi-dimensional data on characters in the anime storyline; S2: Based on the collected multi-dimensional data, multi-dimensional evaluation coefficients are calculated, including emotional tendency coefficient, role relationship tension coefficient, emotional fluctuation coefficient, scene emotional matching coefficient, and conflict accumulation coefficient; S3: The comprehensive evaluation coefficient is obtained by weighted summation of the evaluation coefficients from multiple dimensions; S4: Emotional transfer distance is calculated based on a comprehensive evaluation coefficient; S5: Predict the direction of the subsequent plot based on the distance of emotional transfer.

2. The method for generating short animated stories driven by emotional resonance through multi-dimensional feature fusion as described in claim 1, characterized in that: The S1 multi-dimensional data includes the character's emotional state value, the sequence of favorability between characters, the current scene atmosphere value, and the plot tension of the current narrative node; The expression for the emotional state value is: ,in, Represents a set of emotional states. Indicates the total number of characters. , and They represent the first The, the The and the first The emotional state value of each character.

3. The method for generating short animated stories driven by emotional resonance through multi-dimensional feature fusion as described in claim 2, characterized in that: The expression for the affinity sequence between the characters is as follows: ； in, This indicates the sequence of affection levels between characters; Indicates the first The first character and the first The favorability rating of each character; Indicates the first The first character and the first The favorability rating of each character; Indicates the first The first character and the first The favorability rating of each character.

4. The method for generating short animated stories driven by emotional resonance through multi-dimensional feature fusion as described in claim 3, characterized in that: The expression for the S2 sentiment index is as follows: ； in, Indicates the sentiment tendency coefficient; Indicates the total number of characters; Indicates the first The emotional state value of each character. Positive values ​​represent positive emotions, while negative values ​​represent negative emotions; This indicates a summation operation.

5. The method for generating short animated stories driven by emotion through multi-dimensional feature fusion as described in claim 4, characterized in that: The expression for the S2 role relationship tension coefficient is as follows: ； in, This indicates the tension level in the relationship between the characters. Indicates the length of the affinity sequence between characters; Indicates the first The first character and the first The favorability rating of each character A positive value indicates a friendly relationship between the two characters, while a negative value indicates a hostile relationship. This indicates a summation operation.

6. The method for generating short animated stories driven by emotion through multi-dimensional feature fusion according to claim 5, characterized in that: The expression for the S2 sentiment fluctuation coefficient is as follows: ； in, Indicates the emotional fluctuation coefficient; This represents the emotional state value of the character at the current node. This represents the emotional state value of the character in the previous node; It represents Euclidean distance.

7. The method for generating short animated stories driven by emotional resonance through multi-dimensional feature fusion as described in claim 6, characterized in that: The expression for the S2 scene sentiment matching coefficient is as follows: ； in, Indicates the scene sentiment matching coefficient; Indicates the total number of characters; Indicates the current scene atmosphere value; Indicates the first The emotional state value of each character; This represents the summation operation; It represents the absolute value.

8. The method for generating short animated stories driven by emotion through multi-dimensional feature fusion as described in claim 7, characterized in that: The expression for the S2 conflict accumulation coefficient is as follows: ； in, Indicates the cumulative conflict coefficient; Indicates recent The number of times a node conflict event occurs; Indicates the magnification factor; Indicates the length of the statistical window; This indicates taking the minimum value.

9. The method for generating short animated stories driven by emotion through multi-dimensional feature fusion as described in claim 8, characterized in that: The expression for the S3 comprehensive evaluation coefficient is as follows: ； in, This represents the comprehensive evaluation coefficient; Indicates the first The weights corresponding to each evaluation coefficient; Indicates the first Each evaluation coefficient This is the sentiment index. This represents the tension coefficient in the relationship between the characters. The emotional fluctuation coefficient. The scene emotion matching coefficient, This is the cumulative conflict coefficient; This indicates a summation operation.

10. The method for generating short animated stories driven by emotion through multi-dimensional feature fusion according to claim 9, characterized in that: The expression for the S4 emotion transfer distance is as follows: ； in, Indicates the distance of emotional transference; This represents the comprehensive evaluation coefficient; Indicates the target threshold; Represents absolute value; The S5 method determines the subsequent plot direction based on the distance of emotional transfer, specifically: when the distance of emotional transfer... If the current plot development is deemed reasonable, the narrative pace will be maintained, and no special events will be triggered; if the first threshold is less than the threshold, the plot will be considered reasonable. If the value is less than the second threshold, then it is determined that the plot direction needs to be adjusted. < If so, a collaboration event is triggered. ≥ This will trigger a conflict event.

Images