Method for Generating Dynamic Expression Elements of a Character and Server Performing the Same

Abstract

A method for generating dynamic expression elements of a character according to the present invention, wherein the method for generating dynamic expression elements of a character is performed by a server, comprises the steps of acquiring character information, confirming a plurality of external feature information of the character from the character information, and generating dynamic expression elements of the character to be applied to video content including the character based on the plurality of external feature information.

Description

Technology Field

[0001] The present invention relates to automatically generating various expression elements based on the appearance information of a character, and more specifically, to a method for generating dynamic expression elements of a character and a server for performing the same. Background Technology

[0002] With the recent advancement of generative AI technology, various content creation tools are emerging, such as text-based image generation, speech synthesis, and motion capture automation. These technologies help partially automate the production process of digital characters used in video, games, and the metaverse.

[0003] However, most current tools are limited to individual expressive elements and fail to automatically guarantee consistency or interoperability between a character's appearance and elements such as voice, movement, and facial expressions. Consequently, inconsistencies may arise between the character's appearance and persona, or there is the inconvenience of having to make manual adjustments.

[0004] Therefore, there is an increasing demand for technology capable of consistently and automatically generating multidimensional dynamic expression elements based on information regarding the external characteristics of characters. Prior art literature

[0005] Korean Patent Publication No. 10-2729504 (Published November 8, 2024) The problem to be solved

[0006] The present invention aims to provide a method for identifying a plurality of external feature information of a character from character information and generating dynamic expression elements of a character to be applied to video content including the character based on the plurality of external feature information.

[0007] In addition, the present invention aims to provide a method for identifying contribution information for each of a plurality of external feature information related to the generation of a dynamic expression element, and generating the dynamic expression element based on the contribution information. means of solving the problem

[0008] A method for generating dynamic expression elements of a character according to the present invention, wherein the method for generating dynamic expression elements of a character is performed by a server, comprises the steps of acquiring character information, confirming a plurality of external feature information of the character from the character information, and generating dynamic expression elements of the character to be applied to video content including the character based on the plurality of external feature information.

[0009] In one embodiment of the present invention, the dynamic expression element may include at least one of voice, movement, facial expression, and tone of voice.

[0010] In one embodiment of the present invention, the plurality of external feature information may include at least one of a mesh form type and a texture skin.

[0011] In one embodiment of the present invention, the step of generating the dynamic expression element may include the step of verifying contribution information for each of the plurality of external feature information related to the generation of the dynamic expression element, and the step of generating the dynamic expression element based on the contribution information.

[0012] In one embodiment of the present invention, the dynamic expression element includes a plurality of types, and the contribution information related to the generation of the plurality of types of dynamic expression elements may be configured differently from each other.

[0013] In one embodiment of the present invention, the contribution information may be determined according to the degree to which each of the plurality of external feature information deviates from a preset standard range.

[0014] In one embodiment of the present invention, the step of confirming the contribution may include the step of obtaining information regarding the contribution information from a user terminal.

[0015] In one embodiment of the present invention, step (c) may include the step of generating a first dynamic expression element based on the plurality of external feature information and the step of generating a second dynamic expression element based on the plurality of external feature information and the first dynamic expression element.

[0016] In one embodiment of the present invention, the dynamic expression element includes a first dynamic expression element and a second dynamic expression element, and generating the first dynamic expression element and the second dynamic expression element in step (c) may include the step of generating the first dynamic expression element and the second dynamic expression element together so that they correspond to each other.

[0017] In one embodiment of the present invention, step (c) may include the step of generating a feature vector for the character based on the plurality of external feature information and the step of generating the dynamic expression element based on the feature vector.

[0018] In one embodiment of the present invention, step (c) may include the step of selecting a generative artificial intelligence model according to the type of the dynamic representation element and the step of generating a prompt to be input to the selected generative artificial intelligence model based on the plurality of external feature information.

[0019] In one embodiment of the present invention, step (a) includes the step of obtaining basic information in the form of text for the character, and step (b) may include the step of determining a plurality of external feature information constituting the character based on the basic information and the step of generating external information of the character based on the plurality of external feature information.

[0020] In one embodiment of the present invention, generating the dynamic expression element in step (c) may be performed based on the plurality of external feature information and the basic information.

[0021] In one embodiment of the present invention, after step (c), the method may further include the steps of obtaining modification information for the dynamic representation element from a user terminal, determining whether there is a discrepancy between the modified dynamic representation element and the plurality of external feature information, and, if the discrepancy is confirmed, changing at least some of the plurality of external feature information based on the modified dynamic representation element.

[0022] A server includes a memory and at least one processor connected to communicate with said memory, and said processor is configured to acquire character information, identify a plurality of external feature information of said character from said character information, and generate dynamic expression elements of said character to be applied to video content including said character based on said plurality of external feature information. Effects of the invention

[0023] According to one embodiment of the present invention, a method can be provided for identifying a plurality of external feature information of a character from character information and generating dynamic expression elements of a character to be applied to video content including the character based on the plurality of external feature information.

[0024] In addition, according to one embodiment of the present invention, a method may be provided for identifying contribution information for each of a plurality of external feature information related to the generation of a dynamic expression element, and generating the dynamic expression element based on the contribution information. Brief explanation of the drawing

[0025] FIG. 1 is a diagram illustrating an example of a network environment according to one embodiment of the present invention. FIG. 2 is a flowchart illustrating a method for a server to generate dynamic representation elements of a character according to an embodiment of the present invention. FIG. 3 is a flowchart illustrating a method for a server to generate dynamic representation elements based on contribution information according to an embodiment of the present invention. FIG. 4 is a flowchart illustrating a method for a server to generate dynamic representation elements of a character according to another embodiment of the present invention. FIG. 5 is a table for explaining a configuration in which a server contributes differently to a plurality of external feature information according to the type of dynamic expression element, according to an embodiment of the present invention. FIG. 6 is a table for explaining the process of a server generating dynamic representation elements in conjunction with each other according to an embodiment of the present invention. FIG. 7 is a drawing illustrating an example of a screen in which a server generates and provides dynamic expression elements based on a plurality of external feature information of a character, according to an embodiment of the present invention. Specific details for implementing the invention

[0026] Hereinafter, embodiments disclosed in this specification will be described in detail with reference to the attached drawings. Identical or similar components regardless of drawing symbols are assigned the same reference number, and redundant descriptions thereof will be omitted. Furthermore, in describing the embodiments disclosed in this specification, if it is determined that a detailed description of related prior art could obscure the essence of the embodiments disclosed in this specification, such detailed description will be omitted.

[0027] Terms including ordinal numbers, such as first, second, etc., may be used to describe various components, but said components are not limited by said terms. These terms are used solely for the purpose of distinguishing one component from another.

[0028] A singular expression includes a plural expression unless the context clearly indicates otherwise.

[0029] In this application, each step described may be performed regardless of the order listed, except where it must be performed in the order listed by a particular causal relationship.

[0030] In this application, terms such as “comprising” or “having” are intended to specify the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0032] The present invention will be described below with reference to the attached drawings.

[0033] FIG. 1 is a diagram illustrating an example of a network environment according to one embodiment of the present invention.

[0034] A network environment according to one embodiment of the present invention illustrated in FIG. 1 may include a server (10), a user terminal (20), and a generative artificial intelligence model (30).

[0035] The server (10) is a device that provides a character automatic implementation service. Here, the character automatic implementation service may provide a function to automatically generate or modify dynamic expression elements based on the external feature information of the character to the user terminal (20).

[0036] Specifically, the character automatic implementation service may include a function to generate or modulate various dynamic expression elements, such as voice, movement, facial expressions, and speech patterns, by utilizing the character's appearance information, feature vectors, contribution information, and content context information. For example, based on basic character information or appearance data input by a user, it may generate and provide voice and animation data suitable for the character.

[0037] More specifically, the server (10) can obtain multiple external feature information through the analysis of character information, and based on the feature information, call a generative artificial intelligence model (30) to generate dynamic expression elements, or inversely correct the external information according to user modifications and then provide the result to the user terminal (20).

[0038] The network is not limited to a specific communication method and may include communication methods utilizing communication networks that the network can include (e.g., mobile communication networks, wired internet, wireless internet, broadcasting networks) and electrical signals (analog signals, digital signals, serial signals, Pulse Width Modulation), as well as short-range wireless communication.

[0039] The server (10) may be implemented as a computer device or a plurality of computer devices that provide commands, code, files, content, services, etc. The server (10) may be a server (10) capable of transmitting and receiving information by communicating with a user terminal (20) and a generative artificial intelligence model (30) through a network.

[0040] The server (10) may include a processor (11), memory (12), and a communication unit (13). The processor (11) may control the overall operation of the memory (12) and the communication unit (13). According to various embodiments of the present invention, the memory (12) functions as a storage medium and may store a number of applications running on the server (10), data for the operation of the server (10), and instructions. In one embodiment, the memory (12) may be provided in hardware in the form of various storage devices such as ROM, RAM, flash drive, hard drive, etc., or in the form of web storage. The communication unit (13) may communicate with a user terminal (20) via a network in a wired or wireless manner.

[0041] The processor (11) can provide a character automatic implementation service by controlling the overall operation of the memory (12) and the communication unit (13).

[0042] The memory (12) functions as a storage medium and can store a number of applications running on the server (10), data for the operation of the server (10), and commands. In one embodiment, an application related to a character automatic implementation service can be stored in the memory (12). This memory (12) may be provided in hardware in the form of various storage devices such as ROM, RAM, flash drive, hard drive, etc., or in the form of web storage.

[0043] The communication unit (13) can communicate with the user terminal (20) and the generative artificial intelligence model (30) via a network in a wired / wireless manner.

[0044] The server (10) of the present invention performs a method for generating dynamic expression elements of a character. Specifically, the server (10) obtains character information, identifies a plurality of external feature information of the character from the character information, and generates dynamic expression elements of the character to be applied to video content containing the character based on the plurality of external feature information.

[0045] Additionally, the server (10), in generating dynamic expression elements, checks contribution information for each of the multiple external feature information related to the generation of dynamic expression elements, and generates dynamic expression elements based on the contribution information.

[0046] Here, a character refers to a virtual person or object expressed in various digital environments, such as video content, games, the metaverse, and virtual reality. A character may be a person, animal, robot, living organism, or a created imaginary being, and can be realized as a subject possessing an independent persona through visual information and dynamic expression elements.

[0047] Here, character information is input data for verifying information on the external characteristics of a character, and may include structured or unstructured data representing the appearance or settings of a character. Character information may be in the form of, for example, a 3D model file, a 2D image, an illustration, a character setting document, a scenario, a text description, etc., and may include data in a single format or multiple formats.

[0048] Specifically, 3D model files may include the character's skeletal structure, surface mesh, texture information, animation data, etc., and can be used to represent a three-dimensional appearance, such as a muscular humanoid warrior. 2D images or illustrations may be used as drawings that represent the character's visual appearance in a two-dimensional manner. Additionally, text information in the form of sentences describing the character's appearance, background settings, personality, etc., may also constitute character information; for example, descriptions such as "a lonely mechanical sage wandering the galaxy, a character with a worn appearance and blue eyes" may be included. Furthermore, scenario documents containing the story or dialogue flow in which the character appears may be included in character information as information for understanding the character's context.

[0049] Here, external feature information refers to specific attribute data that constitutes the visual identity of a character. External feature information is an element extracted from or identified from character information and may include various visual elements that can be expressed quantitatively or qualitatively, such as the character's skeletal structure, body proportions, surface texture, color, facial features, and attachments. For example, external feature information may consist of a mesh form type representing the character's overall body shape or proportions, a texture skin representing surface textures such as skin or armor, facial features including the shape and placement of eyes, nose, and mouth, attachment information such as wings or tails, components of independent physical simulations such as fur or hair, and material attributes including physical characteristics such as light reflectivity or transmittance.

[0050] Here, dynamic expression elements refer to time-based expression elements that impart vitality and individuality to a character. Specifically, dynamic expression elements are visual or auditory changes that occur when a character acts or speaks within video content or a virtual environment, and can serve as core elements in constituting the character's persona. Dynamic expression elements may include, for example, voice including the character's tone, intonation, and speech speed; movements such as gait, gestures, and body movements; components and simulations of independent physical simulations such as fur and hair; changes in facial expressions; and speech patterns or speech styles.

[0052] The user terminal (20) is a device for accessing the character automatic implementation service and interacting with the server (10).

[0053] The user terminal (20) can be configured to allow visual verification of detailed attributes of the character's appearance, adjustment sliders for expression elements, and modification reflection results through a UI linked to the server (10). Additionally, the user can experience the main functions of the character automatic implementation service in real time through the user terminal (20) and perform repetitive feedback and modification work.

[0054] The user terminal (20) may include a communication unit (21), an input unit (22), an output unit (23), a memory (24), and a processor (25).

[0055] The communication unit (21) can communicate with the server (10) or other terminals via wired or wireless means.

[0056] The input unit (22) can receive various information through user operation and input actions. Such an input unit may be a touchscreen module, keyboard, mouse, button, camera, stylus, microphone, etc.

[0057] The user terminal (20) can receive user interaction through the input unit (22). Interaction refers to the user operating the input unit (22) to input information reflecting the user's selection or intention into the user terminal (20). For example, the interaction may be a touch on a touchscreen, a click on a mouse, typing on a keyboard, sound input from a microphone, image capture by a camera, motion recognition by a motion sensor, etc.

[0058] The output unit (23) can output various information. The output unit (23) may be a display device, a speaker, a vibration generating device, a tactile generating device, etc. In some cases, the output unit (23) may be a device (e.g., Bluetooth earphones) that receives and outputs a signal by being connected to a user terminal (20) via wired or wireless communication (e.g., short-range wireless communication such as Bluetooth).

[0059] The memory (24) functions as a storage medium and can store a number of applications running on the user terminal (20), data for the operation of the user terminal (20), and commands. This memory may be provided in hardware in the form of various storage devices such as ROM, RAM, flash drive, hard drive, etc., or in the form of web storage. In one embodiment, an application related to a character automatic implementation service (hereinafter referred to as 'application') may be stored in the memory (24).

[0060] The processor (25) can execute an application by controlling the overall operation of the communication unit (21), input unit (22), output unit (23), and memory (24). At this time, if graphics processing is required, the processor (25) can operate in conjunction with a separate GPU (Graphics Processing Unit) or an SoC (System on Chip) with integrated GPU functionality.

[0062] The generative artificial intelligence model (30) is an artificial intelligence-based model used to generate dynamic expression elements of a character based on data received from the server (10).

[0063] The generative artificial intelligence model (30) is connected to communicate with the server (10), receives external feature information or feature vectors of a character as input, and can output expression data such as voice, movement, facial expression, and tone of voice corresponding thereto.

[0064] Generative artificial intelligence models (30) can be classified into different types depending on the type of expression element. For example, a style-transformation-based Text-to-Speech (TTS) model may be used to generate voice, and a physics-based or data-based motion generation model may be applied to generate movement. In addition, GAN or Transformer-based models for generating facial expressions and speech patterns, or A2F (Audio-to-Face) models that generate facial expressions and mouth shapes from voice input may be utilized.

[0065] The server (10) can select an appropriate generative artificial intelligence model (30) based on the external characteristics information of the character and configure and input a prompt or parameter in a format suitable for the model.

[0066] The generative artificial intelligence model (30) may include a processor (31), memory (32), and a communication unit (33). Each detailed component included in the generative artificial intelligence model (30) performs substantially the same function as each detailed component included in the server (10).

[0068] FIG. 2 is a flowchart illustrating a method for a server to generate dynamic representation elements of a character according to an embodiment of the present invention.

[0069] In step (a), the server (10) obtains character information.

[0070] Specifically, character information can be used as input data for analyzing the external characteristics of a character, and the server (10) can receive various forms of character information through one or more paths.

[0071] According to one embodiment, the server (10) can directly receive and obtain input data containing character information from a user terminal (20). For example, when a user accesses a character automatic implementation service platform and uploads a configuration file or image of a character being created, or inputs a related description as text, the server (10) can receive the data in real time and record it in an internal storage.

[0072] In another embodiment, the server (10) can automatically retrieve or collect character information from a previously registered external content repository or API server. For example, if a user specifies a project folder stored in a cloud-based work environment, the server (10) can search for and automatically obtain character-related data within that folder.

[0073] In another embodiment, the server (10) can automatically extract character information based on logs generated during the content development process, metadata from a version control system (such as Git), or configuration information linked with a scenario writing tool. For example, the server (10) can indirectly collect information regarding the appearance of the character through a character configuration change history recorded during production or a scenario structure containing concept tags.

[0075] In step (b), the server (10) identifies multiple external feature information of the character from the character information.

[0076] Specifically, the server (10) can analyze the character information obtained in step (a) to extract or identify visual attributes constituting the appearance of the character as quantitative or qualitative form of appearance feature information.

[0077] In addition, multiple external feature information may include at least one of a mesh form type, a texture skin, and an independent component.

[0078] Here, the mesh form type refers to shape information that defines the character's overall silhouette and skeletal structure. This includes the character's body shape, body proportions, joint structure, and contour shape, and can be used as a criterion for determining the structural identity of the appearance. For example, mesh form types can be classified into 'humanoid,' 'animaloid,' 'non-geometric,' 'deformed type (5-6 head-to-body ratio characters),' and 'super deformed type (2-3 head-to-body ratio characters),' and each type may include detailed attributes such as body proportions, arm and leg ratios, and head size.

[0079] Here, texture skin refers to visual attribute information that defines the texture, material, color, pattern, etc., that constitute a character's surface. It reflects surface attributes applied to the character's skin, armor, clothing, and equipment, and can be used as an element that visually determines the appearance's atmosphere and concept. For example, texture skins can be classified as 'glossy metallic armor,' 'swamp-like organic skin,' 'aged wood texture,' 'rusted iron plate texture,' and so on.

[0080] Here, components refer to external auxiliary elements that can be individually attached to or operated independently outside the character's basic body, such as fur, hair, horns, wings, and tails. These components may exist independently of mesh form types and texture skins, or be combined to enhance visual characteristics. For example, 'double helical mechanical horns,' 'translucent energy wings,' and 'long mane-shaped aquatic creature hair' contribute to visually revealing each character's individuality and can be utilized as key elements for appearance customization. Components are defined by their position, orientation, size, material properties, etc., and can be applied as a modular design basis for expanding the character's visual identity.

[0081] According to one embodiment, if the character information is in the form of a 3D model, the server (10) can parse the model file to check the skeletal structure, mesh shape, surface mesh density, material composition, texture application information, etc., and organize them into external feature information such as mesh form type, material attribute, texture skin, etc. For example, the server (10) can configure the external features into quantified data in the form of "overall body shape: 2.5 heads tall, head size ratio: 40%, mesh structure: simple curve type."

[0082] In another embodiment, when character information is a text description, the server (10) can analyze keywords and phrases within the sentence using natural language processing (NLP) technology, extract expressions corresponding to visual attributes, and convert them into external feature information. For example, external features such as 'iron material', 'asymmetric structure', and 'light source effect' can be identified from the sentence "The body is made of old iron plates, and one eye glows green."

[0083] In another embodiment, when a 2D image or illustration is included, the server (10) can analyze color distribution, contour shape, facial structure, mounted attachment elements, etc. by utilizing computer vision technology or an image recognition model. For example, external feature information such as 'red metal skin', 'mechanical wings', and 'bizarre eyes' can be identified from the character image.

[0085] In step (c), the server (10) generates dynamic expression elements of a character to be applied to video content containing the character, based on multiple external feature information.

[0086] Specifically, dynamic expression elements are time-based expression elements for expressing the personality and liveliness of a character, and may include at least one of voice, movement, facial expression, and tone of voice. The server (10) can automatically generate dynamic expression elements that match the character's appearance by analyzing external feature information and the correlation between the features.

[0087] According to one embodiment, the server (10) can generate, based on external feature information such as an 'asymmetric mesh structure', a 'rusted metal texture', and 'red light source eyes', a 'low and rough voice mixed with mechanical sounds', a 'limping and heavy gait', and a 'blank expression with emphasized light blinking'.

[0088] In another embodiment, if the server (10) is confirmed to have external features such as a ‘small and round body shape’ and ‘bright pastel-toned skin,’ it can generate ‘high and cheerful speech,’ ‘light and rhythmic movements,’ and ‘expressions that frequently show blinking eyes and smiles.’

[0089] In step (c), the server (10) can generate a first dynamic representation element based on a plurality of external feature information, and generate a second dynamic representation element based on the plurality of external feature information and the first dynamic representation element.

[0090] Specifically, the server (10) can analyze multiple external feature information to first generate a first dynamic expression element that represents the character's personality, atmosphere, impression, etc. The first dynamic expression element may generally be an element that directly reveals the character's identity, for example, voice or tone of voice, and these may serve as criteria that influence the style determination of subsequent expression elements.

[0091] According to one embodiment, the server (10) can analyze the external characteristic information as a character having a 'calm and intellectual atmosphere' and, accordingly, generate a voice with a low tone, slow and clear pronunciation as a first dynamic expression element. Subsequently, the server (10) can generate a movement as a second dynamic expression element based on characteristics such as the rhythm, tone, and speed of the voice. For example, a movement of walking slowly rather than quickly, while keeping the body upright and maintaining stable eye contact, can be output as a second dynamic expression element.

[0092] In another embodiment, the server (10) may determine from the appearance information that it is a ‘slender character with a cunning impression,’ generate a whispering, thin voice as a first dynamic expression element, and then generate a ‘cat-like gait with minimized footsteps’ as a second dynamic expression element, reflecting the tension and secrecy inherent in the voice. At this time, the stride length, center of gravity shift, and head rotation speed of the movement are adjusted to harmonize with the atmosphere of the voice.

[0093] In another embodiment, the first dynamic expression element may be a facial expression. For example, if the server (10) analyzes a face with a 'blank and sharp impression' and generates a 'static expression with almost no emotion,' the tone of voice, which is the second dynamic expression element generated thereafter, may be reflected as 'concise speech consisting mainly of short sentences' and 'a tone with almost no intonation.'

[0094] Alternatively, the dynamic expression element includes a first dynamic expression element and a second dynamic expression element, and in step (c), the server (10) may generate the first dynamic expression element and the second dynamic expression element together so that the first dynamic expression element and the second dynamic expression element correspond to each other.

[0095] Specifically, the server (10) can generate a first dynamic expression element and a second dynamic expression element by considering them simultaneously based on multiple external feature information, and the two elements can be set to have a corresponding relationship to harmonize with each other. This method contributes to forming a unified character by sharing related meanings, emotions, rhythms, etc., rather than each dynamic expression element being generated independently.

[0096] According to one embodiment, the server (10) extracts the characteristic of being a "timid and small-bodied robot" from the external information and, accordingly, generates a "high and trembling voice" as a first dynamic expression element and simultaneously generates a "timid movement of frequently looking around" as a second dynamic expression element. In this case, the voice and the movement are composed of pairs that reflect each other's psychological and behavioral characteristics, and can consistently express the character's anxious tendencies.

[0097] In another embodiment, the server (10) can simultaneously generate a ‘low, hard tone of voice’ and a ‘straight gait with chest puffed out’ for a ‘confident, dignified warrior-type character.’ In this case, the two expression elements are composed of a corresponding pair expressing the same identity of integrity, leadership, and composure.

[0098] In another embodiment, there may be cases where facial expressions and speech patterns are matched. For example, if the external facial features of a character include ‘round eyes and drooping corners of the mouth,’ the server (10) may set a ‘constantly worried expression’ as a first dynamic expression element and simultaneously generate a ‘stuttering speech pattern showing anxiety’ as a second dynamic expression element corresponding to it.

[0099] In addition, dynamic expression elements including simulations of components (fur, hair, etc.) can be generated together. These components are external adjuncts that can be attached to the outside of the character's basic body or operate independently, and can enhance visual identity by being linked to movements or emotional expressions.

[0100] For example, if the server (10) analyzes the character's external characteristic information as an 'aquatic character swimming in water,' it can generate a 'slowly swaying walking motion' as a first dynamic expression element and, as a corresponding second dynamic expression element, generate a simulation in which the character's component 'mane' spreads out like waves. At this time, the mane is set to react organically according to the character's movement speed, direction, and surrounding environmental factors.

[0101] As another example, for a 'combat character equipped with mechanical wings,' a 'strong metallic voice' is generated as the first dynamic expression element and a 'straight and decisive gait' as the second dynamic expression element, while simultaneously, 'mechanical vibrations of the wings' are simulated and visually expressed whenever the character moves or speaks dialogue.

[0103] FIG. 3 is a flowchart illustrating a method for a server (10) to generate dynamic representation elements based on contribution information according to an embodiment of the present invention.

[0104] In step (310), the server (10) checks contribution information for each of the plurality of external feature information related to the creation of dynamic representation elements.

[0105] Specifically, the contribution information is information that quantifies the relative weight of the influence of external feature information on a specific dynamic expression element, and can be used as a criterion for determining how much each feature influences the result. The server (10) can dynamically calculate the contribution of the feature based on the degree to which the external feature information deviates from the standard range.

[0106] According to one embodiment, the server (10) can evaluate how prominent a specific external feature is by comparing it with predefined standard range data and assign a contribution value accordingly. For example, if the standard body shape range is 7 to 8 heads tall and a mesh form type corresponding to 12 heads tall is input, the server (10) may determine that the feature is a major factor dominating the impression of the character and assign a high contribution (e.g., 90%). On the other hand, a feature such as 'blue eyes' extracted together may be determined to provide visual individuality but have a low influence on the overall persona, so a low contribution (e.g., 10%) may be set.

[0107] In another embodiment, the server (10) can estimate the contribution by analyzing the correlation between external features and expression elements that appear in multiple character cases through a learning-based statistical model or an artificial intelligence algorithm. In this case, a high weight may be reflected for features frequently used in similar characters based on data.

[0108] Dynamic expression elements include multiple types, and the server (10) can configure contribution information related to the creation of multiple types of dynamic expression elements differently.

[0109] For example, when generating a 'voice,' external features centered on the upper body, such as facial shape, mouth size, and head proportions, may have a relatively higher contribution. On the other hand, when generating 'movement,' structural characteristics of the entire body, such as leg length, body proportions, and center of gravity, may be composed of having a high contribution.

[0110] Additionally, when generating 'expressions,' detailed facial feature information such as eye shape, pupil size, eyebrow angle, and mouth width is considered an important factor and can have a high contribution, whereas in the case of 'speech tone,' the overall atmosphere or the emotional attributes of the texture often have an influence, so a relatively high contribution may be assigned to texture skin or color contrast information.

[0111] As a specific example, in the case of a ‘robot character with a small, round appearance and made of metal material,’ the mesh form type of this character can be defined as a ‘3-head-tall spherical structure’ and the texture skin as a ‘glossy light gray metallic surface.’

[0112] When the server (10) generates the voice of this character, the external characteristic of being ‘metallic’ is directly connected to the texture of the voice, so a high contribution (e.g., 70%) can be given to that element. On the other hand, ‘round body shape’ has a relatively less direct influence on the characteristics of the voice, so it can have a low contribution (e.g., 20%).

[0113] Conversely, when generating movement, a 'round body shape' can be assigned a relatively high contribution (e.g., 65%) as it closely affects the way the body shakes during movement and the inertia of rotation.

[0114] The contribution information verified by the server (10) can be determined based on the degree to which each of the multiple external feature information deviates from a preset standard range.

[0115] Specifically, the server (10) can evaluate how much each external feature differs from the general standard range by referring to predefined standard data, and determine that the greater the difference, the greater the influence that feature has on the formation of the character's impression, and thus assign a high contribution.

[0116] For example, if the mesh form type defined as the standard range is a humanoid structure of 7 to 8 heads tall, a super deformed form of 3 heads tall is recognized as an appearance that deviates significantly from the standard range, and thus a relatively high contribution (e.g., 85%) may be assigned to the creation of dynamic expression elements. On the other hand, a 'light gray skin color' identified in the same character is judged not to show a significant difference within the standard range, so a low contribution (e.g., 10%) may be set.

[0117] In addition, the standard range may be defined differently for each type of feature. For example, for 'material,' parameter standards such as reflectance, roughness, and luminescence characteristics may exist, while for 'skeletal structure,' a range may be defined based on the number of joints, asymmetry, limb ratios, etc.

[0118] The server (10) can obtain information about contribution information from the user terminal (20).

[0119] For example, if a user wants a ‘deep, low-toned, mysterious voice’ to add an element of surprise to a ‘small, cute fairy’ character, the server (10) can receive adjustment information from the user terminal (20) to lower the contribution to ‘body shape’ and increase the contribution to ‘wing material’. Through this, the server (10) can regenerate dynamic expression elements by reflecting the modified contribution information or modify the existing generated result.

[0120] Additionally, the user terminal (20) can intuitively adjust the contribution to each external feature information through UI elements such as sliders, number inputs, and checkboxes, and the server (10) can update the expression result by reflecting these user inputs in real time.

[0122] In step (320), the server (10) generates dynamic representation elements based on contribution information.

[0123] Specifically, the server (10) can generate dynamic expression elements by applying a calculated contribution value to each of the multiple external feature information as a weight, thereby adjusting the degree to which the corresponding feature is reflected in the final expression result.

[0124] For example, when generating a voice, if the external feature information of ‘metal material’ is set to a high contribution of 70% and the feature of ‘long neck’ is set to a medium contribution of 20%, the server (10) can add metallic noise to the tone according to this contribution (e.g., a hard and resonant tone), the intonation can be a restrained tone, and the speech speed can be slow and heavy.

[0125] When generating a gait, if the ‘heavy body type with a lower center’ has a high contribution of 80%, the server (10) can apply a heavy gait with a low center and dragging feet. Conversely, if the same character has the ‘long legs’ feature with a high contribution, it can be expressed in a form with a wide stride.

[0126] When generating a gesture, if the ‘mechanical arm structure’ has a high contribution, the server (10) can generate a linear and simple mechanical gesture that moves the entire arm rather than moving the fingers finely. For example, the gesture can be designed by extending the gripper of the robot arm instead of the finger when pointing at something.

[0127] When generating movement, if the ‘asymmetrical skeleton’ has a high contribution, the server (10) can generate an unbalanced motion in which the lengths of the arms or legs move differently and the center of gravity is unstable. At this time, abnormal timing is intentionally inserted into the movement to implement a non-biological creature style.

[0128] When generating facial expressions, if 'eye luminescence intensity' or 'facial structure rigidity' contribute significantly, inhuman-like changes in expression may occur, such as only the eye color flashing or emotions being conveyed through subtle mouth vibrations. For instance, when expressing surprise, the mouth or eyebrows remain motionless while the color of the pupils changes temporarily.

[0129] When generating speech patterns, if ‘animal-shaped ears’ and ‘round face shape’ have a high contribution, the server (10) can generate a childlike and cheerful speech pattern, for example, a style that repeats the last word or exaggerates the pitch. On the other hand, if the same character is ‘wearing combat equipment,’ the speech pattern can be generated mainly with short, concise sentences by giving a contribution to that element.

[0131] FIG. 4 is a flowchart illustrating a method for a server to generate dynamic representation elements of a character according to another embodiment of the present invention.

[0132] Step (a) may include a step (410) in which the server (10) obtains basic information in the form of text about the character.

[0133] In step (410), the server (10) can obtain basic information about the character in the form of text.

[0134] Specifically, the basic information may be a text-based description describing the appearance or settings of a character, and may be input through a user terminal (20) or retrieved from an external document. The basic information may be natural language text containing a narrative context, such as, for example, a character setting sheet, a scenario, or a character introduction phrase.

[0135] According to one embodiment, a user may input a sentence such as "a huge gatekeeper with a metallic body guarding ancient ruins." In this case, the server (10) analyzes the sentence through a natural language processing (NLP) module and extracts key keywords (e.g., metallic, huge, gatekeeper) to use in a subsequent step for deriving external feature information.

[0136] In another embodiment, the user may provide basic information structured by item, such as “Genre: SF, Background: Mars, Role: Silent Assassin, Characteristics: Dark Cloak, Red Luminous Eyes”. In this case, the server (10) may configure preliminary information for generating appearance information by distinguishing each item and converting keywords and attribute values ​​into a structured form.

[0137] In another embodiment, the basic information may be metadata of a previously created character, and the server (10) may automatically retrieve the relevant basic information from a database. For example, if there is character setting data stored in a specific user account, the server (10) may automatically retrieve the data and use it as basic information.

[0139] Step (b) may include a step (420) in which the server (10) determines a plurality of external feature information constituting the character based on basic information, and a step (430) in which the character's external feature information is generated based on the plurality of external feature information.

[0140] In step (420), the server (10) can determine multiple external feature information constituting the character based on basic information. The server (10) analyzes the previously acquired basic information to extract core attributes that form the visual identity of the character and converts them into external feature information. In this process, the server (10) can utilize natural language processing techniques and semantic analysis algorithms to identify specific keywords or descriptive content in the input text and identify corresponding visual attributes.

[0141] For example, if a user inputs the text “an alien assassin wearing a dark cloak and with red glowing eyes,” the server (10) extracts information such as “dark cloak,” “red eyes,” “alien,” and “assassin” from this sentence. Based on this, “dark cloak” can be interpreted as a dark-colored fabric texture, “red eyes” as an eye shape with strong luminescence, and “assassin” as a mask-wearing human body structure or a stealth attachment. In this way, the server (10) determines various individual attributes that constitute the appearance of the character and determines the external characteristic information corresponding to each attribute.

[0142] As another example, if a user provides basic information based on items such as “Genre: Fantasy, Role: Forest Spirit Guardian, Characteristics: Transparent wings and green eyes,” the server (10) can derive information on appearance characteristics such as “non-humanoid fairy body type,” “translucent wings attached,” “plant-based texture,” and “bright green eye color” according to each item.

[0143] In step (430), the server (10) can generate appearance information of a character based on multiple appearance feature information. At this time, the appearance information includes 3D model data for concretizing the visual representation of the character, and is output in the form of a single integrated 3D model by combining attributes such as a skeletal structure, surface mesh, texture, color, and material derived from multiple appearance feature information.

[0144] For example, if information regarding external features such as ‘non-humanoid fairy body type’, ‘translucent wings’, ‘plant-based texture’, and ‘green eyes’ is determined, the server (10) can generate information regarding an appearance implemented as a ‘forest spirit character with wings and transparent skin’ by loading pre-built mesh parameters, texture skin data, material attributes, etc. corresponding to each of these and synthesizing them.

[0145] When generating appearance information, the server (10) may include an automatic adjustment function to maintain consistency in mesh composition and material application, going beyond simply combining individual elements. For example, if a 'strong metallic texture' is selected, realistic visual effects can be achieved by adjusting physical attribute values ​​such as light reflectivity and surface roughness together according to the area where the surface is applied.

[0147] In step (c), generating dynamic expression elements can be performed according to step (440) based on multiple external feature information and the basic information.

[0148] In step (440), the server (10) can perform processing to generate dynamic expression elements of a character based on multiple external feature information and basic information.

[0149] The server (10) can derive categories and characteristics of dynamic expression elements to be generated by integrating and analyzing the appearance information generated earlier and the semantic context embedded in the basic information. For example, if the appearance information of a character is ‘muscular body type’, ‘metal armor’, and ‘sharp eyes’, and the basic information is ‘desert warrior’ and ‘blunt and taciturn personality’, the server (10) can synthesize this data to derive dynamic expression elements suitable for the character, such as ‘a voice with a rough masculine tone’, ‘a slow but heavy gait’, ‘an expressionless face with little change in expression’, and ‘a short and neat way of speaking’.

[0151] After step (c), the server (10) may include the step (450) of obtaining modification information for a dynamic representation element from a user terminal (20), the step (460) of determining whether there is a discrepancy between the modified dynamic representation element and a plurality of external feature information, and the step (470) of changing at least some of the plurality of external feature information based on the modified dynamic representation element if a discrepancy is confirmed.

[0152] In step (450), the server (10) can obtain modification information for dynamic representation elements from the user terminal (20).

[0153] Specifically, the user may input a request for modification regarding some or all of the dynamic expression elements, such as voice, movement, facial expression, and tone of voice, which are automatically generated by the server (10). This modification request may be input through a UI provided on the user terminal (20), and may be implemented to allow for detailed adjustment through various interface elements such as sliders, dropdowns, and direct input.

[0154] For example, if a user requests to change the character's default voice to a "slightly lower and rougher tone," the server (10) may receive the request via text input or tone slider operation. As another example, for a character whose default generated gait is a "mechanical straight walk," if a user wants a "slightly leisurely curved walk," a modification request may be entered to change the type of movement or adjust its speed and pattern.

[0155] In step (460), the server (10) can determine whether there is a discrepancy between the modified dynamic representation element and the multiple external feature information.

[0156] Specifically, the server (10) analyzes whether the corresponding dynamic representation element harmonizes with the existing external feature information based on the modification information received from the user terminal (20). This analysis can be performed through a pre-learned association model or rule-based judgment logic. The server (10) evaluates whether a discrepancy or conflict exceeding a specific standard is detected by referring to the mutual association rules or consistency criteria set between the external feature information and the dynamic representation element.

[0157] For example, if a ‘light and high-pitched voice’ is originally generated based on the external features of ‘small and slender body shape’ and ‘transparent wings,’ but the user modifies it to a ‘low and thick-pitched voice,’ the server (10) determines that the voice setting may not match the external features. As another example, if the movement of a character with a ‘slime-like soft outer shell’ is originally generated as a ‘flexible and flowing motion,’ but the user changes it to a ‘mechanical and angular motion,’ the server (10) can detect a discrepancy between the physical texture and the movement.

[0158] In step (470), if a discrepancy is detected, the server (10) may change at least some of the multiple external feature information based on the modified dynamic representation element.

[0159] The server (10) can automatically identify parts of the character's external feature information that do not harmonize with the corresponding dynamic expression elements based on the discrepancy judgment result confirmed in step (460), and adjust or replace those parts to ensure the overall character consistency. At this time, the change can be made automatically based on a pre-set consistency mapping table or the recommendation result of a trained AI model, and in some cases, it can be implemented by suggesting the change to the user and obtaining approval.

[0160] For example, if the user modifies the voice to a ‘low and deep tone,’ the server (10) can change the existing external feature of a ‘slender body type’ to a ‘broad shoulders and medium build’ mesh form type to harmonize with the voice, or adjust the external feature to strongly emphasize the facial contour or jaw shape as a ‘visual point matching the intensity of the voice.’ As another example, if the user modifies the movement to a ‘mechanical angular motion,’ the server (10) can change the texture skin of a ‘soft jelly texture’ among the external information to a ‘metallic outer shell,’ or automatically add attachment information so that a mechanical structure is visually exposed at the joint area.

[0162] FIG. 5 is a table for explaining a configuration in which a server contributes differently to a plurality of external feature information according to the type of dynamic expression element, according to an embodiment of the present invention.

[0163] Referring to Fig. 5, the table quantifies the contribution (%) of each of the external feature information to each dynamic expression element and specifically illustrates how the actual generated result changes according to the contribution value.

[0164] In the case of the first external characteristic information, body type (e.g., a massive 12-head body), it has varying degrees of influence on voice, movement, and facial expressions. In particular, a massive body type contributes significantly to voice (85%) and movement (95%), resulting in the automatic generation of a 'low and heavy-toned voice' and 'slow and weighty gait.' On the other hand, it shows a low contribution to facial expressions (20%), demonstrating that the influence of body type is less than that of facial structure.

[0165] The second characteristic, material properties (e.g., metal skin), affects various expressive elements such as voice (70%), movement (50%), and gestures (30%), and the physical properties unique to the material are reflected as the mechanical nature of the movement or the metallic nature of the tone. For example, a metal skin induces linear movements where the joints move stiffly, and a metallic friction sound is mixed into the tone.

[0166] The third characteristic, facial contour (e.g., an angular jawline), influences facial expressions (85%) and speech patterns (30%). A sharp, angular face shape reinforces the decisiveness of facial expressions and can reflect a neat and strong tone in speech.

[0167] The fourth feature, a curve-centered silhouette (e.g., round shape, round eyes), contributes to speech (40%) and movement (70%). A curve-centered silhouette is an appearance in which a large number of curves are visually identifiable, and as a result, it can contribute significantly to generating smooth and flexible movements. Additionally, this appearance can influence speech to a certain degree by reflecting the character's emotional atmosphere. For example, a round face shape or round joint structure induces flexible and flowing walking or small twitching gestures, and a calm and friendly tone can be naturally expressed in speech. The contribution information listed in the table may be automatically determined by the server (10), but it can also be adjusted directly by the user. For example, if the user wants a 'soft and warm voice despite the character having a metal shell,' the user can manually lower the contribution of the material attribute for the voice through the user terminal, and this can be reflected in a way that causes the system to readjust the consistency between the appearance and the expression.

[0169] FIG. 6 is a table for explaining the process of a server generating dynamic representation elements in conjunction with each other according to an embodiment of the present invention.

[0170] Referring to FIG. 6, the table specifically presents a method in which a first dynamic expression element and a second dynamic expression element are linked together according to external feature information. Each case is classified into sequential linkage or corresponding linkage between dynamic expression elements, and explains how the server (10) operates to strengthen the identity and expression consistency of the character.

[0171] In the first example, based on physical characteristic information such as a "large, muscular physique, strong jawline, and red eyes," a "heavy, low voice" is generated first, followed by a "slow, heavy gait" that matches the impression of that voice. This is a sequential linkage method in which the result of the first element—voice generation—is reflected in the second element—movement generation—thereby naturally expressing a strong and intimidating character.

[0172] The second example is also an example of a sequential linkage method, where an introverted and cautious disposition is inferred from the information 'small and thin-looking body, hunched shoulders,' and accordingly, a 'low, whispering voice' is generated first, followed by a 'cautious and slow gait.' In this way, the character's anxious personality is highlighted as the voice and movement are generated while maintaining psychological consistency.

[0173] The third example corresponds to the reactive linkage method. The characteristics of a "smooth metal exterior and exposed joints" entail physical constraints along with visual attributes. Accordingly, "metallic mechanical sounds" and "rigid, linear movements" are considered and configured simultaneously; rather than being generated independently, they are created together by considering each other, thereby reinforcing the cold and mechanical character.

[0174] The fourth case also belongs to the responsive linkage method. The characteristics of ‘round and flexible appearance, large eyes, and short arms’ imply cuteness and richness of emotional expression, and accordingly, ‘high and fast speech’ and ‘short and exaggerated gestures’ are harmoniously generated together. The server (10) can effectively create a humorous and lively atmosphere by setting voice and gestures complementarily while considering the overall appearance and emotional elements simultaneously.

[0176] FIG. 7 is a drawing illustrating an example of a screen in which a server generates and provides dynamic expression elements based on a plurality of external feature information of a character, according to an embodiment of the present invention.

[0177] FIG. 7 is a diagram illustrating a UI example in which, according to an embodiment of a character automatic implementation service, a server generates dynamic expression elements based on multiple external feature information of a character and visually provides them to a user. The diagram is configured so that a user can input character information, check the results of the external information analysis, and edit and control the generated dynamic expression elements.

[0178] The 'Character Information Input' area performs the function of providing character information to the server by allowing the user to directly upload character images or enter text descriptions. This input information is used as input data to extract multiple external feature details and is implemented in a form capable of accommodating text-based or visual-based information.

[0179] The 'Character Appearance Information' area is an area that visually summarizes and provides appearance feature information automatically analyzed by the server from character information. Referring to Fig. 7, key visual elements of the character, such as body shape (e.g., SD style, 2.2 heads tall, large head and short limbs), material (e.g., rusty metal skin, rough surface), facial structure (e.g., exaggerated mechanical eyes, grid-like mouth structure), and joints (e.g., exposed multi-stage rotating joints), are presented in a structured form.

[0180] The 'Contribution by Appearance Element' section quantifies and visually represents the influence of each appearance feature information on individual dynamic expression elements (e.g., voice, movement, facial expressions, etc.). Referring to Figure 7, the contributions of voice (70%), movement (90%), and facial expressions (10%) are presented for the 'body shape' among the appearance information. This illustrates that a single piece of appearance information can contribute differently to multiple expression elements, intuitively demonstrating the correlation structure between appearance and expression. These contributions can be automatically calculated by the server or designed to be determined dynamically based on the degree of deviation from a standard range. For example, if a specific body shape or joint structure deviates significantly from general standards, that item is evaluated as having a high influence on movement or voice, and is assigned a high contribution. Users can adjust these values ​​directly or follow the system's recommended values.

[0181] The 'Voice Editing Area' provides the ability to fine-tune the voice, which is one of the generated dynamic expression elements. In this area, system-automated initial values ​​are provided for items such as timbre, speed, and speech tone; these values ​​are set based on the character's appearance information and the results of the contribution analysis for each element.

[0182] Users can manually adjust these elements or combine various generated timbres to create a new voice. For example, if a character's appearance has both a mechanical nature and a friendly atmosphere, users can create a hybrid timbre by setting 'rough mechanical voice' to 30% and 'soft neutral voice' to 70% to mix the two.

[0183] The 'Select Generation Method' area is a switch interface that allows selecting the method of linkage between dynamic expression elements. Referring to FIG. 7, one can determine how the expression elements of a character will be combined by selecting either sequential linkage or corresponding linkage. If sequential linkage is selected, for example, a gait or facial expression that matches the timbre is subsequently determined after the timbre is generated, and if corresponding linkage is selected, the voice, gestures, facial expressions, etc., can be generated together to harmonize simultaneously.

[0184] Additionally, the server (10) can perform combination between final outputs based on generated dynamic expression elements. Specifically, the server (10) can combine multiple final outputs, each generated independently, to create a new single output. This combination can be performed between outputs that share the same type of dynamic expression elements, and the nature of the result can be adjusted by applying the reflection ratio of each output as a weight during the combination process.

[0185] For example, the user can select multiple outputs that are ultimately produced through the 'Generate' function of FIG. 7 and directly set the contribution ratio between each output. If the user selects existing output A and output B and specifies a ratio of 30% to 70%, the server (10) generates a final result by combining the two outputs according to the ratio. In another embodiment, the server (10) may receive multiple final outputs as input, automatically calculate multiple predetermined ratio combinations (e.g., A:40%-B:60%, A:50%-B:50%, A:20%-B:80%), and present candidate results generated for each combination to the user.

[0187] The technical features disclosed in each embodiment of the present invention are not limited to that embodiment only, and as long as they are not mutually incompatible, the technical features disclosed in each embodiment may be combined and applied to different embodiments.

[0188] Therefore, in each embodiment, the technical features are described primarily, but as long as the technical features are not mutually incompatible, they may be combined and applied together.

[0189] The present invention is not limited to the embodiments described above and the attached drawings, and various modifications and variations may be possible from the perspective of those skilled in the art to which the present invention belongs. Accordingly, the scope of the present invention should be defined not only by the claims of this specification but also by equivalents thereof. Explanation of the symbols

[0190] 10: Server 20: User terminal 30: Generative AI Models

Claims

Claim 1 A method for generating dynamic expression elements of a character, performed by a server, comprising: (a) a step of acquiring character information; (b) a step of identifying a plurality of external feature information of the character from the character information; and (c) a step of generating dynamic expression elements of the character to be applied to video content including the character based on the plurality of external feature information, wherein step (a) comprises a step of acquiring basic information in the form of text about the character, and step (b) comprises a step of determining a plurality of external feature information constituting the character based on the basic information; and a step of generating external information of the character based on the plurality of external feature information. Claim 2 A method for generating a dynamic expression element of a character according to claim 1, wherein the dynamic expression element comprises at least one of a voice, movement, facial expression, and tone of voice. Claim 3 A method for generating dynamic representation elements of a character, wherein, in claim 1, the plurality of external feature information comprises at least one of a mesh form type and a texture skin. Claim 4 A method for generating a dynamic expression element of a character according to claim 1, wherein generating the dynamic expression element in step (c) comprises: a step of confirming contribution information for each of the plurality of external feature information related to the generation of the dynamic expression element; and a step of generating the dynamic expression element based on the contribution information. Claim 5 A method for generating dynamic expression elements of a character, wherein, in paragraph 4, the dynamic expression elements include a plurality of types, and contribution information related to the generation of the plurality of types of dynamic expression elements is configured differently from each other. Claim 6 A method for generating dynamic expression elements of a character, wherein the contribution information is determined according to the degree to which each of the plurality of external feature information deviates from a preset standard range. Claim 7 A method for generating dynamic representation elements of a character, wherein the step of verifying the contribution in paragraph 4 includes the step of obtaining information regarding the contribution information from a user terminal. Claim 8 A method for generating dynamic expression elements of a character, wherein, in claim 1, step (c) comprises: generating a first dynamic expression element based on the plurality of external feature information; and generating a second dynamic expression element based on the plurality of external feature information and the first dynamic expression element. Claim 9 A method for generating a dynamic expression element of a character, wherein, in claim 1, the dynamic expression element includes a first dynamic expression element and a second dynamic expression element, and generating the first dynamic expression element and the second dynamic expression element in step (c) includes the step of generating the first dynamic expression element and the second dynamic expression element together so as to correspond to each other. Claim 10 A method for generating dynamic expression elements of a character, wherein, in claim 1, step (c) comprises: generating a feature vector for the character based on the plurality of external feature information; and generating the dynamic expression elements based on the feature vector. Claim 11 A method for generating dynamic expression elements of a character, wherein, in claim 1, step (c) comprises: selecting a generative artificial intelligence model according to the type of dynamic expression element; and generating a prompt to be input to the selected generative artificial intelligence model based on the plurality of external feature information. Claim 12 delete Claim 13 A method for generating dynamic expression elements of a character, wherein, in step (c) above, generating the dynamic expression elements is performed based on the plurality of external feature information and the basic information. Claim 14 A method for generating a dynamic expression element of a character, wherein, after step (c), the method further comprises: a step of obtaining modification information for the dynamic expression element from a user terminal; a step of determining whether there is a discrepancy between the modified dynamic expression element and the plurality of external feature information; and, if the discrepancy is confirmed, a step of changing at least some of the plurality of external feature information based on the modified dynamic expression element. Claim 15 A server comprising: a memory; and at least one processor connected to communicate with said memory, wherein the processor is configured to acquire character information; acquire basic information in the form of text about said character; identify a plurality of external feature information of said character from said character information; determine a plurality of external feature information constituting said character based on said basic information; generate external information of said character based on said plurality of external feature information; and generate dynamic expression elements of said character to be applied to video content containing said character based on said plurality of external feature information. Claim 16 A computer program stored on a storage medium in combination with hardware to execute the method of any one of claims 1 to 11 or claims 13 to 14.

Images