System for generating advertisement image from text data by using large language model and immediately using same for promotion, and method therefor

Abstract

A method for generating an advertisement image from text data may comprise the steps of: acquiring first text data including advertisement text from a terminal device of an advertiser; extracting a plurality of associative keywords by inputting, into a large language model, a first prompt generated on the basis of the first text data; on the basis of the plurality of associative keywords, extracting associative images corresponding to the plurality of keywords; calculating an importance score of each keyword on the basis of response information of the advertiser for each of the plurality of associative keywords and the associative images; on the basis of the importance score of each keyword, generating an advertisement image including the associative images; and displaying the generated advertisement image.

Description

A system and method for generating advertising videos from text data using a large language model and immediately utilizing the videos for promotion.

[0001] The present invention relates to a system and method for generating advertising videos from text data using a large-scale language model and immediately utilizing the same for promotional purposes. Specifically, the present invention relates to a method for analyzing and parsing advertising information obtained from advertisers using a large-scale language model and producing a video based on the analysis and parsing of the information. Furthermore, the present invention distinguishes between visible areas where important elements within the generated advertising video should be exposed and invisible areas where they should not be, thereby generating an advertising video optimized for the playback environment, thereby enabling immediate promotional use.

[0002] Traditionally, commercials are produced through multiple stages, including planning, filming, and editing. The advertiser communicates the purpose and message of the ad, and the production team develops the concept based on this. The final video is then completed through filming and editing. This process is time-consuming and expensive, and it often struggles to quickly reflect the advertiser's needs.

[0003] Recently, various attempts have been made to improve the efficiency of the advertising production process. In particular, with the advancement of artificial intelligence technology, methods for automatically generating advertising videos by analyzing text data are gaining attention. This offers the potential to shorten advertising production time and reduce costs.

[0004] Existing automated advertising video generation technologies often fail to adequately reflect advertiser needs or suffer from low-quality generated videos. For example, methods that rely on simple keyword matching for image selection or fixed templates have limitations in producing personalized, high-quality advertising videos. Furthermore, videos generated without considering the playback environment are unlikely to achieve optimal advertising effectiveness.

[0005] A related prior art document is Republic of Korea Patent Publication No. 10-2380972 (March 31, 2022).

[0006] The purpose of the present invention is to propose a system that generates advertising images from text data using a large language model and immediately utilizes them for promotion.

[0007] In order to solve the above-described problem, the method for generating an advertising image according to embodiments may include the steps of: obtaining first text data including an advertising text from an advertiser's terminal device; inputting a first prompt generated based on the first text data into a large language model to extract a plurality of associative keywords; extracting associative images corresponding to the plurality of keywords based on the plurality of associative keywords and the scored result; calculating an importance score of each keyword based on the advertiser's response information for each of the plurality of associative keywords and the associative images; generating an advertising image including the associative images based on the importance score of each keyword; and displaying the generated advertising image.

[0008] Furthermore, the step of calculating the importance score may be characterized by setting the priority of each keyword based on the advertiser's response information, and the step of generating the advertising video may be characterized by determining the exposure time, exposure size, and exposure position of the associative image corresponding to each keyword based on the importance score of each keyword.

[0009] Furthermore, the advertising video according to the embodiments may include a plurality of segments, each segment meaning a partial image corresponding to one keyword, and the step of generating the advertising video may be characterized by generating partial images for the plurality of keywords and joining the generated partial images based on the priority.

[0010] In addition, the step of generating an advertising image according to embodiments may display the associative images in a visible area within the advertising image based on a display environment of a display device on which the advertising image is displayed, and the visible area may indicate an area of ​​the advertising image exposed by the display device.

[0011] Additionally, when the advertising image according to the embodiments is displayed by a plurality of LEDs attached to one or more rotating wings, the visible area may represent an area that is further away from the center point of the advertising image than a first distance and less away from the center point than a second distance.

[0012] With this configuration, the system according to the embodiments can help generate advertising videos as intended by the advertiser and maximize the effectiveness and expected performance of the advertising videos.

[0013] In this way, the advertising video generation system can train a large-scale language model based on learning data and utilize planning information and feedback provided by advertisers to generate optimal advertising videos. Through these components and their interactions, the advertising video generation system according to the embodiments can efficiently generate high-quality advertising videos.

[0014] The advertising video generation system according to the embodiments can generate an advertising video by considering the environment in which the advertising video is played due to this configuration, and helps to clearly convey important information to the viewer by considering the position and size of the visible area.

[0015] The advertising video generation system according to the embodiments defines a visible area and an invisible area, and generates an advertising video by considering the visible area and the invisible area, thereby generating an advertising video optimized for a playback environment and providing the effect of being able to use the generated advertising video as a promotional video suitable for a display environment immediately after generation.

[0016] Figure 1 is a drawing showing the overall configuration and operation of an advertising video generation system according to embodiments.

[0017] FIG. 2 is a diagram illustrating a method of creating a prompt for video generation and generating a video based on a conversation history between an advertiser and a large language model, according to embodiments of the advertising video generation system.

[0018] FIG. 3 illustrates a method by which a system according to embodiments generates an advertising video by taking into account the environment in which the advertising video is played.

[0019] Figure 4 is a flowchart illustrating an example of a method for generating an advertising video according to embodiments.

[0020] Figure 5 is an example of a configuration diagram of a system or server according to embodiments.

[0021] The present invention is susceptible to various modifications and embodiments. Specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, but rather to encompass all modifications, equivalents, and alternatives falling within the spirit and technical scope of the present invention. Throughout the description of each drawing, similar reference numerals have been used to designate similar components.

[0022] Terms such as first, second, A, and B may be used to describe various components, but the components should not be limited by the terms. The terms are used solely to distinguish one component from another. For example, without departing from the scope of the present invention, the first component could be referred to as the second component, and similarly, the second component could also be referred to as the first component. The term "and / or" includes any combination of multiple related items described or any item among multiple related items described.

[0023] When a component is referred to as being "connected" or "connected" to another component, it should be understood that it may be directly connected or connected to that other component, but that there may be other components intervening. Conversely, when a component is referred to as being "directly connected" or "connected" to another component, it should be understood that there are no other components intervening.

[0024] The terminology used in this application is only used to describe specific embodiments and is not intended to limit the present invention. The singular expression includes the plural expression unless the context clearly indicates otherwise. In this application, it should be understood that the terms "comprise" or "have" indicate the presence of a feature, number, step, operation, component, part, or combination thereof described in the specification, but do not exclude in advance the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts, or combinations thereof.

[0025] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. Terms defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and shall not be interpreted in an idealized or overly formal sense unless explicitly defined herein.

[0026] The present invention relates to a system (hereinafter referred to as a "system according to embodiments") that generates an image intended by an advertiser based on text data received from the advertiser (or a conversation history between the advertiser and a large language model). The system according to the embodiments generates an advertisement image by considering the environment in which the advertiser intends to advertise, the environment in which the advertisement image is played, the location of the advertisement, etc., such that important information is included in the important portions of the advertisement image, and includes a process of generating the advertisement image and immediately utilizing the advertisement image for promotional purposes.

[0027] With this configuration, the system according to the embodiments can help generate advertising videos as intended by the advertiser and maximize the effectiveness and expected performance of the advertising videos.

[0028] Below, the overall flow, specific configuration, and flowchart of the system according to the embodiments are described in detail.

[0029] Figure 1 is a drawing showing the overall configuration and operation of an advertising video generation system according to embodiments.

[0030] Specifically, FIG. 1 is a diagram illustrating the configuration and operational relationship of the overall system including an advertising image generation device (10) according to embodiments. Referring to FIG. 1, the advertising image generation device (10) communicates with an advertiser's terminal device (11) and obtains information about the image intended by the advertiser from the advertiser.

[0031] An advertising video generation device (10) obtains planning information for an advertising video from an advertiser's terminal device (11). The advertising video generation device (10) analyzes the obtained planning information based on a large language model, generates an advertising video based on the analysis results, and transmits the generated video to the advertiser's terminal device (11) or to a display unit (12) on which the advertising video is to be displayed.

[0032] Referring to FIG. 1, the advertising video generation device (10) includes a data acquisition unit (100), an associative keyword extraction unit (101), an importance calculation unit (102), a masking data generation unit (103), an image generation unit (104), a prompt processing unit (105), a database unit (106), and a communication unit (107).

[0033] The data acquisition unit (100) acquires planning information (e.g., information about the advertising video intended by the advertiser) from the advertiser's terminal device (11). The planning information may, for example, include all information about the advertising video intended by the advertiser. The planning information may, for example, include a conversation history (context information) between the advertiser's terminal device (110) and a large language model in the form of a chatbot within the advertising video generation device (which may be a large language model provided within the video generation unit (104)).

[0034] The associative keyword extraction unit (101) can extract multiple keywords from the planning information acquired from the data acquisition unit (100). The keywords may be, for example, keywords deemed important in the planning information, keywords that must be included in an advertising video, or keywords that represent images corresponding to the keywords. The associative keyword extraction unit (101) may include a large-scale language model or artificial intelligence model for extracting multiple keywords.

[0035] The importance calculation unit (102) can calculate the importance for each keyword extracted by the associative keyword extraction unit (101). The importance may refer to a numerical and normalized score indicating the degree of relevance that a keyword occupies in the planning information.

[0036] Meanwhile, the advertising image generation device (10) according to the embodiments needs to consider an area where important elements within the advertising image should be exposed (e.g., a visible area according to the embodiments) and an area where important elements should not be exposed (e.g., an invisible area according to the embodiments) in order to immediately reproduce the advertising image within a display environment. Accordingly, the advertising image generation device (10) according to the embodiments may further include a masking data generation unit (103) to generate an advertising image optimized for the reproduction environment and to immediately utilize it for promotional purposes.

[0037] The masking data generation unit (103) can determine the areas that are importantly exposed in the advertisement video or the visible areas that are exposed in the advertisement video when the advertisement video is displayed by the display unit (12). That is, the masking data generation unit (103) determines the visible areas and calculates the non-visible areas (i.e., the areas that are not exposed by the display environment of the display unit (12) in the advertisement video). The masking data generation unit (103) controls the image generation unit (104) or the prompt processing unit (105) so that the image generation unit (104) does not include important information (such as keywords or images / videos corresponding to keywords) in the non-visible areas.

[0038] The image generation unit (104) receives keywords extracted by the associative keyword extraction unit (101), importance scores for each keyword calculated by the importance calculation unit (102), and visible areas (or invisible areas) generated by the masking data generation unit (103) to generate an advertising image. The image generation unit (104) is equipped with a large language model that has learned (or fine-tuned) learning data including a plurality of advertising images and a plurality of keywords for each image.

[0039] That is, the large language model provided by the image generation unit (104) may be a language model trained (fine-tuned) based on predetermined learning data. The predetermined learning data includes an advertisement video, multiple keywords for the advertisement video, an importance score for each keyword, and a pre-labeled important region (which can be set by the advertiser) within the advertisement video.

[0040] Furthermore, the large language model provided by the video generation unit (104) may be a language model that has been additionally prompt-engineered for advertising video generation. For example, the large language model provided by the video generation unit (104) may be a large language model instructed to generate an advertising video such that images or videos corresponding to keywords sorted in descending order of importance score are sequentially displayed.

[0041] The image generation unit (104) includes a large language model that has been prompted to perform various advertising image generation tasks in advance. The image generation unit (104) receives keywords extracted by the associative keyword extraction unit (101), importance scores of each keyword calculated by the importance calculation unit (102), and information on visible areas (or invisible areas) generated by the masking data generation unit (103) to generate a final advertising image. Therefore, the LLM dynamically adjusts the exposure time, exposure size, exposure location, etc. of keywords according to the importance scores of each keyword to compose an advertising image.

[0042] For example, the video generation unit (104) may generate an advertisement video by placing keywords or images / videos related to keywords with high importance scores at the beginning of the advertisement video or in key locations (which may be set by the advertiser) to attract the viewer's attention. Furthermore, the video generation unit (104) may generate an advertisement video by placing important phrases or images in the visible area and placing less important elements in the non-visible area, taking into account the playback environment of the advertisement video.

[0043] The video generation unit (104) can quickly reflect the advertiser's requirements through this prompt engineered LLM and efficiently generate high-quality advertising videos.

[0044] The prompt processing unit (105) generates prompts to be used in the image generation unit (104) and optimizes them. The prompt processing unit (105) can generate one or more prompts for performing additional prompt engineering on a pre-learned (fine-tuned) and prompt-engineered LLM.

[0045] Specifically, the prompt processing unit (105) generates a prompt that is converted into a directive by summarizing important matters intended by the advertiser (e.g., whether a specific phrase is required to be included, the concept of the advertising video, a request to require the inclusion of a specific image / video, the length of the advertising video, the overall color of the advertising video, etc.) within the data acquired by the data acquisition unit (100). The prompt processing unit (105) transmits the converted and generated prompt to the video generation unit (104). With this configuration, the system according to the embodiments helps the LLM to produce optimal results when generating an advertising video.

[0046] The database unit (106) stores and manages various data required for creating advertising videos. The database unit (106) may include advertising planning information, associative keywords, importance scores, image data, learning data, etc. The database unit (106) may store images corresponding to the extracted multiple keywords, or store assets for deriving images corresponding to the multiple keywords.

[0047] The communication unit (107) is responsible for data communication between the advertising image generation device (10), the advertiser's terminal device (11), and the display unit (12). Through this, the advertising image generation device (10) can exchange necessary data in real time and transmit the generated advertising image.

[0048] The advertising video generation device (10) quickly reflects the advertiser's requirements due to this configuration and efficiently generates and provides high-quality advertising videos.

[0049] FIG. 2 is a diagram illustrating a method of creating a prompt for video generation and generating a video based on a conversation history between an advertiser and a large language model, according to embodiments of the advertising video generation system.

[0050] The large language model (LLM) provided in the image generation unit (104) of Fig. 1 can be performed through 20 to 23 steps. The large language model can be performed by the learning unit (not shown in Fig. 1) of the advertising image generation device.

[0051] First, referring to FIG. 2, the training data (20) includes data used by the advertising video generation system to train a large language model. The training data (20) includes advertising videos for training the LLM, a plurality of keywords (label information) corresponding to each advertising video, an importance score for the keywords (label information), and information on visible and invisible areas (label information - not shown).

[0052] Afterwards, the learning unit can formally parse the learning data through learning data parsing (21) and prepare it for LLM learning. In this step, the learning unit analyzes the structure of the learning data, extracts necessary information, and converts it into structured data.

[0053] Next, the learning unit can perform a data cleansing and structuring (22) process, which involves filtering out unnecessary or erroneous data from the parsed learning data, leaving only the necessary data, and creating refined data. This process allows the learning unit to maintain data consistency and organize the data in a state optimized for learning.

[0054] Afterwards, the learning unit can perform fine-tuning (23) of the large language model or the first prompt engineering stage. The learning unit can train a large language model based on refined training data or fine-tune an existing model. This stage learns the patterns and knowledge necessary to optimize the LLM for the advertising video generation task, and the first prompt engineering stage is also performed.

[0055] A large language model (LLM) prepared based on the above-described method can be loaded into an image generation unit (104) according to embodiments. Thereafter, the advertising image generation device (10) according to embodiments can perform the following process through the image generation unit (104) and the prompt processing unit (105).

[0056] First, the advertising video generation device (10) can obtain planning information (24a). The planning information (24a) includes planning information for an advertising video provided by an advertiser.

[0057] If the advertiser provides a certain feedback for the video generated by the video generation unit (104), the video generation unit (104) can further obtain feedback information (24b) including the feedback. The feedback information (24b) refers to feedback provided by the advertiser after the advertisement video is generated. The advertisement video generation device (10) can generate a prompt including the feedback (24) based on the feedback information (24b), and perform secondary prompt engineering of a large language model (25) based on the generated prompt. The video generation unit (104) can modify or newly produce (26) the advertisement video based on this. That is, the prompting (24) step is a process of converting the planning information provided by the advertiser into a command format to create a prompt that the LLM can understand. This step can be performed in the prompt processing unit (105) of FIG. 1. In addition, the language model secondary prompt engineering (25) step is a process of performing additional prompt engineering necessary for the LLM to generate the advertisement video based on the prompted information. Finally, the advertising video generation (26) step is where LLM finally generates the advertising video. In this step, the video generation unit (104) according to the embodiments generates and outputs the advertising video as intended by the advertiser based on all information and prompts prepared in the previous step.

[0058] In this way, the advertising video generation system can train a large-scale language model based on learning data and utilize planning information and feedback provided by advertisers to generate optimal advertising videos. Through these components and their interactions, the advertising video generation system according to the embodiments can efficiently generate high-quality advertising videos.

[0059] FIG. 3 illustrates a method by which a system according to embodiments generates an advertising video by taking into account the environment in which the advertising video is played.

[0060] FIG. 3(A) and FIG. 3(B) are diagrams illustrating 'visible areas' according to embodiments within generated advertising images (300a, 300b). FIG. 3(A) illustrates the visible area in a single rotating blade display, and FIG. 3(B) illustrates the visible area in a plurality of rotating blade display devices.

[0061] First, referring to FIG. 3(A), an advertising image (300a) is shown as being displayed by a single wing display (i.e., a device that displays by multiple LEDs attached to one or more rotating wings). The display environment (301a) includes rotating LED wings, and the visible area (302a) is indicated by a gray translucent area in FIG. 3. The visible area (302a) represents an area within the advertising image (300a) that can actually be exposed to the viewer, and the remaining areas may be invisible areas. The advertising image generation device (10) according to embodiments can generate an advertising image by placing an image corresponding to an important phrase or keyword in the planning information, taking into account the visible area (302a). In addition, the advertising image generation device (10) can maximize the effect of the advertising image by placing less important elements in the invisible area or generating a blank image without any content.

[0062] Referring to Fig. 3(B), an advertising image (300b) is shown in a situation where it is displayed by a plurality of wing displays (i.e., a device that displays by a plurality of LEDs attached to one or more rotating wings). Fig. 3(B) shows an environment (301b) displayed by a plurality of wing displays, each display environment including a rotating LED wing. The visible area (302b) within each display environment may be a gray translucent area, and the area where the translucent areas are combined as a whole may be the visible area. In this case, the advertising image generating device (10) according to the embodiments generates an advertising image by considering the visible area (302b) of each display environment. That is, important phrases or corresponding images can be placed by considering the above-mentioned visible area, and less important elements can be placed in the non-visible area, or a blank image with no content can be generated, thereby maximizing the effect of the advertising image.

[0063] Meanwhile, the advertising video generation device according to the embodiments may generate multiple segments, each corresponding to a plurality of keywords. The advertising video generation device may then concatenate the segments based on the importance scores for the multiple keywords in a predetermined order (e.g., descending order). Each segment may represent a partial video corresponding to a single keyword.

[0064] That is, the advertising image generation device (10) according to the embodiments can generate partial images for a plurality of keywords and can combine the generated partial images based on the above priorities.

[0065] As shown in FIG. 3, the advertising image generation device (10) according to the embodiments can display the associative images in a visible area within the advertising image based on the display environment of the display device on which the advertising image is displayed, and at this time, the visible area can be an area exposed by the display device among the advertising images.

[0066] For example, when an advertising image is displayed by a plurality of LEDs attached to one or more rotating wings, as shown in FIG. 3, the visible area may represent an area that is further away from the center point of the advertising image than a first distance and less away from the center point than a second distance.

[0067] When the advertising image is displayed by a plurality of LEDs attached to one or more rotating wings, the visible area may be a circular or donut-shaped area, or a combination of a plurality of circular or donut-shaped areas.

[0068] The advertising video generation system according to the embodiments can generate an advertising video by considering the environment in which the advertising video is played due to this configuration, and helps to clearly convey important information to the viewer by considering the position and size of the visible area.

[0069] Figure 4 is a flowchart illustrating an example of a method for generating an advertising video according to embodiments.

[0070] Specifically, FIG. 4 can be performed by an advertising video generation system according to embodiments (see FIGS. 1 to 3 ). Referring to FIG. 4 , the advertising video generation system according to embodiments can perform some or all of steps 400 to 405 below. The execution process of each step is described below.

[0071] First, the advertising video generation system according to the embodiments may perform a step (400) of acquiring first text data including advertising text from the advertiser's terminal device. In this step, the advertiser provides text data desired to be included in the advertising video, and the advertising video generation system according to the embodiments may collect this.

[0072] Next, the advertising video generation system according to the embodiments may input the first prompt generated based on the first text data into a large language model and perform a step (401) of extracting a plurality of associative keywords. In this step, the advertising video generation system according to the embodiments may analyze the collected text data to derive key keywords to be included in the advertising video.

[0073] Next, the advertising video generation system according to the embodiments may perform a step (402) of extracting associative images corresponding to a plurality of keywords based on the plurality of associative keywords and the scored results. Based on the extracted keywords, images associated with each keyword may be searched and selected from a database.

[0074] Thereafter, the advertising video generation system according to the embodiments may perform a step (403) of calculating an importance score for each keyword based on the advertiser's response information for each of the plurality of associative keywords and the associative images. The advertiser may provide feedback on each keyword and image, and based on this, the advertising video generation system according to the embodiments may evaluate the importance of each keyword and assign a score.

[0075] At this time, the advertising video generation system according to the embodiments can set the priority of each keyword based on the advertiser's response information in this step (403).

[0076] Thereafter, the advertising video generation system according to the embodiments may perform a step (404) of generating an advertising video including the aforementioned associative images based on the importance score of each keyword. In this step, the advertising video generation system according to the embodiments may place images corresponding to keywords with high importance scores in important locations of the advertising video, and the exposure time and size may also be adjusted according to the importance.

[0077] At this time, the advertising video generation system according to the embodiments can determine the exposure time, exposure size, and exposure location of the associative image corresponding to each keyword based on the importance score of each keyword in this step (404).

[0078] Meanwhile, the generated advertising image may include a plurality of segments, and each segment may mean a partial image corresponding to one keyword. The system according to the embodiments may generate partial images for a plurality of keywords in this step (404) and may stitch the generated partial images based on the priority. In addition, the system according to the embodiments may display the associative images in a visible area within the advertising image based on the display environment of the display device on which the advertising image is displayed, wherein the visible area may be an area of ​​the advertising image exposed by the display device.

[0079] For example, when an advertising image is displayed by a plurality of LEDs attached to one or more rotating wings, as shown in FIG. 3, the visible area may represent an area that is further away from the center point of the advertising image than a first distance and less away from the center point than a second distance.

[0080] Finally, the advertising video generation system according to the embodiments may perform a step (405) of displaying the generated advertising video. The advertising video generation system according to the embodiments may control the generated advertising video to be played on the advertiser's terminal device or display device.

[0081] With this configuration, the system according to the embodiments can help generate advertising videos as intended by the advertiser and maximize the effectiveness and expected performance of the advertising videos.

[0082] In this way, the advertising video generation system can train a large-scale language model based on learning data and utilize planning information and feedback provided by advertisers to generate optimal advertising videos. Through these components and their interactions, the advertising video generation system according to the embodiments can efficiently generate high-quality advertising videos.

[0083] The advertising video generation system according to the embodiments can generate an advertising video by considering the environment in which the advertising video is played due to this configuration, and helps to clearly convey important information to the viewer by considering the position and size of the visible area.

[0084] Figure 5 is an example of a configuration diagram of a system or server according to embodiments.

[0085] Referring to FIG. 5, the server (500) includes an input unit (510), an output unit (520), a control unit (530), a storage unit (540), and a communication unit (550).

[0086] The input unit (510) receives commands or information from an administrator. The input unit (510) may include one or more of a microphone and a key input unit for receiving audio signals.

[0087] The output unit (520) outputs command processing results or various information to the administrator. For example, the output unit (520) generates an advertising image from text data using a large language model and outputs it. For this purpose, the output unit (520) may include a display, a speaker, a haptic output unit (520), and an optical output unit (520), although not illustrated in the drawing. The display may be provided in the form of a flat panel display, a flexible display, an opaque display, a transparent display, electronic paper (E-paper), or any other form well known in the art to which the present invention pertains. A touch pad may be laminated on the display to form a touch screen, and a touch key may be implemented through this touch screen. In addition to the display and the speaker, the output unit (520) may also be configured to further include any form of output means well known in the art to which the present invention pertains.

[0088] The control unit (530) connects and controls components within the server (500). For example, it generates an advertising image from text data using a large language model and controls each component so that it can be output through the output unit (520). As another example, when judgment information is input by an administrator, the control unit (530) generates a response signal including the judgment information. The control unit (530) may be configured to include a CPU (Central Processing Unit), an MPU (Micro Processor Unit), an MCU (Micro Controller Unit), a GPU (Graphics Processing Unit), or any other type of processor well known in the art of the present invention.

[0089] The storage unit (540) stores data, programs, applications, etc. required for the server (500) to operate. The storage unit (540) may include non-volatile memory, volatile memory, a hard disk, an optical disk, a magneto-optical disk, or any type of computer-readable recording medium well known in the art to which the present invention pertains.

[0090] The communication unit (550) generates an advertising image from text data using a large language model via a wired or wireless network and communicates with a system or other system that uses the image for immediate promotion.

[0091] The embodiments of the present invention disclosed in this specification and drawings are intended only to provide specific examples to facilitate understanding and easily explain the technical content of the present invention, and are not intended to limit the scope of the present invention. It will be apparent to those skilled in the art that other variations based on the technical concepts of the present invention are possible in addition to the embodiments disclosed herein.

[0092] Although the present invention has been described above with reference to preferred embodiments thereof, it will be understood by those skilled in the art that various modifications and changes may be made to the present invention without departing from the spirit and scope of the present invention as set forth in the claims below.

Claims

1. A method for generating an advertising video from text data, A step of obtaining first text data including an advertisement text from an advertiser's terminal device; A step of inputting a first prompt generated based on the first text data into a large language model to extract multiple associative keywords; A step of extracting associative images corresponding to the plurality of keywords based on the plurality of associative keywords; A step of calculating an importance score of each keyword based on the advertiser's response information for each of the plurality of associative keywords and the associative images; A step of generating an advertising video including the above associative images based on the importance score of each keyword; and A method for generating an advertising image, comprising: a step of displaying the generated advertising image; 2. In paragraph 1, The step of calculating the above importance score sets the priority of each keyword based on the advertiser's response information, A method for generating an advertising video, characterized in that the step of generating the advertising video determines the exposure time, exposure size, and exposure position of an associative image corresponding to each keyword based on the importance score of each keyword.

3. In paragraph 1, The above advertising video includes multiple segments, and each segment means a partial video corresponding to one keyword. A method for generating an advertising video, characterized in that the step of generating the advertising video generates partial videos for the plurality of keywords and joins the generated partial videos based on the priority.

4. In paragraph 1, The step of generating the above advertising image displays the associative images in a visible area within the advertising image based on the display environment of the display device on which the advertising image is displayed, A method for generating an advertising image, wherein the above visible area indicates an area of ​​the advertising image exposed by a display device.

5. In paragraph 4, A method for generating an advertising image, wherein the advertising image is displayed by a plurality of LEDs attached to one or more rotating wings, and the visible area represents an area that is further away from the center point of the advertising image by a first distance and less than a second distance from the center point.

Images