Manga name generation system, manga name generation method, and manga name generation program

Abstract

To provide a system, method and program for generating a manga name, which allows even a beginner to create a manga name by calculating the relevance between information included in a script based on the script of the manga. [Solution] The manga name generation system includes a receiving unit that inputs a script related to the generation of a manga name, a generation unit that uses a generation AI to generate a manga name based on the input script, with the frame layout and text and image materials to be placed within the frames contained within the frame layout defined, and an output unit that outputs the manga name.

Description

[Technical field]

[0001] The present invention relates to a system, method, and program for generating manga names using technology such as generation AI based on the text material of a script. [Background technology]

[0002] A manga name is a rough outline of the background, speech bubbles, dialogue, character expressions or actions and their placement, panel layout, etc., that is drawn before the manga is drawn on manga manuscript paper. In order to create a manga within a set number of pages, manga artists need to rewrite the manga's story, the way the story will unfold, the distribution of dialogue, the placement of illustrations, etc. many times in advance.

[0003] In recent years, artificial intelligence technology has been used in a wide range of fields, and the manga industry in particular has been influenced by AI image generation technology, including generative AI, and AI text generation technology.

[0004] The comic data editing method disclosed in Patent Document 1 can generate comics that include movement as well as comics that include only still images. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] JP 2017-204290 A Summary of the Invention [Problem to be solved by the invention]

[0006] However, the manga generation system described in Patent Document 1 does not generate manga names, does not use generation AI or other technologies, and may be difficult to operate for users who have no experience creating manga. In addition, the learning model described in Patent Document 1 cannot automatically generate manga images based on text information contained in a script.

[0007] Therefore, the present invention has been made in consideration of the above problems, and has an object to provide a manga name generation system that can generate manga names simply by inputting a script. [Means for solving the problem]

[0008] One embodiment of the present invention includes a receiving unit that inputs a script for generating a manga name, a generation unit that uses a generation AI to generate a manga name based on the input script, with the frame layout and text and image materials to be placed within the frames contained within the frame layout defined, and an output unit that outputs the manga name.

[0009] The above-mentioned manga name generation system may further include a classification unit that classifies text material included in a script into one or more pieces of dialogue information and one or more pieces of non-dialogue information, a calculation unit that calculates a first information relevance indicating the degree of relevance between the one or more pieces of dialogue information, and an arrangement unit that arranges the dialogue information and the non-dialogue information classified by the classification unit within a frame, wherein the calculation unit calculates a first information relevance indicating the degree of relevance between one or more pieces of dialogue information that are consecutive in the script based on the order in which the dialogue information appears according to the chronological order of the script, and the arrangement unit arranges the consecutive pieces of dialogue information within the frame as text material if the first information relevance is higher than a predetermined threshold value. When the first information relevance is lower than a predetermined threshold, the placement unit may separate consecutive pieces of dialogue information and place them in the frames as text material. The calculation unit may calculate a second information relevance indicating the degree of relevance between dialogue information and information other than dialogue information that is consecutive in the script in accordance with the chronological order of the script, and the placement unit may place the information other than dialogue information within a frame located in the dialogue information if the second information relevance is higher than a predetermined threshold value. The calculation unit calculates a second information relevance indicating the degree of relevance between consecutive dialogue information and information other than dialogue in the script in accordance with the chronological order of the script, and if the second information relevance is lower than a predetermined threshold, the placement unit may place at least one of stage directions, onomatopoeia, scene descriptions, and background descriptions contained in the information other than dialogue within the frame. The generation unit may further include an image material generation unit that generates image material based on information other than the dialogue in the frame, and the image material generation unit may generate character material based on information indicating characteristics of the character from the information other than the dialogue, thereby placing the character material in the frame as image material in accordance with the dialogue information. The image material generation unit may use morphological analysis to break down information other than dialogue contained in a frame, excluding character characteristics, and use each broken down word as a keyword to generate images required for the manga name as image material, and the placement unit may place the image material in the frame. The generating unit may generate a manga name by rearranging the frames according to a predetermined rule in accordance with the chronological order of the script and in accordance with paragraphs in which dialogue information and non-dialogue information are located in the script. The output unit may output the manga name in an editable format based on an input from the user.

[0010] This is a manga name generation method in which a computer executes an input step of inputting a script for generating a manga name, a generation step of using a generation AI to generate a manga name based on the input script, in which the frame layout and the text and image materials to be placed in the frames contained within the frame layout are defined, and an output step of outputting the manga name.

[0011] This is a manga name generation program that executes an input function for inputting a script for generating manga names into a computer, a generation function for using a generation AI to generate a manga name with predetermined frame layouts and text and image materials to be placed within the frames contained within the frame layout based on the input script, and an output function for outputting the manga name. Effect of the Invention

[0012] According to a manga name generation system according to one embodiment of the present invention, a manga name can be generated and provided by inputting a script using a generation unit. [Brief description of the drawings]

[0013] [Figure 1] 1 is a diagram illustrating a schematic configuration of an embodiment of the present invention. [Diagram 2] 1 is a block diagram showing an example of the internal configuration of a manga name generation system according to an embodiment of the present invention. [Diagram 3] FIG. 11 is a flowchart showing an example of a process for arranging information contained in a script in a frame. [Figure 4] FIG. 11 is a diagram illustrating an example of a process for classifying information included in a script. [Diagram 5] FIG. 13 is a diagram showing an example of an image in which information contained in a script is arranged in frames. [Figure 6] FIG. 11 is a flowchart showing an example of processing relating to frame size editing and rearrangement. [Figure 7] 13A to 13C are diagrams showing examples of processing results relating to frame size editing and rearrangement. [Figure 8] FIG. 13 is a diagram showing an example of image generation based on information to be arranged in a frame. [Figure 9] FIG. 13 is a diagram showing an example of generating a manga name based on information within each frame. [Figure 10] FIG. 13 is a flowchart showing an example of a user re-editing a manga name. [Figure 11] FIG. 2 is a diagram showing an example of an interface of the manga name generation system. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0014] Hereinafter, an embodiment of the comic name generation system 100 will be described with reference to the drawings. Note that the comic name generation system 100 is not limited to the embodiment. Also, some components that are not important for the description will be omitted in the drawings.

[0015] Furthermore, the script according to this embodiment is not limited to headings, scene change instructions, characters, main text, cover, stage directions, and dialogue, but may also include information such as story titles, scene introductions, stage introductions, production effects, production orders, and psychological descriptions of characters (hereinafter referred to as text materials). As an example, in a certain script, character descriptions of characters appearing in the script are written first, and the story contained in the script is divided into scenes for each chapter or story. Furthermore, each scene may include the dialogue of the characters in the scene, background descriptions, psychological descriptions, titles, production orders, etc.

[0016] <Summary> 1 is a schematic diagram illustrating a configuration of one embodiment of the present invention. In this embodiment, a manga name generating system 100 is communicably connected to a user terminal 200 via the Internet NW. A user prepares in advance manga name generating materials including various text materials (for example, script information including at least one of information for describing characters, information for describing scenery, information for describing the character's psychology, sound effects, etc., hereinafter referred to as script information) in the user terminal 200, and inputs the materials to the manga name generating system 100 via the Internet NW to generate a manga name.

[0017] The manga name generation system 100 according to this embodiment can generate a manga name based on script information received from a user terminal 200. The manga name generation system 100 can also convert the manga name into a form that can be edited by the user, and transmit the obtained data to the user terminal 200. After the user re-edits the manga name on the user terminal 200, the edited manga name can be re-input into the manga name generation system 100, which then outputs the edited manga name.

[0018] The manga name generation system 100 can also generate a final draft of the manga from the edited manga name. The final draft of the manga according to this embodiment may be richer in image material and text material than the manga name. For example, if the image material "galaxy" is included in the manga name, the manga name generation system 100 may re-edit this image material "galaxy" to make it a real "galaxy", and add beautiful stars, sparkling gaseous interstellar matter, etc. to the image material "galaxy" in the manga name. In other words, the manga name generation system 100 may generate supplementary material according to the text material and image material included in the manga name and add it to the manga name to generate the final draft of the manga.

[0019] The user terminal 200 according to this embodiment is not limited to functions such as uploading information input by the user and downloading the manga name generated by the manga name generation system 100, and the user can also re-edit the manga name generated by the manga name generation system 100 on the user terminal 200. Furthermore, the embodiment of the user terminal 200 is not limited to a personal computer, and may be a mobile terminal such as a smartphone (a so-called mobile phone) or a tablet.

[0020] The editing in this embodiment may be divided into a plurality of types depending on the editing target. When the user is about to edit the output image material, the user may adjust the position, size, composition, character design, color, etc. of the image material on the user terminal 200. When the user is about to edit the output text material, the user may adjust the position, size, font, character, color, etc. of the image material on the user terminal 200.

[0021] <Configuration> Next, the internal structure of the comic name generation system 100 will be described with reference to FIG. 2. The comic name generation system 100 according to this embodiment includes a receiving unit 110, a classification unit 121, a calculation unit 122, an arrangement unit 123, a generation unit 124 including an image material generation unit 125, a storage unit 126, and an output unit 130. The comic name generation system 100 is typically a computer system, and functions are realized by a CPU (Central Processing Unit), a GPU (Central Processing Unit), and the like. Note that FIG. 2 is only an example, and the configuration of the comic name generation system 100 is not limited to that shown in the figure. Each functional unit of the comic name generation system 100 processes information to automatically generate a comic name based on script information received from a user terminal 200.

[0022] The receiving unit 110 can receive script information prepared in advance from the user terminal 200. Furthermore, the script information received from the user terminal 200 may be directly passed to the intermediate processing unit 120, or may be stored in the storage unit 126 as information for generating a manga name.

[0023] The intermediate processing unit 120 may include a classification unit 121, a calculation unit 122, a placement unit 123, a generation unit 124, an image material generation unit 125, and a storage unit 126. The intermediate processing unit 120 can generate a manga name based on the script information received from the user terminal 200. This will be described in detail below.

[0024] The classification unit 121 can classify the script received from the user terminal 200 into dialogue information and non-dialogue information according to the characteristics of the text material. The method of classifying script information may be divided into three. In the first method, the classification unit 121 may recognize and classify information within symbols that normally indicate dialogue in a script (for example, quotation marks, double quotation marks, brackets, double brackets, parentheses, corner brackets, square brackets, angle brackets, etc.) as dialogue information. In this case, the remaining information may be classified as non-dialogue information. In addition, when a character's psychological description is included in a script, the symbols indicating the psychological description are normally displayed separately from dialogue, so the manga name generation system 100 may identify the symbols indicating the psychological description and extract the information within the symbols. After that, the psychological description may be classified as dialogue information or non-dialogue information according to a pre-setting of the manga name generation system 100. For example, when psychological description is set as a type of dialogue information, if information indicating dialogue is shown in quotation marks (") and information indicating psychological description is shown in parentheses (()), the classification unit 121 does not determine the characteristics of the information enclosed in quotation marks and parentheses, and after detecting the quotation marks (") and parentheses (()), it may automatically treat the information contained within the parentheses as dialogue information, and treat the part not enclosed in parentheses as information other than dialogue.

[0025] Secondly, the classification unit 121 may classify the script information received from the user terminal 200 by machine learning. Specifically, the manga name generation system 100 may use dialogue information and non-dialogue information in a script prepared in advance as learning data, and may load the information into a learning model installed in the manga name generation system 100. In this case, when a script is input to the manga name generation system 100, the manga name generation system 100 may classify information in parentheses and information not in parentheses in the script as features based on the learning results, thereby separating the information into dialogue information and non-dialogue information. In addition, the learning model may use whether or not information is in parentheses as learning data. The features of the information in this embodiment may include, but are not limited to, the presence or absence of a line break between each piece of information (for example, a line break at the switch between dialogue information and non-dialogue information), the use of symbols, and the like.

[0026] Third, the classification unit 121 may classify information included in the script into spoken dialogue information and written information. Written language (written language) in this embodiment may be language often used when writing sentences or scripts, and spoken language (spoken language) in this embodiment may be language often used when directly talking with another person. As an example, if the script received from the user terminal 200 includes "My husband stopped by a convenience store on his way home from work, but he didn't seem to buy anything" and "My husband stopped by a convenience store on his way home from work, but he didn't buy anything," the classification unit 121 may detect "mono no" and "yoda datta" from "My husband stopped by a convenience store on his way home from work, but he didn't seem to buy anything" and then compare it with the data on written language stored in the storage unit 126. When a keyword included in a sentence matches the written language data stored in the storage unit 126, the classification unit 121 can treat "My husband stopped by a convenience store on his way home from work, but he didn't seem to buy anything" as information other than lines. Conversely, since "but" and "desu yo" are generally used in colloquial speech, the classification unit 121 may detect "but" and "desu yo" from "My husband stopped by a convenience store on his way home from work, but he didn't buy anything" and then compare it with the data related to colloquial speech stored in the storage unit 126. When a keyword included in a sentence matches the written language data stored in the storage unit 126, the classification unit 121 can treat "My husband stopped by a convenience store on his way home from work, but he didn't buy anything" as line information.

[0027] The calculation unit 122 can calculate a first information relevance indicating the information relevance between pieces of dialogue information, and a second information relevance indicating the information relevance between dialogue information and information other than dialogue. Furthermore, the objects for which the first information relevance and the second information relevance are calculated are those that are chronologically continuous in the script information (i.e., the order in which the information appears in the script). Details will be described below.

[0028] As an example, the calculation unit 122 can calculate the degree of relevance between information included in each paragraph by calculating the first information relevance and the second information relevance between each part by morphological analysis. As an example, the calculation unit 122 divides information included in the script by a period, an exclamation mark, a question mark, or the like, and sets the end of a sentence. Then, the calculation unit 122 may treat consecutive sentences according to the order of appearance of the lines as one set, and may decompose each sentence into nouns, verbs, particles, auxiliary verbs, conjugated forms, or the like by morphological analysis. As an example, when sentences that are consecutive in chronological order in the script are sentences A, B, and C, the calculation unit 122 may treat sentences A and B as one set, and sentences B and C as another set. After that, the calculation unit 122 may treat one set as one unit, extract morphemes (i.e., the smallest unit having meaning in language) of two sentences included in the set, and calculate the degree of relevance by detecting the presence or absence of related keywords. If the degree of relevance is higher than a predetermined threshold, it can be determined that the degree of relevance between the preceding and following sentences is relatively high. Conversely, if the value is lower than the predetermined threshold, it can be determined that the relevance to the surrounding sentences is relatively low.

[0029] The method of setting the threshold for determining the presence or absence of relevance is not limited, and the relevance between contexts may be compared based on the presence or absence of a characteristic word in a divided context. As an example, between sentence A "Hey, you! Are you a subordinate of Yamata no Orochi?" and sentence B "What is Yamata no Orochi?", "Yamata no Orochi" appears as a noun in both conversations, so it may be determined that these two conversations are related. As another example, between sentence C "I want to be a subordinate of Yamata no Orochi!" and sentence D "I'm sleeping all the time," the characteristic words are different, so it may be determined that the relevance between the contexts is low.

[0030] After the calculation unit 122 calculates a first information relevance indicating the relevance between pieces of dialogue information and a second information relevance indicating the relevance between the dialogue information and information other than dialogue, the arrangement unit 123 may arrange each text material in a frame generated by the generation unit 124 based on the result of comparing the first information relevance and the second information relevance with a predetermined threshold. If the first information relevance and the second information relevance are higher than the predetermined threshold, the relevance of the consecutive sentences (i.e., sentences as a set) is high, so they may be arranged in one frame. Conversely, if the second information relevance is lower than the predetermined threshold, the relevance of the consecutive sentences is low, so the set may be divided and the text materials may be arranged in separate frames. The generation of frames and frame allocation will be described below.

[0031] The frames according to this embodiment refer to each frame in a manga, and the frame layout may refer to the layout of the frames. The generation unit 124 can generate a manga name in which the frame layout, text material to be placed in the frame layout, and image material are determined based on the script received from the user terminal 200. The generation unit 124 may also include an image material generation unit 125 that generates an image suitable for the text material using a generation AI technology. The generation AI technology according to this embodiment may be a normal one.

[0032] In this embodiment, the generating unit 124 may generate a frame template of a predetermined size according to the size of manuscript paper recommended for a publisher, web page, application, etc. when submitting a manga. In this case, the generating unit 124 may generate a frame layout by dividing the frame into four or six equal parts according to the frame of the predetermined size, or each divided frame may be completely identical and the same size, but is not limited to these. The generating unit 124 can divide the frame template generated according to the manga genre and generate a four-frame or six-frame manga frame layout. The generating unit 124 may generate a space between frames (hereinafter referred to as a margin) by changing its appearance according to the manuscript paper size, but the generating unit 124 can also enlarge or reduce the margin according to an editing instruction from the user.

[0033] The image generating unit 125 is equipped with an image generating AI, and can generate image data to be used in a comic based on text material included in each frame and information included in a script. The generating AI is an AI that can automatically generate an image by inputting text indicating the content of the target to be generated. The image generating unit 125 may generate an image data by inputting text material that describes a character appearing in a comic included in a script to the image generating AI, for example. The image generating unit 125 may also accept input of image data of an image data generated by a user, and generate an image data according to the content of the text material based on the image data. As an example, when image data of a "quiet residential area" is accepted as image data, and a sentence "People are rushing home in the rain with umbrellas" is input as text material, an image of a person holding an umbrella walking in a residential area in the rain may be generated.

[0034] The storage unit 126 is a storage medium that stores various information and programs required in the comic name generation system 100. The storage unit 126 may realize its functions using, for example, a hard disc drive (HDD), a solid state drive (SSD), a flash memory, etc., but is not limited to these, and may include a flash memory, a random access memory (RAM), a read only memory (ROM), etc.

[0035] The storage unit 126 can store various information that can be used in the manga name generation system 100. For example, each information relevance calculated by the calculation unit 122, the manga frames generated by the generation unit 124, the AI ​​image materials generated by the image material generation unit 125, the manga names generated by the generation unit 124, etc. may be stored, but are not limited to these, and the storage unit 126 may store what is obtained after the user re-edits the manga name on the user terminal 200.

[0036] The output unit 130 can output a manga name with a set specification in which text material and image material are arranged in a frame. In addition, what is output to the output unit 130 is not limited to this, and it may output text material, image material, a manga frame having text material and image material, a manga frame having only text material before generating image material, etc.

[0037] The output unit 130 may output the final draft of the manga. After the manga name generation system 100 generates the manga name and transmits it to the user terminal 200, the user may input the manga name to an image generation AI installed in the manga name generation system 100 or to an externally connected artificial intelligence application software. Next, the manga name generation system 100 may generate supplementary materials based on the materials included in the manga name and add them to the manga name to generate the final draft of the manga. As an example, when the manga name generation system 100 generates a manga name including non-dialogue information 1 "The wind is blowing strongly" and non-dialogue information 2 "A person is walking in heavy snow", the user may re-edit the template, specifications, etc. of the manga name on the user terminal 200, and then input the manga name including the non-dialogue information 1 "The wind is blowing strongly" and non-dialogue information 2 "A person is walking in heavy snow" to an image generation AI installed in the manga name generation system 100 or to an externally connected artificial intelligence application software. Here, the image generation AI installed in the manga name generation system 100 or artificial intelligence application software connected externally may detect non-dialogue information 1 "The wind is blowing strongly" and non-dialogue information 2 "A person is walking in heavy snow" contained in the edited manga name, and generate image material related to non-dialogue information 1 "The wind is blowing strongly" (for example, the leaves are swaying while the trees are swaying and making a sound) and image material related to non-dialogue information 2 "A person is walking in heavy snow" (for example, footprints left on the snow-covered ground) and add them to the manga name as supplementary material, thereby generating the final draft of the manga.

[0038] <Examples of operation and display screen> Next, an embodiment of the comic name generation system 100 will be described with reference to a flow chart and sample drawings, etc. Fig. 3 shows an example of an operation for classifying information contained in a script into dialogue information and non-dialogue information and arranging the information in frames after receiving the script from the user terminal 200, but the classification method and arrangement procedure are not limited to this.

[0039] First, the comic name generation system 100 receives script information from the user terminal 200 (S101). In this embodiment, the information is not limited to script information, and may include short stories, full-length novels, essays, etc. Furthermore, the information included in the script information may be text material. After that, the classification unit 121 classifies the information into dialogue information and non-dialogue information according to the characteristics of the information based on the script received from the user terminal 200 (S102).

[0040] Thereafter, the calculation unit 122 calculates a first information relevance between pieces of dialogue information that are consecutive in chronological order in the script, and a second information relevance between the dialogue information and information other than dialogue, and proceeds to the next step (S103). Finally, the arrangement unit 123 arranges the text material within the frame based on the first information relevance and the second information relevance (S104). After the arrangement unit 123 has arranged the text material within the frame, the operation of a portion of the intermediate processing unit 120 ends.

[0041] In addition, since there is a possibility that non-dialogue information and non-dialogue information may appear consecutively for several lines in a script in response to a chronological order, in this case, the calculation unit 122 may separate each piece of non-dialogue information with a period, an exclamation mark, a question mark, or the like, and calculate a third information relevance indicating a relevance between pieces of non-dialogue information that are consecutive in a chronological order. The calculation method of the third information relevance is the same as the first information relevance and the second information relevance, and the arrangement unit 123 may determine whether each text material related to the non-dialogue information is arranged in the same frame or in different frames based on the third information relevance. As an example, when non-dialogue information X, non-dialogue information Y, and non-dialogue information Z appear consecutively in a script, the calculation unit 122 may calculate a third information relevance between the non-dialogue information X and the non-dialogue information Y, and between the non-dialogue information Y and the non-dialogue information Z. When the third information relevance between the non-dialogue information X and the non-dialogue information Y is higher than a predetermined threshold value and the third information relevance between the non-dialogue information Y and the non-dialogue information Z is lower than a predetermined threshold value, the arrangement unit 123 may arrange the non-dialogue information X and the non-dialogue information Y in the same frame and arrange the non-dialogue information Z in a single frame. In addition, the calculation unit 122 may calculate the second information relevance with the preceding and following sentences by collecting several lines of non-dialogue information into one sentence (i.e., an information set of non-dialogue information). As an example, when the script contains not only the non-dialogue information X, the non-dialogue information Y, and the non-dialogue information Z, but also the dialogue information O before the non-dialogue information X, the calculation unit 122 may directly treat the non-dialogue information X, the non-dialogue information Y, and the non-dialogue information Z in the script that continues for several lines as an information set having information relevance. In this case, the calculation unit 122 may compare the second information relevance between the information set and the dialogue information O with the predetermined threshold value by calculating the second information relevance. The placement unit 123 may place each piece of information within a frame based on the result of the comparison.

[0042] On the other hand, the method of calculating each information relevance is not limited to the above, and the calculation unit 122 may divide the script into scenes and calculate the information relevance between the information described in each scene as one unit. As an example, a script may include scenes 1, 2, and 3, and the calculation unit 122 may detect a keyword "scene" and then divide the script into "scene 1," "scene 2," and "scene 3." After calculating the information relevance between the information included in each divided scene, the arrangement unit 123 may arrange the information included in each scene in a frame according to the information relevance between the information included in each scene. In this case, since the information relevance between the scenes is relatively high (i.e., one story exists in one scene), the method of dividing the script into scenes and calculating the information relevance between the information included in each scene is expected to reduce the effort required for calculation, compared to calculating the information relevance for the entire script.

[0043] 4 illustrates an example of classifying information included in a script into dialogue information and information other than dialogue. A script example 400 may generally include at least one of a story title, character descriptions of characters appearing in the story, scenes for developing the story, stage directions, stage information, background information, dialogues, monologues, psychological descriptions between characters, and an ending. After receiving the script example 400 from the user terminal 200, the receiving unit 110 stores the script example 400 in the storage unit 126, and the classification unit 121 may classify information included in the script example 400 into dialogue information example 410 and information other than dialogue information example 420 based on the script example 400 stored in the storage unit 126.

[0044] The dialogue information example 410 may generally include speech of the characters placed in the script, but is not limited to this, and may also include onomatopoeia that imitates the voices or actions of people or animals (for example, "boom" that imitates the sound of an explosion or impact, or "mee-mee" that imitates the sound of a sheep), dialogue that shows a psychological description of the character, etc.

[0045] Furthermore, the non-dialogue information example 420 may include information other than the dialogue information example 410 in the script example 400. In general, the non-dialogue information example 420 may include information including the title of the story included in the script example 400, the clothing, hairstyle, appearance, etc. of the characters appearing in the story, scenes for developing the story, stage directions, setting information, background information, etc.

[0046] The method of determining whether the information in the parentheses is dialogue information or information other than dialogue is not limited. The classification unit 121 may determine whether the information in the parentheses is dialogue information by detecting various types of parentheses (hereinafter referred to as a parentheses detection method), or may determine whether the information is dialogue information by analyzing the context of information included in each scene or information included in different scenes in the script information using morphological analysis. As an example, in the script example 400, a portion showing a character's psychological description is enclosed in parentheses. When the classification unit 121 detects parentheses, the information in the parentheses may be automatically recognized as a character's psychological description and may be a type of dialogue information. When the classification unit 121 detects brackets, the information in the brackets may be the character's remarks or lines. The classification unit 121 is not limited to these, and may determine whether the information is dialogue information by analyzing the context of the sentence in question using morphological analysis (hereinafter referred to as a context analysis method) based on the information included in each scene.

[0047] On the other hand, when the classification unit 121 classifies information included in a script into dialogue information and non-dialogue information, both the parentheses detection method and the context analysis method may classify information about the person who speaks the dialogue into non-dialogue information. As an example, when sentence E is written as "A: (I'm not motivated yesterday and today either.)", the classification unit 121 may classify the part "I'm not motivated yesterday and today either.)" as dialogue information by the parentheses detection method, but since "A:" is not enclosed in parentheses, there is a risk that it will be classified as non-dialogue information. Therefore, after classifying the dialogue information into non-dialogue information, the classification unit 121 may detect whether or not there is a line break in the dialogue information. When there is a line break before and after the part where the dialogue information is located, the whole part where the dialogue information is located may be stored in the storage unit 126. In the above example, "A: (I'm not motivated yesterday and today either.)" may be stored in the storage unit 126 in its entirety.

[0048] 5 shows an example of an operation for arranging dialogue information and non-dialogue information in a frame. The arrangement unit 123 according to this embodiment may classify each frame into a "dialogue frame 510," a "monologue frame 520," a "composite frame 530," and a "depiction frame 540" according to the information arranged in the frame.

[0049] The "dialogue-type frame 510" according to this embodiment is a frame in which the text material arranged therein is a dialogue between two or more characters. When the first information relevance is higher than a predetermined threshold, the arrangement unit 123 may arrange consecutive dialogue information calculated based on the first information relevance as text material in the frame.

[0050] The "monologue-type frame 520" according to the present embodiment is a frame in which the text material arranged in the frame is composed of a monologue of one character or a psychological description showing the true feelings of the character. When the first information relevance is lower than a predetermined threshold, the arrangement unit 123 may separate the consecutive dialogue information for calculating the first information relevance and arrange it in separate frames as text material. As an example, three people are talking to each other in a script, and the calculation unit 122 may calculate the first information relevance between character A and character B and the first information relevance between character B and character C, respectively. Based on the results calculated by the calculation unit 122, when the first information relevance between the dialogue information of character A and character B is higher than a predetermined threshold and the first information relevance between the dialogue information of character B and character C is lower than a predetermined threshold, the dialogue information of character A and character B may be arranged in a frame as text material, and the dialogue information of character C may be arranged alone as text material in another frame.

[0051] The "composite frame 530" according to this embodiment is a frame in which the text material arranged within the frame is composed of the character's dialogue information and information other than dialogue. As an example, in a script, a scene showing a part of a story includes at least one of a scenic description, a psychological description, dialogue, etc., and when the second information relevance between the dialogue information and the information other than dialogue that are consecutive in the script chronologically is higher than a predetermined threshold, the dialogue information and the information other than dialogue may be arranged together as text material within the frame. In addition, when the third information relevance indicating the information relevance between the information other than dialogue is higher than a predetermined threshold, the information other than dialogue that is consecutive in the chronological order may be arranged in the composite frame 530 together with the character's dialogue information.

[0052] A "depiction frame 540" according to this embodiment is a frame in which the text material arranged therein consists only of information other than dialogue. As an example, when the second information relevance between dialogue information and non-dialogue information that are consecutive in chronological order in the script is lower than a predetermined threshold, the dialogue information and non-dialogue information may be arranged as text material within the frame. Also, when the third information relevance indicating the information relevance between information other than dialogue is lower than a predetermined threshold, the non-dialogue information that is consecutive in chronological order may be separated and arranged separately in the depiction frame 540.

[0053] The rules for arranging the text material in a frame are not particularly limited, and an upper limit on the number of characters of the text material contained in each frame may be set to make it easier for the reader to read. Specifically, if the number of characters of the text material contained in a frame is lower than a predetermined upper limit, the text material contained in the frame is easier for the reader to read, and the arrangement unit 123 does not need to rearrange the text material. Conversely, if the number of characters of the text material contained in a frame is higher than the predetermined upper limit, the arrangement unit 123 may rearrange the existing text material by separating it with a period, an exclamation mark, or a question mark, so that the text material contained in the frame is easier for the reader to read.

[0054] As an example, when dialogue information a, dialogue information b, dialogue information c, and non-dialogue information d, which are highly related and chronologically continuous, are arranged in a frame, the calculation unit 122 can first calculate the number of characters included in each of the dialogue information a, dialogue information b, dialogue information c, and non-dialogue information d, while calculating the total number of characters of the information included in the frame. Next, the calculation unit 122 compares the total number of characters of the information arranged in the frame with a predetermined upper limit of the number of characters, and if it is higher than the predetermined upper limit of the number of characters, calculates the difference between the total number of characters of the information arranged in the frame and the predetermined upper limit of the number of characters, and re-compares the difference with the predetermined upper limit of the number of characters. If the difference is lower than the predetermined upper limit of the number of characters, the calculation unit 122 may calculate the number of phrases by using periods, exclamation marks, and question marks for the information in the frame at the current time. In addition, the arrangement unit 123 may equally divide the number of phrases according to the time series and arrange them in two frames, respectively. Conversely, if the difference value is higher than the predetermined upper limit of the number of characters, the difference value may be continuously compared with the predetermined upper limit of the number of characters to calculate a second difference value between the difference value and the predetermined upper limit of the number of characters. The calculation unit 122 may repeatedly calculate the difference value, and if the final difference value is lower than the predetermined upper limit of the number of characters, the arrangement unit 123 may arrange the number of phrases in the "calculation number + 1" frames in chronological order, but is not limited to this.

[0055] FIG. 6 is an example of a flowchart showing the process of adjusting the size of each frame after each line of dialogue information and non-dialogue information are placed within the frame, but the manga name generation system 100 according to this embodiment is not limited to this.

[0056] First, the amount of information contained in each frame is valued based on the number of characters of information arranged in each frame calculated by the calculation unit 122 (S201). Then, the calculated amount of information arranged in the frame is compared with a predetermined threshold (S202), and it is determined whether the amount of information is higher than the predetermined threshold (S203). If the amount of information is higher than the predetermined threshold (YES in S203), the comic name generating system 100 adjusts the size of the frame to be larger so that the text material can fit within the frame (S204). On the other hand, if the amount of information is lower than the predetermined threshold (NO in S203), since the text material and image material are already contained within the frame, the comic name generating system 100 may adjust the size of the frame to be smaller, or may not adjust it as it is (S205). Next, the comic name generating system 100 acquires the frame after the size adjustment and stores it in the storage unit 126 (S206). Then, the manga name generation system 100 rearranges the frames after adjusting the size in chronological order according to the size of the manuscript paper (S207). The method of rearranging the frames according to this embodiment is not limited, and the frames may be rearranged from top to bottom, or in an S-shape from the top right to the bottom left, or may be rearranged according to a predetermined rule. Finally, the image material generation unit 125 causes the image generation AI to learn the dialogue information and non-dialogue information arranged in the frames, and generates an AI image material (S208). The generated AI image material may be inserted into the frame, and the size of the AI ​​image material may be adjusted depending on the size of the frame.

[0057] The comic name generating system 100 can adjust the size of the frames according to a comic expression method set in advance. The comic expression method may be divided into one-panel comic, four-panel comic, one-page comic, and one-shot comic. FIG. 7 shows frames rearranged on a manuscript paper 600 based on information included in a script example 400, using a one-page comic as an example. After the classification unit 121 divides the information described in the script into a plurality of types of frames, the calculation unit 122 calculates the amount of information arranged in each frame, and the comic name generating system 100 can adjust the size of the frames according to the result of comparison between the calculation result and a threshold value. As an example, if the information included in the frame 610 is higher than a predetermined threshold value and the information included in the frame 620 is lower than a predetermined threshold value, the size of the frame 610 may be adjusted to be relatively large, and the size of the frame 620 may be adjusted to be relatively small.

[0058] In addition, the generative AI in this embodiment can learn the patterns and relationships of various data contained in the script and generate possible AI image materials to be used in the manga name. As an example, if the script contains the statement "A...lightning wraps around his body," the generative AI may automatically detect "lightning" and "wrap around his body" in the statement, and then generate an AI image related to "lightning" and an AI image having the state of "wrap around his body," and automatically generate a character whose body is wrapped around lightning. In addition, the image generating AI according to this embodiment may learn the connection between the character material and the corresponding image material as teacher data based on the image generating material (which may include the character material and the corresponding image material) uploaded in advance to the manga name generating system 100. In this case, the more image generating materials uploaded, the higher the accuracy of the generated image. In addition, the method of generating AI image materials may be a normal one.

[0059] Next, the image generating unit 125 inputs the included information to the image generating AI according to this embodiment, and the image generating AI generates AI image material corresponding to each piece of information based on the information arranged in the frame. FIG. 8 shows an example in which the image generating unit 125 generates AI image material based on the information arranged in the frame. When a title 611, a scene 612, a stage 613, and a psychological description 614 are arranged in a frame 610, the image generating unit 125 may generate a title image material 615, a speech bubble material 616, and a person image material 617 and a scene image material 618 from the above descriptions via the image generating AI.

[0060] Furthermore, the image material generating unit 125 may generate different image materials depending on the type of information. As an example, when information indicating a title is input to the image generating AI, the image material generating unit 125 may not change the information included in the title, but may modify the style, size, etc. and place it at a predetermined position within the frame. As another example, when information indicating dialogue is input to the image generating AI, the image material generating unit 125 may not change the dialogue information, but may generate speech bubbles, etc., and place them around the human image.

[0061] 9, the image material generation unit 125 may generate an edit frame 700 based on the information arranged in the frame, and then replace the edit frame 700 with the original frame 610. The manga name generation system 100 can generate a manga name by repeating the above-mentioned process of generating an AI image and the process of replacing frames.

[0062] Next, an example of a process for outputting edited data will be explained using Figure 10. Note that, although the operations in the following flowchart will be mainly performed by the user terminal 200, operations such as data conversion and output of edited data may be executed by each functional section of the manga name generation system 100.

[0063] 10, the comic name generation system 100 converts the comic name into data that can be edited in the user terminal 200 (S301). The editable data according to this embodiment may be data that can be recognized by the user terminal 200, but is not limited to this, and may also be data that can be edited by an external comic homepage, comic app, comic system, etc.

[0064] After the comic name is converted, the output unit 130 of the comic name generating system 100 according to the present embodiment outputs the converted comic name as converted data (S302). After the user checks the converted data on the user terminal 200 (S303), it is determined whether or not the comic name generated by the comic name generating system 100 according to the present embodiment needs to be re-edited (S304). If the converted data needs to be re-edited (YES in S304), the user may adjust information in the converted data on the user terminal 200 (S305). Adjustable information is not specified, and may include adjustment of text material (e.g., may include the content of the text material, the number of characters, character size, etc.), adjustment of image material (e.g., may include the size and position of the image), and adjustment of the position or corresponding information of the speech bubble. After the converted data is edited, it may be input to the comic name generating system 100 as edited data, and the final comic name may be output via the output unit 130 mounted on the comic name generating system 100 (S306). Conversely, if there is no need to re-edit the converted data (NO in S304), the edited data may be directly re-input into the manga name generation system 100 as edited data, and the final manga name may be output via the output unit 130 installed in the manga name generation system 100 (S306).

[0065] FIG. 11 is a diagram showing an example of the interface of the comic name generation system 100. In the interface 800 of the comic name generation system 100, the user presses an input button 810 to upload previously prepared script information to the comic name generation system 100. In this embodiment, the user may input each scene included in the script to the comic name generation system 100. After receiving the input script information, the comic name generation system 100 may reflect the script information in the text display field 840. Then, information such as paragraphs and settings located in the "scene" may be extracted from the script using context analysis or morphological analysis, and the information may be reflected in the location and time display field 820.

[0066] Furthermore, in order to generate a manga name suitable for the script, the user may input supplementary settings for generating a manga name to the manga name generation system 100 via the supplementary settings field 830 on the user terminal 200. In this case, various supplementary settings input by the user may be simultaneously reflected in the production order display field. The supplementary settings according to this embodiment may include information such as character descriptions (e.g., character A is a demon king, character B is a villain, etc.), story development (e.g., character A and character B are in a fierce battle and the weather is getting worse, etc.), etc.

[0067] The comic name generation system 100 may automatically extract information indicating the title using the above-mentioned morphological analysis or context analysis method and place it in the title column 850. Information included in the place and time display column 820 may also be automatically placed in the chapter column 860 of the comic name. In addition, the classification unit 121 installed in the comic name generation system 100 may extract dialogue information from a script input by a user, and then the generation unit 124 may generate speech bubble material 870 suitable for the dialogue information and place it in the comic name. The position of the generated speech bubble material 870 is not limited, and the comic name generation system 100 may detect a character image generated by the image generation AI, and then associate the dialogue information corresponding to the character in the script with the character image generated by the image generation AI to form a set.

[0068] Furthermore, the comic name generation system 100 may automatically place stage directions and supplemental settings included in the script in the information confirmation field 880. In this case, the user can determine whether or not to re-edit while checking the information confirmation field 880 on the user terminal 200.

[0069] The method of generating a manga name from a script is not limited to the method described in the above embodiment. Other methods may be used as long as they can generate a manga name or a manga from a script. As an example, a learning model that has learned the correspondence between a script and an image of a manga name to be generated based on the script may be prepared, and the manga name generation system 100 may be configured to input a new script into the learning model and generate a manga name. Alternatively, a learning model that has learned the correspondence between a script and an image of a manga to be generated based on the script may be prepared, and the manga name generation system 100 may be configured to input a new script into the learning model and generate a manga.

[0070] The comic name generating system 100 may also include a learning unit (not shown) for generating the learning model, and may be configured to generate a learning model for each user. That is, by learning the correspondence between a script created by a user and a comic name or comic generated by the user based on the script, a learning model that outputs a comic name or comic that suits the user's taste can be prepared. The learning unit can generate the learning model by learning using data that associates text data of the script with image data of a comic name or image data of a comic to be generated based on the script as teacher data. The learning unit can use existing algorithms to generate the learning model, and examples include, but are not limited to, neural networks, support vector machines, VAE (Variational Auto Encoder), GAN (Generative Adversarial Network), and CNN (Convolutional Neural Network).

[0071] Furthermore, the learning model may be prepared to realize a part of the process of generating a manga name or manga from a script as described above. For example, a learning model for generating a manga from a generated manga name may be prepared, and the manga name generation system may utilize the learning model. The learning model for generating a manga from a manga name may be a model that has learned the correspondence between image data of the manga name and image data of the manga, and may be a model that outputs image data of the manga by inputting image data of the manga name (the manga name generated by the manga name generation system).

[0072] The manga name generation system may also have a function of accepting input of a script and converting the format of the input script into a format compatible with the manga name generation system if the script is not in a format that the manga name generation system can convert into a manga name, and a learning model may be used to realize this function. In this case, the learning model is a model that has learned the correspondence between a script written in a format that is not compatible with the manga name generation system and a script that has been modified to be compatible with the manga name generation system. By providing such a learning model, the manga name generation system can generate a manga name or manga even from a script written in a format that it cannot support.

[0073] Note that the operations shown in the above flowcharts are merely examples of the operations of the information processing device 1, and do not limit the order of processing. For example, a specific process may be executed in an order or at a timing different from the processing order in the flowcharts.

[0074] Although an embodiment of the present invention has been described above with reference to the drawings, the specific configuration is not limited to this embodiment, and various modifications may be made without departing from the spirit of the present invention.

[0075] <Supplementary Information> Although one embodiment of the comic name generating system 100 has been described, it goes without saying that the invention is not limited to this embodiment. Various modified examples and functions that can be installed will be described below.

[0076] (1) In the above embodiment, the manga name generation system 100 may periodically collect information related to work production from the Internet NW or personal devices to generate a work plan. The work plan may be a plan to consider the possibility of generating a manga before the genre of the manga or a summary of the manga is determined. As an example, the manga name generation system 100 is connected to a large-scale language model LLM (Large Language Model), and information collected from the Internet NW or personal devices may be input to the large-scale language model to learn, and a script may be automatically generated, or a character design may be generated based on the information. In this case, the manga name generation system 100 may generate a manga name through a manga name generation function according to the present embodiment based on the script and character design generated by the large-scale language model. In addition, the large-scale language model LLM may read the production intention of the work specified by the name, and the work may be generated by an image generation AI based on a diffusion model (i.e., a diffusion model).

[0077] (2) In the above embodiment, the manga name generation system 100 may have a function of automatically adjusting the relative positions between image material and text material included in the generated manga name. As an example, when a frame contains multiple materials, the manga name generation system 100 calculates the distance between each material. The manga name generation system 100 may compare the distance between each material and adjust it to be appropriate for the size of the frame. In addition, after receiving a command from a user to adjust the material position, the manga name generation system 100 may adjust the design and overall layout in the manga name according to the user's personal preferences, etc.

[0078] (3) In the above embodiment, the manga name generation system 100 may analyze the input script information by pairing multiple items using a large-scale language model to divide the script. As an example, the manga name generation system 100 may divide the script information using characters and their appearance order, themes and messages, conflicts in events and their resolutions, genres and tones, etc., contained in the script information as dividing points. In this case, since there is a relationship between each divided part, a manga name with a matching style may be generated based on the style of the divided parts.

[0079] (4) In the above embodiment, the manga name generation system 100 may generate a video based on the generated manga. As an example, the manga name generation system 100 may generate a final draft of the manga and then introduce the final draft of the manga to the video generation unit. The video generation unit may generate character actions based on the relevance of each frame in the final draft of the manga and visualize it. [Explanation of symbols]

[0080] 100 Manga Name Generation System 200 User terminals 110 Receiving unit 120 Intermediate Processing Section 121 Classification Department 122 Calculation section 123 Placement section 124 Generation part 125 Image material generation section 130 Output section 400 Script Examples 410 Line information example 420 Examples of non-dialogue information 510 Interactive Top 520 Monologue-type frame 530 Composite Top 540 Descriptive Frame 600 manuscript paper 611 Titles 612 scenes 613 Stage 614 Psychological description 615 Title Image Material 616 Dialogue balloon material 617 People Images 618 Scene Image Material 700 Editing Frames 800 Manga Name Generation System Interface 810 Introduction button 820 Location and time display column 830 Replenishment setting field 840 Text display field 850 Title column 860 Chapter column 870 Speech bubble material 880 Information confirmation column

Claims

1. A receiving unit for inputting a script related to the generation of a manga name; A generation unit that uses a generation AI to generate a manga name in which a frame layout and text materials and image materials to be placed in the frames included in the frame layout are determined based on the input script; a classification unit that classifies the text material included in the screenplay into one or more pieces of dialogue information and one or more pieces of non-dialogue information; a calculation unit that calculates a first information relevance degree indicating a relevance degree between one or more pieces of dialogue information; an arrangement unit that arranges the dialogue information and the non-dialogue information classified by the classification unit within the frame; and an output unit that outputs the manga name, The calculation unit is calculating the first information relevance degree indicating a relevance degree between one or more pieces of consecutive dialogue information in the script, based on an appearance order of the dialogue information in accordance with a chronological order of the script; The placement unit is When the first information relevance is higher than a predetermined threshold, the consecutive pieces of dialogue information are arranged in the frame as the text material. Manga name generation system.

2. The placement unit is If the first information relevance is lower than a predetermined threshold, the consecutive pieces of dialogue information are separated and arranged in the frames as the text material.

2. The comic name generating system according to claim 1 .

3. The calculation unit is calculating a second information relevance degree indicating a relevance degree between the dialogue information and the non-dialogue information, the non-dialogue information being consecutive in the script, in accordance with a chronological order of the script; The placement unit is If the second information relevance is higher than a predetermined threshold, the information other than the dialogue is arranged in the frame located in the dialogue information.

3. The comic name generating system according to claim 2.

4. The calculation unit is calculating the second information relevance degree indicating a relevance degree between the dialogue information and the information other than the dialogue, which are consecutive in the script, in accordance with a chronological order of the script; The placement unit is When the second information relevance is lower than a predetermined threshold, at least one of stage directions, onomatopoeia, scenery description, and background description included in the information other than the dialogue is arranged in the frame.

4. The comic name generating system according to claim 3.

5. The generation unit is an image material generating unit that generates the image material based on information other than the dialogue in the frame, The image material generating unit generates character material based on information indicating characteristics of the character from information other than the dialogue, and places the character material in the frame as the image material in accordance with the dialogue information.

5. The comic name generating system according to claim 4.

6. The image material generating unit decomposing information other than the lines, excluding the characteristics of the characters, contained in the frames by morphological analysis, and using each decomposed word as a keyword to generate an image required for the manga name, thereby forming the image material; The placement unit is Place the image material in the frame 6. The comic name generating system according to claim 5.

7. The generation unit is According to the chronological order of the screenplay, and depending on the paragraphs in which the dialogue information and the non-dialogue information are present in the screenplay, the frames are rearranged according to a predetermined rule to generate the manga name.

7. The comic name generating system according to claim 6.

8. The output unit is The manga name is output in an editable format based on an input from a user.

2. The comic name generating system according to claim 1 .

9. The computer An input step of inputting a script related to the generation of a manga name; A generation step of generating a manga name using a generation AI, in which a frame layout and text material and image material to be placed in the frame included in the frame layout are determined based on the input script; a classification step of classifying the text material included in the screenplay into one or more dialogue information and one or more non-dialogue information; A calculation step of calculating a first information relevance degree indicating a relevance degree between one or more pieces of dialogue information; an arrangement step of arranging the dialogue information and the non-dialogue information classified in the classification step within the frame; An output step of outputting the manga name is performed, The calculation step includes: calculating the first information relevance degree indicating a relevance degree between one or more pieces of consecutive dialogue information in the script, based on an appearance order of the dialogue information in accordance with a chronological order of the script; The placing step includes: When the first information relevance is higher than a predetermined threshold, the consecutive pieces of dialogue information are arranged in the frame as the text material. How to generate manga names.

10. On the computer, An input function for inputting a script related to the generation of a manga name; A generation function that uses a generation AI to generate a manga name in which a frame layout, text material and image material to be placed in the frame included in the frame layout are determined based on the input script; a classification function for classifying the text material included in the screenplay into one or more dialogue information and one or more non-dialogue information; A calculation function for calculating a first information relevance degree indicating a relevance degree between one or more pieces of dialogue information; an arrangement function that arranges the dialogue information and the non-dialogue information classified by the classification function within the frame; An output function for outputting the manga name is executed. The calculation function is calculating the first information relevance degree indicating a relevance degree between one or more pieces of consecutive dialogue information in the script, based on an appearance order of the dialogue information in accordance with a chronological order of the script; The placement function is When the first information relevance is higher than a predetermined threshold, the consecutive pieces of dialogue information are arranged in the frame as the text material. Manga name generation program.

Images