An AI-Assisted Graphic Novel Generation System and Method Based on a Creation Platform
By building an AI-assisted system based on the creation platform, it supports multiple input and generation modes, realizes the integration of text and images and the generation of structured data, solves the problems of complex operation, poor display effect and control of existing tools, meets diverse creative needs, reduces trial and error costs and provides convenience for video conversion.
Patent Information
- Application Number
- CN202611316001.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-27
- Publication Date
- 2026-11-10
AI Technical Summary
Existing AI-assisted creation tools have high operating thresholds, lack a unified platform, poor image and text integration display effects, difficulty for users to control generated content, limited input methods, inability to adapt to diverse creative needs, and isolated image and text generation that cannot form a structured data system.
We have built an AI-assisted system based on the creation platform, which supports multiple input methods and generation modes, and realizes the display of integrated text and images. Users can customize role settings and modify prompt words, and generate structured data in a synchronous manner to provide auxiliary information for video conversion.
It enables the entire creation process to be completed under a unified interface, improves the reading experience and production control, reduces trial and error costs, supports diverse creation needs, and provides a convenient data foundation for video conversion.
Smart Images

Figure FT_1 
Figure FT_2 
Figure FT_3
Abstract
Description
Technical Field
[0001] This invention belongs to the field of generative artificial intelligence (GI), specifically involving text generation based on large language models, image generation based on diffusion models, and image-text fusion display technology. This invention is particularly applicable to online creation platforms for novel writing, digital content production, and image-text reading services. Background Technology
[0002] In recent years, generative artificial intelligence technology has made significant breakthroughs, with large language models and text-based graph models being widely applied in content creation. However, existing technologies still have the following shortcomings: First, current AI-assisted creative tools are mostly locally deployed or single-function software. Users need to build their own environment and combine multiple different tools to complete the entire creative process from text generation to image generation, which has a high operating threshold and a large learning cost. There is a lack of an integrated online creative platform that allows users to complete the entire process of novel idea input, AI generation, text and image layout, and publishing and reading under a unified interface, which seriously restricts the content production efficiency of ordinary creators.
[0003] Secondly, existing AI-assisted writing tools typically only offer plain text generation. Even those that support images often simply piece together text and pictures, lacking the immersive text-image integration effect of "text floating above images," making it difficult to deliver a high-quality text-image reading experience. Furthermore, most tools do not support systematic customization of the generated character's appearance, clothing, and other visual elements, resulting in significant discrepancies between the AI-generated character's appearance and the user's preconceived notions.
[0004] Third, most existing AI content creation tools adopt a "black box" generation mode. Users can only input their initial requirements and cannot view or understand the specific prompts used by the AI to generate the content, let alone modify or adjust them during the generation process. This makes it difficult for users to accurately control the direction and details of the generated content, resulting in a significant deviation between the generated results and user expectations, a poor user experience, and high trial-and-error costs.
[0005] Fourth, in existing technologies, the methods for users to input creative information are relatively limited, typically only supporting the filling in of conceptual elements item by item, which cannot meet the diverse needs of creators such as "importing and optimizing existing complete novel manuscripts" or "precisely controlling the creation process chapter by chapter." At the same time, AI text generation also lacks two optional modes: "generating the entire book at once" and "confirming generation chapter by chapter," making it unable to adapt to different creative habits.
[0006] Fifth, existing technologies for generating images from story text typically aim to produce single or multiple images of characters and scenes, outputting them as image files rather than creating a complete novel for readers to enjoy on a platform. This fails to address the technical challenges of integrating text and images and providing a seamless reading experience on a platform. Furthermore, the information generated during the text-image generation process in existing solutions is isolated and lacks a structured data system, thus failing to provide a convenient foundation for subsequent video conversion.
[0007] To address the aforementioned problems, this invention proposes an AI-assisted graphic novel generation system and method based on a creation platform. Summary of the Invention
[0008] Purpose of the invention This invention aims to provide an AI-assisted graphic novel generation system and method based on a creation platform, the core objectives of which include: First, we will build an integrated online creation platform to provide users with a complete online creation environment, from inputting novel ideas, generating the entire novel / chapter by chapter using AI, displaying and arranging text and images, to reading text and images on the platform.
[0009] Second, it supports multiple input methods and multiple generation modes to meet the needs of different creative habits and creative stages.
[0010] Third, it supports various text and image integration display methods, such as floating and non-floating display, with the final output form being a text and image integrated novel available for readers to read on the platform.
[0011] Fourth, it supports users to customize the appearance of novel characters in detail, generate character visual identity profiles and force their use throughout the entire process, ensuring that the character image remains highly consistent across all images.
[0012] Fifth, it empowers users to view, edit, and modify the prompts used in the AI generation process, enabling "human-machine collaborative creation" and enhancing users' precise control over the generated content.
[0013] Sixth, during the text and image generation process, structured video generation auxiliary information (including but not limited to storyboards, scene descriptions, image prompts, etc.) is generated simultaneously, providing a convenient data foundation for the subsequent video conversion of the novel content and avoiding repetitive work for creators during the video conversion stage. Technical solution
[0014] To achieve the above objectives, the present invention provides the following technical solution:
[0015] Users access this module through the web or mobile interface of the content creation platform and input their content creation requests. This module supports multiple input methods to accommodate different creation habits and stages:
[0016] Users input settings at different levels of novel creation according to the structured templates provided by the platform: (1) The hierarchical structure of the entire novel (overall approach) At this level, users set the overall creative elements of the entire novel, including but not limited to: story theme and core idea, world view and historical background setting, narrative style of the entire novel, overall plot structure, character system setting, and overall visual style.
[0017] (2) Chapter hierarchy setting (chapter ideas) At this level, users can refine the settings for specific chapters to guide the generation of text and illustrations for that chapter, including but not limited to: the core plot of this chapter, the characters appearing in this chapter, the scene environment of this chapter, the key shots / scenes of this chapter, and the narrative rhythm of this chapter.
[0018] (3) Inheritance and Overriding Relationship between Global Settings and Chapter Settings The system automatically applies the "overall novel hierarchy setting" as the global default parameter to the text and image generation of all chapters; the specific settings made by the user in the "chapter hierarchy setting" will take priority over the conflicting parts in the global settings, forming a creative control system with "unified global tone and distinctive chapter personality".
[0019] Users can directly upload complete text files of existing novels (supporting formats such as txt, doc, docx, and pdf). The system automatically recognizes and imports the entire novel's text content. After uploading, the system performs text parsing and chapter structure division of the entire novel, automatically recognizing chapter titles and paragraph structures, and organizing subsequent image generation tasks by chapter. Users can review and revise the content of each chapter online after uploading.
[0020] Users can upload novel text files chapter by chapter, and the system will import the text content of each chapter in the order of upload. This mode is suitable for scenarios where the creator has already written some chapters and needs to import them chapter by chapter and generate illustrations for each chapter.
[0021] Users can combine the above methods.
[0022] Regardless of which input method the user chooses, this module allows for detailed custom descriptions of the physical characteristics of each character in the novel.
[0023] For file uploads, after importing the text, the system uses a large language model to automatically perform semantic analysis on the entire novel, extracting information such as the names of each character, descriptions of their appearance, and descriptions of their clothing from the text, and generating initial character setting suggestions for users to confirm or modify.
[0024] The system automatically converts user-submitted character appearance descriptions or character settings extracted from text into structured character description prompts, which are then fully displayed to the user through the platform interface. Users can view, edit, modify, and optimize the description prompts for any character. After modification, the system regenerates or updates the character's "Visual Identity Profile" based on the user's revised prompts.
[0025] Based on a large language model, the module generates novel text content according to the creative information received from the user input module. This module supports two independent generation modes:
[0026] Based on the user-provided hierarchical structure of the entire novel, the system generates the text of all chapters from beginning to end in one go. Once generated, the system presents all chapters sequentially on the platform, allowing users to browse the entire text and conduct a comprehensive review. Users can revise and polish any chapter, or request the system to partially regenerate specific chapters.
[0027] The system generates text content chapter by chapter based on the overall hierarchy of the novel and the chapter hierarchy of the currently specified chapter. After each chapter is generated, the system displays its content on the platform for user review, revision, and confirmation; once the user confirms the current chapter, the next chapter is generated in the same manner.
[0028] Users can switch between the two generation modes at any time during the creation process. The two generation modes are independent of the user input method and can be freely combined.
[0029] After the novel text is generated, several images matching the story content are generated for each chapter: Storyboard generation unit: Based on the novel text, a large language model is used to parse and generate storyboards; The prompt word construction unit generates prompt words by constructing images based on the storyboard and the user-defined character image, environment setting, and visual style, and incorporates the feature information from the "character visual identity profile" into the prompt words; Prompt display and editing interface: The completed image generates prompts and displays them to the user through the platform interface. The user can view, edit, modify, and optimize the prompts. After the user confirms the modifications, the system regenerates the image using the modified prompts. Image generation unit: Generates high-quality images based on a large model of text-based images; Character Consistency Maintenance Unit: Based on the "Character Visual Identity Profile", ensures that the image of the same character remains highly consistent in different chapters and different pictures; User adjustment interface: Creators can preview, select and modify each generated image, and can specify an image to view its prompts, modify the prompts and regenerate; Video Generation Auxiliary Information Generation Unit: During the storyboard generation, prompt word construction, and image generation processes, the system simultaneously structures and stores data such as storyboards, character settings, scene descriptions, and image prompt words for each chapter, forming an auxiliary information dataset that can be used by video generation tools. This provides a convenient data foundation for the subsequent video conversion of the novel content. User modifications to the prompt words are also recorded and stored in the auxiliary information dataset.
[0030] This module is responsible for integrating and displaying AI-generated novel text with corresponding chapter images on the platform. All presentation is completed within the platform's interface. Assign the text content of each chapter to the associated scene images by paragraph; Supports multiple display methods for combining text and images:
[0031] The text is displayed as a floating overlay on top of the image, creating a stacked display relationship with the text in front and the image behind.
[0032] It should be noted that, in this invention, "text floating above an image" means that the display layer containing the text content is higher in display hierarchy than the display layer containing the image, so that the text content is at least partially displayed above the image's display area. This "floating" display does not require the text background to be completely opaque or completely obscure the image; the text background can be set to any of the following: opaque, semi-transparent, or completely transparent. Regardless of the text background transparency setting, as long as the text layer is in front of the image layer, it falls under the definition of "text floating above an image" in this invention.
[0033] Displaying text content and images side-by-side or vertically on the same page, including but not limited to: Side-by-side display: Text content is displayed to the left or right of the image, with the text and image presented side-by-side at the same visual level; Vertical arrangement: Text content is displayed above or below the image, with the text and image arranged in a vertical order.
[0034] Regardless of whether a floating or non-floating display method is used, the text and images maintain a content-related correspondence, meaning that the text of the same chapter and the images generated for that chapter are presented together on the same display page.
[0035] Users can adjust the layout of images and text online, including switching display modes and adjusting the order of images and text. After the creator confirms that the text and images are arranged correctly, they will be integrated into the novel page and provided to readers through the platform for text and image reading services; The system outputs graphic works in the form of complete novels that can be read by readers on the platform.
[0036] Automatic compliance checks are performed on the content at each stage of text generation and image generation.
[0037] Beneficial effects
[0038] Compared with the prior art, the present invention has the following significant advantages:
[0039] Users do not need programming skills or to install any software to complete all operations from novel text generation, image generation, text-image fusion and layout, to platform reading display within a unified interface. The system's final output is a complete novel available for readers to enjoy on the platform.
[0040] This invention supports two display modes for text and images: floating and non-floating. The floating display layer overlays the text layer on top of the image layer, creating a text-first, image-later arrangement that provides an immersive reading experience. The non-floating display layer arranges the text and images side-by-side or vertically, suitable for reading scenarios where both text and images are important and the image needs to be fully displayed. Users can freely choose or switch between display modes as needed, enhancing the flexibility of the reading experience.
[0041] Through the "Character Visual Identity Profile" mechanism, users can precisely control the visual characteristics of characters, such as appearance, clothing, and physique, through detailed custom descriptions. This profile is forcibly invoked throughout the entire image and text generation process, ensuring that the character image remains highly consistent across all images in the novel, fundamentally solving the core pain point of "inconsistent character image" in AI-generated content.
[0042] This invention transforms the traditional "black box" generation mode of AI creation tools into a "white box" transparent mode. Users can view and modify the complete prompts used by the AI, achieving fine-grained adjustments that are "what you change is what you get." This gives users precise control over the generated content and significantly reduces the cost of trial and error.
[0043] It supports three input methods and two AI generation modes, which users can freely choose and combine to meet diverse needs, from "importing and optimizing existing complete manuscripts" to "refining creation chapter by chapter".
[0044] This invention simultaneously generates and structurally stores video generation auxiliary information such as storyboards, character settings, scene descriptions, and image prompts during the text and image generation process. When creators decide to further transform novel content into video, they can directly access the aforementioned auxiliary information without having to perform repetitive work such as text decomposition and storyboard design. This significantly reduces the initial preparation costs and technical barriers to video development, achieving seamless data integration from text and image creation to video development. Attached Figure Description
[0046] Figure 1 This is a block diagram of the overall system structure of the present invention; Figure 2 This is a schematic diagram of the input method selection interface in the user input module of the present invention; Figure 3 This is a schematic diagram of the role prompt word editing interface in the user input module of the present invention; Figure 4 This is a schematic diagram of the image generation prompt word display and editing interface in the AI image generation module of the present invention; Figure 5 This is a flowchart illustrating the workflow of the AI image generation module of the present invention. Figure 6 This is a schematic diagram illustrating the floating display effect of the graphic-text fusion display module of the present invention; Figure 7 This is a schematic diagram illustrating the non-floating display method of the graphic fusion display module of the present invention; Figure 8 This is a schematic diagram illustrating the platform reading display of the novel page that integrates text and images according to the present invention. Detailed Implementation
[0047] Example
[0048] The following example, a complete novel creation process, will be used to illustrate in detail the specific implementation of this invention on the platform.
[0049] Setting: Creator "Zhang" logs into the creation platform of this invention, planning to create a graphic novel with a martial arts theme.
[0050] Zhang selected the "Itemized Input" method through the platform interface and entered the hierarchical and character settings for the entire novel item by item. The system generated character description prompts and displayed them to Zhang on the platform. After reviewing them, Zhang felt that the protagonist's description lacked detail and supplemented the description through the prompt editing interface. After the modification was completed, the system updated the character's visual identity profile based on the revised prompts.
[0051] Zhang selected "Chapter Generation Mode" on the platform interface, generated and confirmed each chapter, and completed the creation of all 30 chapters.
[0052] Zhang clicks "Generate Image" on the platform interface. The AI image generation module generates images for each chapter. When previewing the images, if Zhang is not satisfied with a particular image, he clicks on it to access the prompt message viewing interface. The system displays the complete image generation prompt message for that image. After modifying the prompt message, Zhang clicks "Regenerate," and the system regenerates the image based on the modified prompt message, achieving a satisfactory result.
[0053] At the same time, the system automatically stores the storyboards, character settings, scene descriptions, and image prompts (including the final version modified by the user) of each chapter in a structured manner, forming a video generation auxiliary information dataset, which provides a data foundation for possible subsequent video development.
[0054] Zhang entered the integrated text and image layout interface. In the first chapter, wanting to emphasize the visual effect of the images, Zhang selected the "floating background image mode," placing the text above the images with a semi-transparent background to create an immersive reading experience. In the second chapter, due to the numerous details in the images, Zhang didn't want the text to obscure the view, so he switched to the "vertical arrangement display mode," displaying the text below the images. Zhang flexibly switched the display method according to the characteristics of different chapters. After completing the text and image layout for all chapters, Zhang presented the integrated text and image novel page to readers through the platform, providing them with a text and image reading service.
[0055] Example
[0056] Scenario setting: Creator "Li" has completed the first draft of a novel on his computer and hopes to import it into the platform for AI illustration and text display.
[0057] After logging into the platform, Li selected "Upload Entire Novel File." The system automatically parsed the text and extracted character information, generating an initial character setting card for Li's confirmation. After Li reviewed and modified the settings extracted by the system, the system updated the character's visual identity profile.
[0058] Steps 2 to 4: Similar to Example 1, complete the generation of illustrations (users can view and modify the prompts for each image) and the integrated display of text and images (users can freely switch between floating and non-floating modes). The system synchronously stores auxiliary information for video generation, and finally forms a complete novel that readers can read on the platform.
Claims
1. An AI-assisted graphic novel generation system based on a creation platform, characterized in that, The system is integrated into an online creation platform and includes: The user input module, located at the platform front end, is used to receive creative information input by creators; the user input module supports at least one of the following input methods: Itemized input settings, including overall novel hierarchy settings and chapter hierarchy settings; Upload the entire novel file; the system will automatically recognize the chapter structure and import the full text. Upload files by chapter and import the text content of each chapter in sequence; The user input module also includes a character appearance customization unit, which is used to receive the user's description of the appearance features of each character and generate a character visual identity profile. The AI text generation module generates novel text content based on a large language model and the creative information received by the user input module; the AI text generation module supports whole-book generation mode and chapter generation mode; The full-book generation mode is based on the full novel's hierarchical settings and generates all chapters of the complete novel's text content at once; The chapter generation mode generates text content chapter by chapter based on the overall novel hierarchy setting and the chapter hierarchy setting of the currently specified chapter. The AI image generation module, located on the platform backend, is used to generate several images matching the storyline for each chapter of the novel's text content; the AI image generation module calls the character visual identity file to ensure that the same character appears consistently in images across different chapters; The AI image generation module also includes an image prompt display and editing interface, which is used to display prompts that guide image generation to the user through the platform interface, receive the user's editing and modification of the prompts, and regenerate the corresponding image based on the modified prompts; The image and text fusion display and platform publishing module is set on the front end of the platform and is used to merge and display the novel's text content with the images. The fusion display includes a floating display mode in which the text is superimposed on the images and / or a non-floating display mode in which the text and images are arranged vertically or horizontally, forming an image and text fusion novel page, and providing image and text reading services to readers through the platform. The system outputs graphic works in the form of complete novels that can be read by readers on the platform.
2. The system according to claim 1, characterized in that, The AI image generation module also includes a video generation auxiliary information generation unit, which generates video generation auxiliary information for each chapter simultaneously while generating images. The video generation auxiliary information includes at least one of storyboard scripts, scene descriptions, and image prompts, and is stored in a structured manner.
3. The system according to claim 1, characterized in that, The user input module also includes a character prompt word display and editing unit, which is used to convert the character appearance feature description into character description prompt words and display them to the user through the platform interface, receive the user's editing and modification of the prompt words, and update the character visual identity file according to the modified prompt words.
4. The system according to claim 1, characterized in that, The image and text fusion display and platform publishing module supports at least one of the following display modes: Text and images are displayed in an overlay manner, with the text layer above the image layer, and the two overlap in at least a portion of the display area; Bottom-fixed floating mode; Top floating mode; Centered floating mode; User-defined floating position mode; Background image floating mode; Text and image separation display mode: Text and images are displayed independently within the display area, arranged vertically or horizontally, with no overlapping areas between them.
5. The system according to claim 1, characterized in that, When a user uploads the entire novel file or chapters, the user input module also includes a text parsing unit, which performs semantic parsing on the uploaded novel text, automatically extracts character information, and generates initial character settings for user confirmation.
6. The system according to claim 1, characterized in that, The AI image generation module includes: The storyboard generation unit is used to generate storyboards based on novel text. The prompt word construction unit is used to generate prompt words by constructing images based on the storyboard and the character visual identity profile; Image generation unit, used to generate images based on text-based image models; The character consistency maintenance unit is used to ensure that the image of the same character is consistent in different pictures based on the character's visual identity profile.
7. A method for generating AI-assisted graphic novels based on a creation platform, characterized in that, Includes the following steps: S1: Receive creative information input by the creator on the platform interface; the input method includes at least one of the following: The input options include settings for the entire novel's hierarchy and chapter hierarchy. Upload the entire novel file, and the system will automatically parse its chapter structure; Upload files by chapter and import the text content of each chapter in sequence; Receive user descriptions of the appearance features of each character and generate a visual identity profile for each character; S2: Generate novel text content based on a large language model according to the generation mode selected by the creator; If the full-book generation mode is selected, the text content of all chapters of the complete novel will be generated at once based on the full novel hierarchy settings; If the chapter generation mode is selected, text content will be generated chapter by chapter based on the overall novel hierarchy settings and the chapter hierarchy settings of the currently specified chapter; S3: Call the character's visual identity file and generate several images that match the storyline for each chapter to ensure that the same character appears consistently in images across different chapters; The platform displays the generated prompts from the guidance images to the user, allows the user to edit and modify the prompts, and regenerates the corresponding images based on the modified prompts. S4: The text and images are integrated and displayed. The integrated display includes a floating display method in which the text is superimposed on the image and / or a non-floating display method in which the text and the image are arranged vertically or horizontally to form a text and image integrated novel page, and the text and image reading service is provided to readers through the platform. The method outputs graphic works in the form of a complete novel that can be read by readers on the platform.
8. The method according to claim 7, characterized in that, Step S3 further includes: while generating images, simultaneously generating video generation auxiliary information for each chapter, the video generation auxiliary information including at least one of storyboard script, scene description and image prompts, and storing it in a structured manner.
9. The method according to claim 7, characterized in that, Step S1 further includes: converting the character appearance feature description into character description prompts and displaying them to the user through the platform interface, receiving the user's editing and modification of the prompts, and updating the character visual identity profile according to the modified prompts.
10. The method according to claim 7, characterized in that, In step S4, the fusion display includes at least one of the floating display method and the non-floating display method.
