Asset reference management method and system of digital picture book generation system and medium
By generating a structured insight package of reference assets through multimodal parsing, the problems of repeated asset parsing and copyright compliance in digital picture book generation systems are solved, achieving efficient resource utilization and stable asset management.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NINGBO JUFENG SYST SOFTWARE CO LTD
- Filing Date
- 2026-04-16
- Publication Date
- 2026-05-15
AI Technical Summary
Existing digital picture book generation systems suffer from resource waste and copyright compliance risks during asset citation. Assets need to be repeatedly parsed at different stages, and inconsistent asset management methods lead to inconsistent prompt word assembly.
Multimodal parsing technology is used to generate a reference asset structured insight package. The package is parsed once when the asset is created or updated to generate a reusable structured insight package, which is then reused throughout the system to verify asset availability and avoid duplicate parsing. When the asset is withdrawn or permissions are changed, the relevant resources are removed.
It enables assets to be parsed repeatedly at different stages without repeated resolution, reducing resource waste, lowering copyright and compliance risks, and ensuring the consistency and stability of asset references.
Smart Images
Figure CN122048533A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of asset management technology, and in particular to an asset reference management method, system and medium for a digital picture book generation system. Background Technology
[0002] A media asset repository is a specialized system / platform for the centralized storage, cataloging, retrieval, management, circulation, and archiving of various digital media content (video, audio, images, text, charts, etc.). It is the core digital asset infrastructure for media organizations and content companies.
[0003] Existing digital picture book generation systems require the creation of assets that fit the actual scene based on the story content when creating digital picture books. These assets generally include character assets, prop assets, and scene assets. There are various ways to acquire these assets. Most of the time, users directly obtain them from existing media asset libraries. For example, they can obtain reference images or page images from existing media asset libraries, save them, and then upload them to the digital picture book generation system. Another type of asset is created by users themselves and then uploaded to the digital picture book generation system. However, whether the assets are obtained from existing media asset libraries or user-created assets from the previous period, they only include the asset's source files and metadata. Digital picture book generation systems typically use these assets across three stages: New Tale (creation), Plan (page-by-page planning), and Edit / Re-generate (page generation / partial regeneration). When the same asset is referenced in all three stages, it requires three parsing processes. The asset parsing results from the asset creation stage cannot be directly reused in the generation chain (corresponding to the page-by-page planning stage and the page generation / partial regeneration stage), leading to repeated parsing of the same asset at different stages and wasting parsing resources. Furthermore, the different structures used in previous asset management methods compared to the digital picture book generation system result in inconsistent prompt word assembly. Additionally, when an asset is withdrawn or its reference permissions change, the historical reference relationships of the digital picture book generation system cannot be automatically cleared, posing copyright and compliance risks.
[0004] Therefore, further improvements to existing technologies are needed. Summary of the Invention
[0005] The first technical problem to be solved by the present invention is to provide an asset reference management method for a digital picture book generation system, which addresses the above-mentioned prior art. This method can achieve full-link reuse by parsing assets only once, without repeatedly parsing assets, and can reduce copyright and compliance risks caused by unauthorized asset references.
[0006] The second technical problem to be solved by the present invention is to provide a digital picture book generation system that applies the above-mentioned asset reference management method.
[0007] The third technical problem to be solved by the present invention is to provide a computer-readable storage medium for implementing the above-described asset reference management method.
[0008] The technical solution adopted by this invention to solve the first technical problem mentioned above is: an asset reference management method for a digital picture book generation system, wherein the assets include at least character assets, prop assets, and scene assets, characterized by including: performing multimodal parsing when an asset is created or updated to obtain a reference asset structured insight package; writing the reference asset structured insight package into the asset storage to achieve end-to-end reuse; verifying the availability of the asset when the digital picture book generation system loads the asset or / and during the process of generating a digital picture book using an asset; if the asset is available, directly referencing the reference asset structured insight package corresponding to the asset to avoid repeated parsing; if the asset is withdrawn or there is no asset reference permission, the digital picture book generation system removes the asset and the reference asset structured insight package corresponding to the asset, and outputs a list of removed assets.
[0009] To enable asset traceability, preferably, an asset reference record table is created. When a digital picture book is created in the digital picture book generation system and a reference asset structured insight package is referenced, the traceability identifier of the asset is recorded in the asset reference record table. When an asset is withdrawn or no asset reference permission is available, the traceability identifier of the asset is deleted from the asset reference record table.
[0010] To enable the creation and updating of assets, a separate asset management library page is set up in the digital picture book generation system. Users can create or update assets through this asset management library page, preprocess the created or updated assets, and then perform multimodal parsing to obtain the corresponding reference asset structured insight package.
[0011] Preferably, the source of the created or updated assets includes reference image transfer, page image transfer, or direct upload.
[0012] Preferably, the method for preprocessing created or updated assets is as follows: first, the assets are format-validated, and then the format-validated assets are compressed into a uniform format and thumbnails are generated.
[0013] Preferably, the process of performing multimodal analysis on assets specifically includes the following steps:
[0014] Step 1: Generate parsing instructions using the created or updated assets;
[0015] Step 2: According to the parsing instructions, call the visual language model to perform visual semantic parsing on the assets and generate a structured JSON file;
[0016] Step 3: Apply appropriate constraints to the structured JSON file generated in Step 2, and output the original reference asset structured insight package;
[0017] Step 4: Perform fault tolerance processing on the original reference asset structured insight package generated in Step 3 and normalize the versions to obtain a standardized reference asset structured insight package;
[0018] Step 5: Persist the standardized reference asset structured insight package obtained in Step 4 to obtain the reference asset structured insight package.
[0019] Through the processing in step 4 above, a standardized reference asset structured insight package can be obtained, which can avoid the problem of inconsistent prompt assembly.
[0020] In order to categorize assets, an asset document is generated simultaneously with the reference asset structured insight package obtained in step 5. The asset document contains the following fields: asset name, visibility, type, path, and timestamp.
[0021] Preferably, in step 4, the fault-tolerant processing of the original reference asset structured insight package generated in step 3 includes syntax fault tolerance processing, field compatibility processing, and anchor point completion processing.
[0022] The technical solution adopted by the present invention to solve the second technical problem mentioned above is: a digital picture book generation system, characterized in that: the above-mentioned asset reference management method is used for asset reference management.
[0023] The technical solution adopted by the present invention to solve the third technical problem mentioned above is: a computer-readable storage medium storing a computer program that can be read and executed by a processor, characterized in that: when the computer program is executed by the processor, it implements the above-mentioned asset reference management method.
[0024] Compared with existing technologies, the advantages of this invention are as follows: Multimodal parsing is performed during asset creation or updating to obtain a reference asset structured insight package. Since this reference asset structured insight package is a reusable structure, it can be reused throughout the entire process. Therefore, the digital picture book generation system can use the same asset multiple times when it needs to be used at different stages or on different generation pages, without repeated VLM parsing, thus reducing the cost of repeated parsing. Furthermore, if an asset is withdrawn or there is no permission to reference it, the digital picture book generation system removes the asset and its corresponding reference asset structured insight package, and outputs a list of removed assets, thereby reducing copyright and compliance risks caused by unauthorized references. Attached Figure Description
[0025] Figure 1 This is a flowchart of the asset reference management method in an embodiment of the present invention. Detailed Implementation
[0026] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.
[0027] The assets in this invention are visual anchor point assets used to generate image consistency control; in this embodiment, the assets include at least character assets, prop assets, and scene assets. The following examples illustrate the assets in this application:
[0028] The asset reference management method of the digital picture book generation system in this embodiment includes: performing multimodal parsing when an asset is created or updated to obtain a reference asset structured insight package; writing the reference asset structured insight package into the asset storage to achieve end-to-end reuse; verifying the availability of the asset when the digital picture book generation system loads the asset or / and during the process of generating a digital picture book using an asset; if the asset is available, directly referencing the reference asset structured insight package corresponding to the asset to avoid repeated parsing; if the asset is withdrawn or there is no asset reference permission, the digital picture book generation system removes the asset and the reference asset structured insight package corresponding to the asset, and outputs a list of removed assets.
[0029] In this embodiment, the reference asset structured insight package is not the original image itself, but a machine-readable image summary. The role of multimodal parsing is to break down the information that can be seen in the original image into structured fields, including character fields, prop fields, scene fields, anchor point constraint fields, etc. In the subsequent generation model of the digital picture book generation system, it is not necessary to repeatedly "re-view the image" (parsing), but to read the reference asset structured insight package as a stable input.
[0030] The process of performing multimodal analysis on assets includes the following steps:
[0031] Step 1: Generate parsing instructions using the created or updated assets;
[0032] In this embodiment, when creating or updating an asset, the user needs to submit information such as the asset name, type, visibility, and optional resolutions. That is, the user submits information such as the asset name, type, and optional resolutions to form a resolution instruction.
[0033] Step 2: According to the parsing instructions, call the visual language model to perform visual semantic parsing on the assets and generate a structured JSON file;
[0034] Step 3: Apply appropriate constraints to the structured JSON file generated in Step 2, and output the original reference asset structured insight package;
[0035] Step 4: Perform fault tolerance processing on the original reference asset structured insight package generated in Step 3 and normalize the versions to obtain a standardized reference asset structured insight package;
[0036] The fault tolerance processing in this embodiment includes syntax fault tolerance processing, field compatibility processing, and anchor point completion processing; the syntax fault tolerance processing includes stripping Markdown code blocks and fixing trailing commas; the field compatibility processing is to unify old fields; the anchor point completion processing is to automatically construct mustMatch for characters / props / scenes if anchors are missing (see the key consistency fields in character assets and prop assets);
[0037] Step 5: Persist the standardized reference asset structured insight package obtained in Step 4 to obtain the reference asset structured insight package; In this embodiment, at the same time as obtaining the reference asset structured insight package in Step 5, an asset document is generated, which includes the following fields: asset name, visibility, type, path, and timestamp.
[0038] The above multimodal parsing is performed in batches. If a large batch of parsing fails, it is split into smaller batches for retry.
[0039] When loading assets, the digital picture book generation system can directly read the corresponding reference asset structured insight package if an asset needs to be referenced in the three stages of New Tale (creation), Plan (page-by-page planning), and Edit / Re-generate (page generation / partial regeneration), thus avoiding the need to re-parse the same asset.
[0040] The following example illustrates the above asset reference management method.
[0041] Character Asset: Guan Yu;
[0042] Item Asset: Green Dragon Crescent Blade;
[0043] Scene Asset: Oath of the Peach Garden;
[0044] Asset reference management method execution process:
[0045] 1. Users upload Guan Yu character images through the asset management library page and trigger multimodal parsing to generate a reference asset structured insight package corresponding to the "Guan Yu" character assets;
[0046] 2. During the digital picture book generation process, in the Plan and Edit stages, when the digital picture book needs to reference the "Guan Yu" character asset in the raw images of pages 1, 4, and 8, the digital picture book generation system will not repeat the VLM parsing of the "Guan Yu" character asset in the Plan and Edit stages, but will directly reference the reference asset structured insight package corresponding to the "Guan Yu" character asset. If the digital picture book generation system needs to generate page 8 again and needs to reference the "Guan Yu" character asset, the digital picture book generation system will also directly reference the reference asset structured insight package corresponding to the "Guan Yu" character asset.
[0047] 3. Users save the "Oath of the Peach Garden" page image as a scene asset and trigger multimodal parsing to generate a reference asset structured insight package corresponding to the "Oath of the Peach Garden" scene asset. If subsequent pages need to use the "Oath of the Peach Garden" scene asset, they can directly reuse the reference asset structured insight package corresponding to the "Oath of the Peach Garden" scene asset.
[0048] 4. If the "Guan Yu" character asset is set to private or withdrawn, the digital picture book generation system removes the "Guan Yu" character asset and its corresponding reference asset structured insight package, and outputs a list of removed assets.
[0049] Results: In the cross-page digital picture book generation system, the character appearance (red face, long beard, green robe) and key props (Green Dragon Crescent Blade) remained stable, and once the assets were withdrawn, they could not be referenced again.
[0050] like Figure 1 As shown, the asset reference management method of the digital picture book generation system in this embodiment involves a front-end input module, an asset service module, and a parsing module. Users create or update assets in the front-end input module; the asset service module generates a parsing instruction after receiving the asset created or updated by the front-end input module; the asset service module sends the parsing instruction to the parsing module; then, after receiving the parsing instruction, the parsing module generates a reference asset structured insight package; the parsing module sends the generated reference asset structured insight package to the asset service module for persistent storage; ultimately, the assets on the asset service module and the corresponding reference asset structured insight packages can be reused throughout the entire process.
[0051] A separate asset management library page is set up in the digital picture book generation system. Users can create or update assets through this page, preprocess the created or updated assets, and then perform multimodal parsing to obtain a reference asset structured insight package. In this embodiment, the sources of created or updated assets include reference image transfer, page image transfer, or direct upload. In addition, the method of preprocessing created or updated assets in this application is as follows: first, the asset format is validated, and then the format-validated assets are compressed into a uniform format and thumbnails are generated.
[0052] In addition, the asset reference management method in this embodiment also includes creating an asset reference record table. When a digital picture book is created in the digital picture book generation system and a reference asset structured insight package is referenced, the traceability identifier of the asset is recorded in the asset reference record table. When an asset is withdrawn or no asset reference permission is available, the traceability identifier of the asset is deleted from the asset reference record table.
[0053] This embodiment also relates to a digital picture book generation system, which uses the above-mentioned asset reference management method to manage asset references.
[0054] In addition, this embodiment also relates to a computer-readable storage medium storing a computer program that can be read and executed by a processor. When the computer program is executed by the processor, it implements the above-mentioned asset reference management method.
[0055] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.
Claims
1. An asset reference management method for a digital picture book generation system, wherein the assets include at least character assets, prop assets, and scene assets, characterized in that... include: Perform multimodal parsing when an asset is created or updated to obtain a structured insight package of the reference asset; The reference asset structured insight package is written into the asset storage to achieve end-to-end reuse. When the digital picture book generation system loads an asset or / and during the process of generating a digital picture book using an asset, the availability of the asset is verified. If the asset is available, the reference asset structured insight package corresponding to the asset is directly referenced to avoid repeated parsing. If the asset is withdrawn or there is no permission to reference the asset, the digital picture book generation system removes the asset and the reference asset structured insight package corresponding to the asset, and outputs a list of removed assets.
2. The asset reference management method of the digital picture book generation system according to claim 1, characterized in that: Create an asset reference record table. When creating a digital picture book in the digital picture book generation system and referencing a structured insight package of a reference asset, record the traceability identifier of the asset in the asset reference record table. When an asset is withdrawn or no longer has asset reference permissions, delete the traceability identifier of the asset in the asset reference record table.
3. The asset reference management method of the digital picture book generation system according to claim 1, characterized in that: A separate asset management library page is set up in the digital picture book generation system. Users can create or update assets through this asset management library page, preprocess the created or updated assets, and then perform multimodal parsing to obtain the corresponding reference asset structured insight package.
4. The asset reference management method of the digital picture book generation system according to claim 3, characterized in that: The sources of the assets created or updated include reference images, page images, or direct uploads.
5. The asset reference management method of the digital picture book generation system according to claim 3, characterized in that: The method for preprocessing created or updated assets is as follows: first, the asset's format is validated, then the validated asset is compressed into a uniform format, and a thumbnail is generated.
6. The asset reference management method of the digital picture book generation system according to claim 1, characterized in that: The process of performing multimodal analysis on assets includes the following steps: Step 1: Generate parsing instructions using the created or updated assets; Step 2: According to the parsing instructions, call the visual language model to perform visual semantic parsing on the assets and generate a structured JSON file; Step 3: Apply appropriate constraints to the structured JSON file generated in Step 2, and output the original reference asset structured insight package; Step 4: Perform fault tolerance processing on the original reference asset structured insight package generated in Step 3 and normalize the versions to obtain a standardized reference asset structured insight package; Step 5: Persist the standardized reference asset structured insight package obtained in Step 4 to obtain the reference asset structured insight package.
7. The asset reference management method of the digital picture book generation system according to claim 6, characterized in that: While obtaining the reference asset structured insight package in step 5, an asset document is generated, which includes the following fields: asset name, visibility, type, path, and timestamp.
8. The asset reference management method of the digital picture book generation system according to claim 6, characterized in that: In step 4, the original reference asset structured insight package generated in step 3 is subjected to fault tolerance processing, including syntax fault tolerance processing, field compatibility processing, and anchor point completion processing.
9. A digital picture book generation system, characterized in that: The asset reference management method described in any one of claims 1 to 8 is used for asset reference management.
10. A computer-readable storage medium storing a computer program and capable of being read and executed by a processor, characterized in that: When the computer program is executed by the processor, it implements the asset reference management method as described in any one of claims 1 to 8.