Procedural 3D asset generation support system

Abstract

To provide a professional 3D asset generation support system for producing a 3D asset only by text input even without advanced expertise such as a 3D engineer.SOLUTION: The automatic 3D asset generation support system 10 selects a 3D asset template from the database 30 based on the text information input to the user prompt 20, generates a 3D asset from the selected template and the generated parameters by the AI, and modifies the generated 3D asset by adding text information to the user prompt 20 so that the AI modifies the parameters of the selected 3D asset template or the AI selects a 3D asset template different from the selected 3D asset template and generates the parameters of the template to modify the 3D asset.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a procedural 3D asset generation support system that enables users who wish to create video games or movies using computer graphics to efficiently create 3D assets that are essential for creating computer graphics. [Background technology]

[0002] 3D assets are an essential element in creating computer graphics, and there are several well-known methods for creating them, including manually creating 3D assets using 3D modeling software, scanning objects with a 3D scanner and using that data to create 3D assets, and creating 3D assets using textures and shapes generated by algorithms (procedural generation).

[0003] Among these, the procedural generation method of creating 3D assets uses a computer program (specific algorithms and parameters) to generate 3D assets. This method is not created by hand by a 3D modeler, but is automatically generated based on algorithms and rules. This method is therefore more efficient than manual work, and has the advantage that different variations can be easily generated by repeating the same procedure.

[0004] As described above, procedural generation is an efficient and powerful method for creating 3D assets. However, appropriate algorithms and parameter settings are important, and attention to detail (such as algorithms and parameters) is required to achieve realistic rendering. Current templates (materials for generating 3D assets) are difficult to use without prior knowledge of parameters, etc., requiring 3D engineers to have advanced specialized knowledge. Furthermore, the initial development of templates for procedural generation takes time. In light of this current situation, the applicants conducted extensive research and development to overcome these drawbacks, resulting in the present invention.

[0005] Patent Document 1 discloses a method for designing a 3D modeled object representing a manufactured product (Patent Document 1: Title of the Invention), with the objective of "providing a computer-implemented method, program, medium, and system for designing a 3D modeled object representing a manufactured product (Patent Document 1: Abstract)," and states that "the method comprises the steps of obtaining a base mesh representing the 3D modeled object; selecting connecting edges in the base mesh to obtain a path of the connecting edges consisting of two endpoints; subdividing the base mesh based on the selected edges; and outputting the subdivided base mesh. The subdivision step includes the steps of obtaining a bevel pattern region spanning the selected path; obtaining a transition region by grouping all faces that share at least one of the two endpoints of the path, excluding those of the calculated bevel pattern region; and remeshing the transition region (Patent Document 1: Abstract)." [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Japanese Patent Application Publication No. 2024-016009 Summary of the Invention [Problem to be solved by the invention]

[0007] The design of a 3D modeled object representing a manufactured product according to Patent Document 1 (Patent Document 1: Title of the Invention) requires various steps, such as obtaining a base mesh, selecting a connecting edge in the base mesh and obtaining the path of the connecting edge consisting of two end points, subdividing the base mesh based on the selected edges, and outputting the subdivided base mesh. As the user must perform work (to create a 3D model) at each step, the skills and knowledge (experience) required to perform each step are required, which makes 3D model creation undesirable as it is time-consuming.

[0008] The object of this invention is to provide a procedural 3D asset generation support system that enables users to set appropriate algorithms and parameters and create 3D assets simply by entering text, even if they do not have the advanced specialized knowledge (including prior knowledge of algorithms, parameters, etc.) of a 3D engineer. [Means for solving the problem]

[0009] In order to solve the above problem, the invention described in claim 1 is a procedural 3D asset generation support system that can efficiently generate 3D assets for users who want to create computer graphics, etc., and a user prompt where the user enters text information; A database that has accumulated numerous 3D asset templates, a template selection AI that selects a plurality of 3D asset templates from the database based on text information entered in the user prompt; a parameter generation AI that selects one of a plurality of 3D asset templates selected from the database based on text information entered in the user prompt, and generates parameters for the selected template; a generation system that generates a 3D asset from the 3D asset template selected by the parameter generation AI and the generated parameters; The system is characterized by being a procedural 3D asset generation support system that modifies the 3D asset using the (re)generated parameters by having the parameter generation AI modify the parameters of the 3D asset template selected by the parameter generation AI by inputting additional text information into the user prompt in order to modify the 3D asset generated by the generation system. Note that in this specification, parameters refer to settings for adjusting the scale (relative size), dimensions (absolute size), position, etc. of the 3D asset to be generated.

[0010] The invention described in claim 2 is characterized in that, in the invention described in claim 1, in order to regenerate a 3D asset different from the 3D asset generated by the generation system, by inputting additional text information to the user prompt, multiple 3D asset templates with combinations different from the 3D asset template selected by the template selection AI are selected, and the parameter generation AI (again) selects one from the multiple 3D asset templates, generates parameters for the selected 3D asset template, and regenerates a 3D asset using the selected 3D asset template and the generated parameters.

[0011] The invention described in claim 3 is characterized in that, in the invention described in claim 1 or claim 2, the parameter generation AI and the generation system are a procedural 3D asset generation support system that generates a scene. Note that in this specification, a scene refers to an environment in which multiple 3D assets (e.g., chairs and desks) exist.

[0012] The invention described in claim 4 is a procedural 3D asset generation support system according to claim 1 or claim 2, which processes the source code and descriptions of the 3D asset templates stored in the database, as well as the user prompts, as learning data, and uses a 3D asset generator LLM trained with the learning data to generate 3D asset templates that do not exist in the database.

[0013] The invention described in claim 5 is a procedural 3D asset generation support system according to the invention described in claim 3, which processes the source code and descriptions of the 3D asset templates stored in the database, as well as the user prompts, as learning data, and uses a 3D asset generator LLM trained with the learning data to generate 3D asset templates that do not exist in the database. [Effects of the Invention]

[0014] The procedural 3D asset generation support system of the present invention allows users to access a database of 3D asset templates. Furthermore, by utilizing AI technology, users can input only text information, and the AI ​​will select the optimal 3D asset template from a large database based on the text information. By setting the parameters of that template, 3D assets can be generated.

[0015] Unlike in the past, users no longer have to search for a specific (necessary) 3D asset template from a large number of templates or deal with complex parameters; they can now generate 3D assets simply by entering text information. In other words, 3D assets can now be generated from text information alone. As a result, the cost of producing 3D assets can be significantly reduced. [Brief explanation of the drawings]

[0016] [Figure 1] FIG. 1 is a diagram illustrating an overview of a procedural 3D asset generation support system. [Figure 2] FIG. 1 is a diagram illustrating an example of 3D asset generation by a procedural 3D asset generation support system. [Figure 3] FIG. 1 is a flowchart illustrating an overview of a procedural 3D asset generation pipeline. [Figure 4] This is a flowchart of the procedural 3D asset generation pipeline (part 1). [Figure 5] This is a flowchart of the procedural 3D asset generation pipeline (part 2). [Figure 6] This is a flowchart of the procedural 3D asset generation pipeline (part 3). [Figure 7] This is a flowchart of the AI-assisted world builder (part 1). [Figure 8] This is a flowchart of the AI-assisted world builder (part 2). [Figure 9] This is a flowchart of the AI-assisted world builder (part 3). [Figure 10] FIG. 1 is a diagram illustrating a server structure in a procedural 3D asset generation support system. [Figure 11] FIG. 10 is a flowchart of the template selection AI. [Figure 12] FIG. 10 is a flowchart of the parameter generation AI. [Figure 13] This is a flowchart of the 3D asset generator LLM. DETAILED DESCRIPTION OF THE INVENTION

[0017] <Outline of the procedural 3D asset generation support system> An embodiment of a procedural 3D asset generation support system according to the present invention will be described in detail below with reference to Figures 1 to 13. Figure 1 is a diagram for explaining an overview of the procedural 3D asset generation support system.

[0018] The procedural 3D asset generation support system 10 according to the present invention can efficiently generate 3D assets for users who want to create computer graphics or the like (who do not have the specialized knowledge of a 3D engineer). As shown in Figure 1, the procedural 3D asset generation support system 10 includes a user prompt 20 into which a user inputs text information (in Figure 1, the user inputs "chozubachi" (water basin)), a database 30 that stores a large number of 3D asset templates (in Figure 1, templates 1, 2, 3, ..., template N), a template selection AI 40 that selects multiple 3D asset templates from the database 30 based on the text information input into the user prompt 20 (in Figure 1, templates 4, 7, and 99 are selected), a parameter generation AI 50 that selects one template from the multiple 3D asset templates selected by the template selection AI 40 based on the text information input into the user prompt 20 and generates parameters (in Figure 1, template 4 is selected and parameters for template 4 are generated), and a generation system 60 that generates a "3D asset (chozubachi)" from the template selected by the parameter generation AI 50 and the generated parameters.

[0019] If the user inputs "chozubachi" (water basin) as text information into the user prompt 20, and after checking the generated 3D asset, needs to modify it (note: it may be regenerated; the AI ​​decides whether to modify or regenerate), that is, to modify the 3D asset generated by the generation system 60, the user can input additional text information into the user prompt 20, and the parameter generation AI 50 will modify the parameters of the (already) selected 3D asset template, thereby modifying the 3D asset.

[0020] On the other hand, when regeneration (decision is made by AI), the template selection AI 40 selects multiple 3D asset templates with a different combination from the (already) selected 3D asset templates from the database 30 (in FIG. 1, a different combination of templates is selected, not limited to templates 4, 7, and 99). Then, the parameter generation AI 50 selects one of the multiple 3D asset templates selected by the template selection AI 40, generates parameters for that template, and can regenerate a new 3D asset different from the 3D asset generated by the generation system 60 from the template selected by the parameter generation AI 50 and the generated parameters.

[0021] FIG. 2 is a diagram showing an example of 3D asset generation by the procedural 3D asset generation support system 10. As shown in FIG.

[0022] When a user inputs the text information "A torii made of stone" into the user prompt 20, the AI ​​(artificial intelligence) determines from the text information input into the user prompt 20 that "the object the user wants to generate is a torii made of stone," and then selects the template that the AI ​​determines to be the most suitable for generating a 3D asset related to the object the user wants to generate from the 3D asset templates stored in the database 30, and generates a 3D asset related to the object the user wants to generate (see Figure 2).

[0023] If the user (looking at this 3D asset) wants to make modifications, they can enter additional text information into the user prompt 20, such as "Make the pillars a little thicker...", and the parameter generation AI 50 will modify the parameters of the already selected 3D asset template to modify the 3D asset generated by the generation system 60. Alternatively, (at the AI's discretion) the template selection AI 40 can select multiple 3D asset templates from the database 30 that are different in combination from the (already) selected 3D asset template, and the parameter generation AI 50 can select one of the multiple 3D asset templates, generate parameters for that template, and regenerate the 3D asset generated by the generation system 60.

[0024] <Procedural 3D asset generation pipeline> FIG. 3 is a flowchart illustrating an “overview of a procedural 3D asset generation pipeline,” which is a series of steps and processes for generating 3D assets (which form the core) in the procedural 3D asset generation support system 10.

[0025] When a user enters the “asset description” they want to generate in a user prompt (user prompt 20), a procedural 3D asset template selection AI (template selection AI 40) selects multiple 3D asset templates based on the text information entered in the user prompt.

[0026] If the AI ​​(template selection AI 40) identifies a template associated with the user prompt, it passes the identified template to the procedural 3D asset parameter generation AI (parameter generation AI 50).

[0027] On the other hand, if the AI ​​(template selection AI 40) cannot identify a template associated with the user prompt, the procedural 3D asset generation support system (procedural 3D asset generation support system 10) sends a message to the user asking them to add more information about the asset they want to generate. The user must provide more information about the asset they want to generate. This flow is repeated until the AI ​​(template selection AI 40) identifies a template.

[0028] The AI ​​(parameter generation AI50) selects one template from the multiple templates identified by the procedural 3D asset template selection AI (template selection AI40), and generates parameters for that template.

[0029] If the AI ​​(parameter generation AI 50) determines that it can generate one template and the parameters for that template, the procedural 3D asset generation system (generation system 60) generates a 3D asset using the template selected by the procedural 3D asset parameter generation AI (parameter generation AI 50) and the parameters generated. If the user determines that a satisfactory asset has been generated, the generated 3D asset is downloaded.

[0030] On the other hand, if the AI ​​(parameter generation AI 50) determines that it cannot generate a template and its parameters, the procedural 3D asset generation support system (procedural 3D asset generation support system 10) sends a message to the user requesting that they add detailed information about the asset they want to generate. The user must provide detailed information about the asset they want to generate. This flow is repeated until the generation system (generation system 60) can generate a 3D asset using the template selected by the AI ​​(parameter generation AI 50) and the parameters generated.

[0031] Even if the AI ​​(parameter generation AI 50) determines that it can generate one template and the parameters for that template, and the procedural 3D asset generation system (generation system 60) generates a 3D asset, if the user does not feel satisfied with the generated asset, the user can add detailed information about the asset they want to generate. This flow is also repeated until the user feels satisfied with the generated asset.

[0032] 4 to 6 are flowcharts (to explain the details) of the procedural 3D asset generation pipeline: Fig. 4 is a flowchart (to explain the details) of the user prompt 20 and template selection AI 40, Fig. 5 is a flowchart (to explain the details) of the parameter generation AI 50, and Fig. 6 is a flowchart (to explain the details) of the generation system 60.

[0033] As shown in Fig. 4, the user inputs a "description of the asset to be generated" in the user prompt 20. At the same time, the user selects whether to bake textures, at what resolution to bake, etc.

[0034] The template selection AI 40 creates embeddings (a technique for converting high-dimensional data into low-dimensional real vectors) from the asset description entered by the user in the user prompt 20. It compares the embeddings of the created asset description with the embeddings of 3D asset templates (stored in the database 30) and measures the similarity. As a result, it selects multiple templates that are highly similar to the embeddings of the created asset description and fit within the context space of the parameter generation AI 50 (the length of text taken into account when generating the generative model and the size of past information held by the model). The 3D asset templates stored in the database 30 are equipped with a function to create embeddings of 3D asset template descriptions, parameters, and materials (for all 3D asset templates) as an automated task.

[0035] The parameter generation AI 50 prepares a prompt containing predefined generation instructions, a user-entered asset description, and a description of the template, parameters, and materials selected by the template selection AI 40, as shown in Figure 5. After preparing the prompt, it selects a 3D asset template and one or more materials to apply based on the asset description, and generates a message for the user and parameters to be used by the generation system 60.

[0036] If the AI ​​can generate parameters for the generation system 60, the parameter generation AI 50 packs (converts models, algorithms, data, etc. into a format that is easily usable by bundling together) the selected 3D asset template, the generated parameters, and the material library as a request to the generation system 60. If the AI ​​cannot generate parameters for the generation system 60, the parameter generation AI 50 prepares a message for the user.

[0037] The generation system 60 loads the 3D asset template, sets generation parameters, and generates meshes and UVs, as shown in Figure 6. If bake materials are enabled, the system bakes materials and packs the mesh into a downloadable file (GLB, USD, FBX, OBJ, etc.). If bake materials are not enabled, the system packs the mesh into a downloadable file (GLB, USD, FBX, OBJ, etc.) without baking materials. The generation system 60 then prepares a message for the user.

[0038] The user checks the message from the AI ​​and the generated 3D asset. If a satisfactory asset is generated, the user downloads it and the 3D asset generation is complete. If the user is not satisfied, they add detailed information about the asset they want to generate through a user prompt 20. The template selection AI 40 then identifies multiple templates related to the user prompt, and the parameter generation AI 50 selects one template and one or more materials to apply based on the added asset description, and generates a message for the user and parameters for the generation system 60. The generation system 60 then repeats this process. This flow is repeated until the user is satisfied (see Figures 4 to 6).

[0039] <AIアシストワールドビルダー> 7 to 9 are flowcharts (for explaining the details) of the AI-assisted world builder. FIG. 7 is a flowchart (for explaining the details) of the user prompt 20 and template selection AI 40, FIG. 8 is a flowchart (for explaining the details) of the parameter generation AI 50, and FIG. 9 is a flowchart (for explaining the details) of the generation system 60. As shown in FIG. 7, the user inputs a "description of the scene to be generated" into the user prompt 20. At the same time, the user selects whether to bake textures, at what resolution to bake, etc. A scene is an environment in which multiple 3D assets exist.

[0040] The procedural 3D asset generation support system 10 creates embeddings (a technique for converting high-dimensional data into low-dimensional real vectors) from a scene description entered by a user in a user prompt 20. The embeddings of the created scene description are compared with the embeddings of 3D asset templates (stored in a database 30) to measure their similarity. As a result, one or more templates that are highly similar to the embeddings of the created scene description and fit within the context space of the parameter generation AI 50 (the length of text taken into account when generating the generative model and the size of past information held by the model) are selected. The 3D asset templates stored in the database 30 are equipped with a function for creating embeddings of 3D asset template descriptions, parameters, and materials for all 3D asset templates as an automated task.

[0041] The parameter generation AI 50 prepares a prompt containing predefined generation instructions, a scene description entered by the user, and descriptions of the templates, parameters, and materials selected by the template selection AI 40, as shown in Figure 8. After preparing the prompt, it selects one or more 3D asset templates and one or more materials to apply based on the asset descriptions, and generates a message for the user and parameters to be used by the generation system 60 of Figure 1.

[0042] If the AI ​​can generate parameters for the generation system 60, the parameter generation AI 50 packs (converts models, algorithms, data, etc. into a format that can be easily used) the selected 3D asset template, generated parameters, and material library as a request to the generation system 60. If the AI ​​cannot generate parameters for the generation system 60, the parameter generation AI 50 prepares a message for the user. (See Figure 8: Parameter Generation AI 50)

[0043] If there are 3D asset templates that need to be processed in addition to the 3D assets generated above (for example, if a 3D chair asset needs to be generated in addition to the 3D desk asset that has already been generated), the procedural 3D asset generation support system 10 loads the 3D asset template (related to a chair), sets the generation parameters, generates a mesh (vertices, edges, and polygons) and UVs (the coordinate system used to apply textures, which are images with color and texture information, to a 3D model), and if bake material is enabled, bakes the material (a process that calculates the effect of textures and light placed on the surface of a 3D model and saves the results as a still image or texture), and sets the position of the asset. If bake material is not enabled, the position of the asset is set as is. If there are no asset templates that need to be processed, the system packs the asset into a downloadable file (GLB, USD, FBX, OBJ, etc.) and prepares a message to the user (see Figure 9: generation system 60).

[0044] The user reviews the message from the AI ​​and the generated 3D scene. If the generated 3D scene is satisfactory, the user downloads it and completes the process. If not, the user adds more information about the 3D scene they want to generate. The template selection AI 40 then identifies a template associated with the user prompt 20, and the parameter generation AI 50 selects one or more templates and one or more materials to apply based on the added scene description, and generates a message for the user and parameters for the generation system 60. The generation system 60 then repeats this process again. This process is repeated until the user is satisfied.

[0045] <3D asset generation method using the procedural 3D asset generation support system> The procedural 3D asset generation support system 10 is a method for efficiently generating 3D models for users who want to create computer graphics and the like (who do not have the specialized knowledge of a 3D engineer).

[0046] The system includes an input step in which a user inputs text information into a user prompt 20, a generation AI step in which a template selection AI 40 identifies a relevant template from a database 30 based on the text information input into the user prompt 20, a parameter generation AI 50 selects one or more templates and one or more materials to be applied based on the text information input into the user prompt 20, and generates a message for the user and parameters for a generation system 60, and the generation system 60 generates a 3D asset using the template and materials selected and the parameters generated by the parameter generation AI 50 of FIG. 1.

[0047] Furthermore, by adding text information to the user prompt 20, the parameter generation AI 50 modifies the parameters of the selected 3D asset template for the 3D asset generated in the generation AI process, or the template selection AI 40 selects multiple 3D asset templates from the database 30 that are different in combination from the selected 3D asset template, and the parameter generation AI 50 selects one or more of the multiple 3D asset templates selected by the template selection AI 40, generates parameters for those templates, and regenerates (the 3D asset generated in the generation AI process) from the 3D asset template selected by the parameter generation AI 50 and the generated parameters.

[0048] <Server structure for procedural 3D asset generation support system> Fig. 10 is a diagram illustrating the server structure in the procedural 3D asset generation support system 10. As shown in Fig. 10, the server structure in the procedural 3D asset generation support system 10 is composed of a distributed DB server 11 (a part that runs behind the scenes of applications and web services and stores the backend responsible for data processing, business logic, database management, etc.), a distributed storage server 12 (storing AI models), a distributed embedding storage server 13 (storing template embeddings), a distributed DB server 14 (storing material detail information), a distributed DB server 15 (storing template detail information), a distributed storage server 16 (storing material generation code), and a distributed storage server 17 (storing template generation code).

[0049] Note that template details information refers to the description and usage of the template, the description and usage of the template parameters (e.g., scale, number of polygons, etc.), and the description and usage of the template's available material slots, and material details information refers to the description and usage of the material parameters (e.g., scale, number of polygons, etc.).

[0050] The relationship between the above server structure and the process flow of the procedural 3D asset generation support system 10 (each element constituting the server structure is described within (*)) will be explained with reference to FIG.

[0051] User (Client PC1, Client PC2,..., Client PCN: Nth PC) → Load balancer (a device that distributes server traffic across multiple servers to even out the load and improve system performance and reliability) → Backend server (Server 1, Server 2,..., Server N (Nth server)): (*Distributed DB server 11 (contains the backend))

[0052] (Continued from 0050) → Load Balancer → Template Selection AI40 (GPU Server 1, GPU Server 2,..., GPU Server N (Nth GPU Server)): (* Distributed Storage Server 12 (Storing AI Models): (* Distributed Embedding Storage Server 13 (Storing Template Embeddings)

[0053] (Continued from 0051) → Parameter generation AI50 (GPU server 1, GPU server 2, , GPU server N (Nth GPU server)): (* Distributed DB server 14 (storing material details information): (* Distributed DB server 15 (storing template details information)

[0054] (Continued from 0052) → Generation system 60 (GPU server 1, GPU server 2, ..., GPU server N (Nth GPU server)): (* distributed storage server 16 (storing material generation code): (* distributed storage server 17 (storing template generation code)). The server structure in the procedural 3D asset generation support system 10 is characterized by the fact that it stores a template selection AI 40 and a parameter generation AI 50 (see Figure 10).

[0055] Figure 11 is a flowchart of the Template Selection AI 40. Multiple template selection via the Template Selection AI 40 is made possible by a process that adds detailed descriptions to every template, including a description and usage of the template, a description and usage of the template's parameters (e.g., scale, polygon count, etc.), and a description and usage of the template's available material slots.

[0056] The Template Selection AI 40 describes the template itself and the parameters and materials available for that template, calculates an embedding (vector representation) for each template, and stores it in an accessible way, allowing it to compare the stored embeddings with embeddings in a user-provided text description of the 3D asset, creating a list of templates scored by relevance.

[0057] 12 is a flowchart of the parameter generation AI 50. The parameter generation AI 50 selects one or more templates and generates parameters by describing the parameters and materials that can be used in the templates and providing this information to the parameter generation AI 50. The parameter generation AI 50 is configured (programmed) to use this information to generate the necessary parameters for each template.

[0058] Figure 13 is a flowchart of the 3D asset generator LLM's training data processing. The 3D asset generator LLM stores the generation code of the procedural 3D asset template in distributed storage. The generation code is converted into source code, and this source code and detailed descriptions of the template are used as training data to train the 3D asset generator LLM's model. The model trained by the procedural 3D asset template selection AI (template selection AI 40) is used to generate 3D asset templates that do not exist in the database.

[0059] <Effects of the Pro-Jal 3D Asset Creation Support System> The pro-active 3D asset generation support system 10 enables the efficient creation of 3D assets (3D images and videos) to be used in digital content production such as game development, animation production, virtual reality (VR) and augmented reality (AR) applications, i.e., 3D assets to be used in three-dimensional space.

[0060] Specifically, by allowing a user to access the database 30 and further utilizing AI, the AI ​​can select the optimal template (at least one or more) from a large database based on the text information, simply by inputting text, and generate a 3D asset from the parameters of the selected template and the generated template.

[0061] Unlike in the past, users no longer have to search for a specific (necessary) template from a large number of templates or deal with complex parameters; they can now create 3D models simply by entering text. In other words, it is now possible to create 3D models from text information alone. As a result, the cost of creating 3D assets has been significantly reduced.

[0062] <Example of changes to the pro-journal 3D asset generation support system> The procedural 3D asset generation system of the present invention is not limited to the aspects of the above-described embodiments, and the configuration of the database, user prompt, template selection AI, parameter generation AI, generation system, etc. can be appropriately modified as needed within the scope of the present invention. [Industrial Applicability]

[0063] The procedural 3D asset generation support system of the present invention has the excellent effects described above, and can therefore be suitably used as a system for creating digital content such as game development, animation production, virtual reality (VR) and augmented reality (AR) applications. [Explanation of symbols]

[0064] 10. 3D asset generation support system 11. Distributed DB server (backend) 12. Distributed storage server (AI model) 13. Distributed Embedding Storage Server (Template Embedding) 14. Distributed DB Server (Detailed Material Information) 15. Distributed DB Server (Template Details) 16. Distributed storage server (material generation code) 17. Distributed storage server (template generated code) 20 User Prompts 30··Database 40··Template Selection AI 50 Parameter Generation AI 60··Generation System

Claims

1. A procedural 3D asset generation support system that can efficiently generate 3D assets for users who want to create computer graphics, etc., a user prompt where the user enters text information; A database that stores a large number of 3D asset templates, a template selection AI that selects a plurality of 3D asset templates from the database based on text information entered in the user prompt; a parameter generation AI that selects one of a plurality of 3D asset templates selected from the database based on text information entered in the user prompt, and generates parameters for the selected template; A generation system that generates a 3D asset from the 3D asset template selected by the parameter generation AI and the generated parameters; A procedural 3D asset generation support system characterized in that, in order to modify the 3D asset generated by the generation system, the parameter generation AI modifies the parameters of the 3D asset template selected by the parameter generation AI by inputting additional text information into the user prompt, thereby modifying the 3D asset using the (re-)generated parameters.

2. The procedural 3D asset generation support system of claim 1, characterized in that in order to regenerate a 3D asset different from the 3D asset generated by the generation system, the user selects multiple 3D asset templates with combinations different from the 3D asset template selected by the template selection AI by additionally inputting text information to the user prompt, and the parameter generation AI (again) selects one from the multiple 3D asset templates, generates parameters for the selected 3D asset template, and regenerates a 3D asset using the selected 3D asset template and the generated parameters.

3. The procedural 3D asset generation support system according to claim 1 or 2, characterized in that the parameter generation AI and the generation system generate a scene.

4. The procedural 3D asset generation support system according to claim 1 or 2, characterized in that the source code and descriptions of the 3D asset templates stored in the database, as well as the user prompts, are processed as learning data, and a 3D asset generator LLM trained with the learning data is used to generate 3D asset templates that do not exist in the database.

5. The procedural 3D asset generation support system according to claim 3, characterized in that the source code and descriptions of the 3D asset templates stored in the database, as well as the user prompts, are processed as learning data, and a 3D asset generator LLM trained with the learning data is used to generate 3D asset templates that do not exist in the database.

Images