Model data format conversion method and system for three-dimensional engine

Abstract

The present application relates to the technical field of data format conversion, and specifically to a model data format conversion method and system for a three-dimensional engine, the method comprising: obtaining a GLB format file of target model three-dimensional data, and parsing to obtain the coordinates of the nodes, the geometric data and texture data contained in the binary buffer; determining the structural association value of each node; obtaining the relative feature data sequence of each node, and determining the relative difference value of each node based on the difference between the relative feature data sequence of each node and the nodes in the local area where it is located; obtaining the associated structural feature vector of each node, and determining the partition association value between any two nodes; clustering all nodes based on the partition association value, segmenting and grading the GLB data based on the clustering result, and using the three-dimensional engine to reconstruct the scene of the target model. The present application aims to improve the accuracy of data segmentation during the model data format conversion process.

Description

Technical Field

[0001] The present application relates to the technical field of data format conversion, and in particular to a method and system for converting model data format for a three-dimensional engine. Background Art

[0002] In today's digital age, 3D technology has been widely used in many fields, such as game development, virtual reality, architectural design, industrial manufacturing, etc. However, different 3D software often use their own unique data formats, which leads to data compatibility and interoperability issues. Converting the model data of 3D design software into the scene data format of the real-time 3D engine can enable models created by different 3D design software to be applied in the real-time 3D engine, breaking down the barriers between software and improving work efficiency. Secondly, through this conversion, the powerful functions of the real-time 3D engine can be fully utilized, such as efficient rendering, real-time interaction, physical simulation, etc., to bring users a smoother experience.

[0003] In the existing model data conversion, there are some defects in the segmentation and classification of GLB data. The segmentation algorithm is not accurate enough, which causes unnatural breaks or deformations in the model at the segmentation point, affecting the visual effect; and the accuracy of data segmentation will directly affect the selection of data classification strategy. If the classification strategy is not selected reasonably, it will not be able to accurately distinguish the important and minor parts of the model, resulting in unbalanced resource allocation during real-time rendering and affecting performance. Nowadays, in the process of converting the model data of 3D design software into the scene data of the real-time 3D engine, the traditional uniform segmentation method based on the decreasing bounding box of the spatial range is used to segment the GLB data, resulting in reduced accuracy of data segmentation, which in turn affects the visual effect and real-time rendering performance of the model after the subsequent data format conversion. Summary of the invention

[0004] In view of the above, it is necessary to provide a model data format conversion method and system for a three-dimensional engine, which improves the accuracy of data segmentation during the model data format conversion process compared to the traditional model data segmentation method:

[0005] In a first aspect, an embodiment of the present application provides a method for converting a model data format for a three-dimensional engine, the method comprising the following steps:

[0006] Get the GLB format file of the target model's 3D data, and parse it to get the node coordinates, the geometric data and texture data contained in the binary buffer;

[0007] Determine the structural association value of each node based on the position distribution of nodes in the local area where each node is located, the difference in geometric data between nodes in the local area, and the difference in texture data;

[0008] Based on the difference in structural association values ​​between each node and the nodes in the local area where it is located, as well as the distance between them, a relative feature data sequence of each node is obtained, and based on the difference in relative feature data sequence between each node and the nodes in the local area where it is located, a relative difference value of each node is determined;

[0009] The structural association value and the relative difference value of each node are combined into an association structural feature vector of each node, and based on the distance between any two nodes and the similarity of the association structural feature vectors between the two nodes, the partition association value between the two nodes is determined;

[0010] All nodes are clustered based on the partition association values, the GLB data are segmented and graded based on the clustering results, and a three-dimensional engine is used to reconstruct the scene of the target model.

[0011] In one embodiment, the process of determining the structure association value is as follows:

[0012] A search sphere with a preset radius is constructed with any node as the center, and all nodes in the search sphere are composed of a set of neighboring nodes of the node;

[0013] Calculate the mean of the measured distances between any two nodes in the neighbor node set of each node, recorded as the first distance mean;

[0014] Calculate the mean of the metric distance of the geometric data between all two arbitrary nodes in the neighbor node set of each node, and the mean of the metric distance of the texture data between them, which are respectively recorded as the first mean and the second mean, and record the mean of the first mean and the second mean of each node as the second distance mean;

[0015] Based on the first distance mean and the second distance mean, a structural association value of each node is determined.

[0016] In one embodiment, the process of determining the structural association value is: calculating the product of the first distance mean and the second distance mean, and taking the reciprocal of the sum of the product and a preset value greater than 0 as the structural association value of each node.

[0017] In one embodiment, the method for acquiring the relative characteristic data sequence is:

[0018] Calculate the difference in structural association value between any node and each node in its neighboring node set, as well as the metric distance between them, recorded as the comparison distance, and arrange all the differences in order from small to large according to the comparison distance to form a relative feature data sequence of any node.

[0019] In one embodiment, the relative difference value is the average of the metric distances between each node and all nodes in the set of neighboring nodes relative to the feature data sequence.

[0020] In one embodiment, the partition association value is determined by the metric distance between any two nodes and the similarity of the associated structural feature vectors therebetween.

[0021] In one embodiment, a sum of the similarity and a preset value greater than 0 is calculated, and a ratio of the metric distance between any two nodes and the sum is used as a partition association value between the any two nodes.

[0022] In one embodiment, the segmentation process is:

[0023] For each cluster, obtain the minimum and maximum coordinate values ​​of all nodes in the cluster in the X, Y, and Z directions to obtain the spatial range of the cluster. , , ,in, , , are the minimum coordinate values ​​of the cluster in the X, Y, and Z directions, respectively. , , are the maximum coordinate values ​​of the cluster in the X, Y, and Z directions respectively;

[0024] The axis-aligned bounding box of the target model is constructed according to the GLB data, and the axis-aligned bounding box of the target model is segmented according to the spatial range of each cluster.

[0025] In one embodiment, the classification method is: counting the number of nodes in each subspace obtained by segmenting the axis-aligned bounding box, and classifying the subspaces in descending order according to the number of nodes.

[0026] In a second aspect, an embodiment of the present application also provides a model data format conversion system for a three-dimensional engine, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of any one of the above-mentioned model data format conversion methods for a three-dimensional engine when executing the computer program.

[0027] This application has at least the following beneficial effects:

[0028] In view of the fact that the complexity characteristics of the three-dimensional spatial structure of the target model in the process of model data format conversion have a great influence on the accuracy of data segmentation and classification, which may cause unnatural fractures or deformations of the model at the segmentation point, the application analyzes the characteristics of differences in the distribution and complexity of geometric shapes between different key structures of the three-dimensional data of the target model, performs local analysis on the coordinate information, geometric information and texture information in the GLB format file of the three-dimensional data of the target model, obtains the structural association value of each node, characterizes the local area characteristics of the local area where the node is located, fully considers the comprehensive information in the local area where the node is located, and facilitates the subsequent analysis of the association between different nodes based on the similarity between the local area characteristics of the local areas where different nodes are located;

[0029] Furthermore, based on the similarity between the local area features of the local areas where different nodes are located, the relative difference value of each node is obtained, and the partition association value between the nodes is further obtained. The local area features of the local area where the node is located, as well as the difference in local area features between the node and the local area where the neighboring nodes are located, can be comprehensively considered to accurately analyze the association features between the nodes of the target model, thereby improving the accuracy of analyzing whether the nodes are components of the same key structure, thereby improving the accuracy of data segmentation and classification during the model data format conversion process, and reducing the impact of data segmentation effects on the model visual effects and real-time rendering performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0031] Figure 1 A flowchart of a method for converting a model data format for a three-dimensional engine provided by an embodiment of the present application;

[0032] Figure 2 It is a schematic diagram of the process of determining the structure association value;

[0033] Figure 3 Schematic diagram of the process of determining the partition association value. DETAILED DESCRIPTION

[0034] In the description of the embodiments of the present application, words such as "exemplary", "or", "for example" and the like are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary", "or", "for example" and the like is intended to present related concepts in a concrete manner.

[0035] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art in the present application. The terms used in the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. It should be understood that, unless otherwise specified, " / " means or.

[0036] It should also be noted that the terms "first" and "second" in the present application are used to distinguish similar objects rather than to describe a specific order or sequence.

[0037] The specific scheme of the model data format conversion method and system for a three-dimensional engine provided by the present application is described in detail below with reference to the accompanying drawings.

[0038] See also Figure 1 , which shows a flowchart of a method for converting a model data format for a three-dimensional engine provided by an embodiment of the present application, the method comprising the following steps:

[0039] Step 1, obtain the GLB format file of the target model's three-dimensional data, and parse it to obtain the node coordinates, the geometric data and texture data contained in the binary buffer.

[0040] The model data format conversion used for the 3D engine in this application is to convert the target model 3D data in the 3D design software PDMS into the scene data of the real-time 3D engine. Usually, in the process of converting the target model 3D data in the 3D design software, the segmentation and classification effect of the data will directly affect the rendering efficiency and performance in the real-time 3D engine. Since the target model 3D data for data segmentation and classification is a GLB format file, the GLB format file of the target model 3D data exported from the 3D design software is obtained. The specific acquisition process is:

[0041] The RVM format file and TXT attribute file of the target model's three-dimensional data are exported from the tree database of the 3D design software PDMS project, wherein the RVM format file contains the geometric shape parameters of the target model, and the TXT attribute file includes the material, color, and structure of each component of the target model; the matching and association results between the RVM format file and the TXT attribute file are obtained using the Autodesk Knowledge Network, and the RVM format file and the TXT attribute file after matching and association are combined and converted into a GLB format file using Aspose.3D, and the GLB file contains the coordinates, geometry, and texture information of the target model's three-dimensional data. Among them, the matching and association process between the RVM format file and the TXT attribute file, and the conversion process of Aspose.3D are both well-known technologies, and this application will not repeat them.

[0042] In order to facilitate the subsequent analysis of the characteristics of the target model's three-dimensional data, Assimp is first used to parse the GLB file of the target model's three-dimensional data to obtain the vertex information in the JSON data part of the GLB file and the geometric data and texture data contained in the binary buffer. It should be noted that the GLB file mainly consists of two parts: the JSON data part and the binary buffer; among them, the vertex information includes node coordinate information, normals, and indices, etc.

[0043] Step 2: determine the structural association value of each node through the coordinate information, geometric information and texture information in the GLB format file of the target model three-dimensional data, and further determine the partition association value between different nodes.

[0044] Generally speaking, the steps of converting the model data of 3D design software into the scene data of real-time 3D engine mainly include: exporting the target model data file, associating and matching the model data, converting the GLB file, segmenting and grading the data through the GLB file, and rendering after the data segmentation and grading. Among them, analyzing the coordinates, geometry and texture information of the target model 3D data stored in the GLB file to achieve accurate segmentation and grading is the key to ensuring the performance of real-time rendering and retaining the visual effect of the original design.

[0045] Since different key structures of the target model are not exactly the same, there are differences in the node distribution and geometric complexity between the corresponding areas of different key structures of the target model in the three-dimensional data. In the process of data segmentation of GLB files based on spatial range, the composition and distribution of the key structures of the target model affect the effect of data segmentation, which may cause unnatural breaks or deformations at the segmentation point, affecting the visual effect. Therefore, this application analyzes the correlation features between the data in the GLB file to make the data segmentation results more consistent with the structure of the target model. The specific analysis process is as follows:

[0046] Step 2.1, based on the position distribution of nodes in the local area where each node is located, the difference in geometric data and the difference in texture data between nodes in the local area, determine the structural association value of each node.

[0047] All nodes in the GLB file of the target model 3D data are grouped into a node information set, a search sphere with a radius of R is constructed with each node in the node information set as the center, and all nodes within the search sphere of each node are grouped into a neighbor node set;

[0048] In this embodiment, the value of R is 10. The value of R is preset manually and can be set by the implementer. This application does not impose any special restrictions.

[0049] Calculate the mean of the Euclidean distances between all arbitrary two nodes in the neighbor node set of each node, recorded as the first distance mean. The smaller the first distance mean is, the denser the distribution of nodes in the local area where each node is located may be. Calculate the mean of the Hamming distance of geometric data and the mean of the Hamming distance of texture data between all arbitrary two nodes in the neighbor node set of each node, recorded as the first mean and the second mean, respectively. The mean of the first mean and the second mean of each node is recorded as the second distance mean. The smaller the second distance mean is, the smaller the difference in geometric features and texture features of the target model in the local area where each node is located is, and the more likely the local area where each node is located is to belong to the same key structure of the target model.

[0050] Based on the analysis results of the local distribution characteristics, geometric feature differences and texture feature differences within the search sphere of each node, the structural association value of each node is determined, and the expression is: ; In the formula, Represents the structural association value of the i-th node; represents the first distance mean of the i-th node; represents the mean of the second distance of the i-th node; β represents a preset value greater than 0, the purpose of which is to avoid the denominator being 0. The value of β is preset manually and can be set by the implementer. In this embodiment, the value of β is 0.01.

[0051] It should be noted that: the more concentrated the nodes are in the local area where the i-th node is located, and the higher the consistency of the nodes in the local area in terms of geometry and texture, the greater the structural association value. The schematic diagram of the process of determining the structural association value is as follows: Figure 2 shown.

[0052] Step 2.2, based on the difference in structural association values ​​between each node and the nodes in the local area, as well as the distance between them, obtain the relative feature data sequence of each node, and based on the difference in relative feature data sequence between each node and the nodes in the local area, determine the relative difference value of each node.

[0053] The structural association value of a node can characterize the node distribution characteristics, geometry and texture consistency of the local area where the node is located. In the process of data segmentation within the spatial range, if the different key structures of the target model are distinguished only by the node spatial distance, the nodes belonging to the same key structure may be divided into different subspaces, which may cause unnatural breaks or deformations at the segmentation point and affect the visual effect. Therefore, the local area characteristics of different nodes can be relatively analyzed. The specific analysis process is as follows:

[0054] Calculate the difference in structural association value between any node in the node information set and each node in its neighboring node set, as well as the Euclidean distance between them, recorded as the reference distance, and arrange all the differences in order from small to large according to the reference distance to form a relative feature data sequence of any node;

[0055] Based on the difference in the relative feature data sequence between each node and the nodes in its neighboring node set, the relative difference value of each node is determined. The relative difference value is obtained by combining the difference in local area features between each node and the local area where the neighboring nodes are located, which accurately reflects the correlation features between nodes when the GLB data of the target model is segmented. The expression is:

[0056] ; In the formula, Represents the relative difference value of the i-th node; and They represent the relative feature data sequences of the ith and tth nodes respectively, Dis() represents the distance function; n represents the number of nodes in the neighboring node set of the ith node.

[0057] In this embodiment, the distance between relative feature data sequences is the DTW distance. As other implementation methods, based on the measurable distance between sequences, the implementer may adopt other existing technologies to obtain the distance between sequences, such as Euclidean distance, Manhattan distance, etc., and this application does not impose any special restrictions.

[0058] It should be noted that: the smaller the distance between the relative feature data sequences, the more similar the local area features between the i-th node and the local area where the neighboring nodes are located are, the more likely it is that the i-th node and the neighboring nodes are a combined association structure, and the smaller the relative difference value is.

[0059] Step 2.3, the structural association value and relative difference value of each node are combined into an association structural feature vector of each node, and based on the distance between any two nodes and the similarity of the association structural feature vectors between them, the partition association value between the any two nodes is determined.

[0060] In order to avoid large segmentation errors of the key structures of the target model in the process of segmenting GLB data based on spatial range, the nodes of the same key structure are segmented into different subspaces, or the nodes of different key structures are segmented into the same subspace. The similarity of local area features and the distance between different nodes are comprehensively considered to determine the partition association value between nodes. The expression is:

[0061] ; In the formula, Represents the partition association value between the i-th and v-th nodes; and Respectively represent the associated structural feature vectors of the i-th and v-th nodes; dp() represents the cosine similarity function; represents the Euclidean distance between the i-th and v-th nodes; γ represents a preset value greater than 0, the purpose of which is to avoid the denominator being 0. The value of γ is preset manually and can be set by the implementer. In this embodiment, the value of γ is 0.01; wherein the associated structural feature vector of each node is composed of the structural association value and the relative difference value of each node.

[0062] It should be noted that the smaller the partition correlation value is, the greater the possibility that the i-th and v-th nodes are components of the same key structure of the target model. The flow chart for determining the partition correlation value is as follows: Figure 3 shown.

[0063] Step 3: cluster all nodes based on the partition association values, segment and classify the GLB data based on the clustering results, and use a three-dimensional engine to reconstruct the scene of the target model.

[0064] A density peak clustering algorithm is used to cluster all nodes in the node information set to obtain clusters, wherein the partition association value between nodes is used as a distance metric. The density peak clustering algorithm is a well-known technology and will not be described in detail in this application.

[0065] Furthermore, based on the clustering results of all nodes in the node information set, the GLB data is segmented and classified within the spatial range; the specific process is as follows:

[0066] For each cluster, obtain the minimum and maximum coordinate values ​​of all nodes in the cluster in the X, Y, and Z directions to obtain the spatial range of the cluster. , , ,in, , , are the minimum coordinate values ​​of the cluster in the X, Y, and Z directions, respectively. , , are the maximum coordinate values ​​of the cluster in the X, Y, and Z directions respectively;

[0067] An axis-aligned bounding box of the target model is constructed according to the GLB data, and the axis-aligned bounding box of the target model is segmented according to the spatial range of each cluster to obtain each subspace, thereby segmenting the GLB data; wherein, the method for constructing the axis-aligned bounding box is a well-known technology and will not be described in detail in this application;

[0068] Count the number of nodes in each subspace, and classify the subspaces in descending order according to the number of nodes, and assign a unique label to each subspace; for example, if the number of nodes in a subspace is 5, then after sorting according to the number of nodes, the label of the subspace is 5-1, where 5 is the number of nodes in the subspace; 1 is the level, which represents the importance of the subspace corresponding to the GLB data. The lower the level, the lower the importance.

[0069] Furthermore, the results of segmentation and grading of the target model's GLB data are used as input, and the i3D Act real-time 3D engine is used to reconstruct the scene of the target model, specifically including: placing the converted data at the correct position and level, setting the lighting and material of the scene, ensuring that the appearance of the model in the engine meets expectations, and adjusting rendering settings as needed, such as anti-aliasing, shadow quality, post-processing effects, etc. The i3D Act real-time 3D engine for scene reconstruction is a well-known technology and will not be elaborated in this application.

[0070] Based on the same inventive concept as the above method, an embodiment of the present application also provides a model data format conversion system for a three-dimensional engine, including a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, the steps of any one of the above-mentioned model data format conversion methods for a three-dimensional engine are implemented.

[0071] In summary, the present application aims at the problem that the complexity characteristics of the three-dimensional spatial structure of the target model in the process of model data format conversion have a great influence on the accuracy of data segmentation and classification, which may cause unnatural fractures or deformations of the model at the segmentation point. By analyzing the characteristics of differences in the distribution and complexity of geometric shapes between different key structures of the three-dimensional data of the target model, the coordinate information, geometric information and texture information in the GLB format file of the three-dimensional data of the target model are locally analyzed to obtain the structural association value of each node, characterize the local area characteristics of the local area where the node is located, and fully consider the comprehensive information in the local area where the node is located, so as to facilitate the subsequent analysis of the association between different nodes according to the similarity between the local area characteristics of the local areas where different nodes are located;

[0072] Furthermore, based on the similarity between the local area features of the local areas where different nodes are located, the relative difference value of each node is obtained, and the partition association value between the nodes is further obtained. The local area features of the local area where the node is located, as well as the difference in local area features between the node and the local area where the neighboring nodes are located, can be comprehensively considered to accurately analyze the association features between the nodes of the target model, thereby improving the accuracy of analyzing whether the nodes are components of the same key structure, thereby improving the accuracy of data segmentation and classification during the model data format conversion process, and reducing the impact of data segmentation effects on the model visual effects and real-time rendering performance.

[0073] The flowchart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to the embodiment of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. In some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. In the description corresponding to the flowchart and the block diagram in the accompanying drawings, the operations or steps corresponding to different boxes can also occur in a different order from the order disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by a dedicated hardware-based system that performs the specified functions or actions, or may be implemented by a combination of dedicated hardware and computer instructions.

[0074] It is obvious to those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and that the present application can be implemented in other specific forms without departing from the basic features of the present application. Therefore, no matter from which point of view, the above embodiments of the present application should be regarded as exemplary and non-restrictive.

Claims

1. A model data format conversion method for a three-dimensional engine, characterized in that: The method comprises the following steps: Get the GLB format file of the target model's 3D data, and parse it to get the node coordinates, the geometric data and texture data contained in the binary buffer; Determine the structural association value of each node based on the position distribution of nodes in the local area where each node is located, the difference in geometric data between nodes in the local area, and the difference in texture data; Based on the difference in structural association values ​​between each node and the nodes in the local area where it is located, and the distance between each node and the nodes in the local area where it is located, a relative feature data sequence of each node is obtained, and based on the difference in relative feature data sequence between each node and the nodes in the local area where it is located, a relative difference value of each node is determined; The structural association value and the relative difference value of each node are combined into an association structural feature vector of each node, and based on the distance between any two nodes and the similarity of the association structural feature vectors between any two nodes, the partition association value between the any two nodes is determined; Clustering all nodes based on the partition association values, segmenting and grading the GLB data based on the clustering results, and reconstructing the scene of the target model using a three-dimensional engine; The process of determining the structure association value is as follows: A search sphere with a preset radius is constructed with any node as the center, and all nodes in the search sphere are composed of a set of neighboring nodes of the node; Calculate the mean of the measured distances between any two nodes in the neighbor node set of each node, recorded as the first distance mean; Calculate the mean of the metric distance of the geometric data between any two nodes in the neighbor node set of each node, and the mean of the metric distance of the texture data between any two nodes in the neighbor node set of each node, record them as the first mean and the second mean respectively, and record the mean of the first mean and the second mean of each node as the second distance mean; Determining a structural association value of each node based on the first distance mean and the second distance mean; The smaller the first distance mean is, the denser the distribution of nodes in the local area where each node is located is.

2. The model data format conversion method for a three-dimensional engine according to claim 1, characterized in that: The process of determining the structural association value is: calculating the product of the first distance mean and the second distance mean, and taking the reciprocal of the sum of the product and a preset value greater than 0 as the structural association value of each node.

3. The model data format conversion method for a three-dimensional engine according to claim 1, characterized in that: The method for obtaining the relative characteristic data sequence is: Calculate the difference in structural association value between any node and each node in its neighboring node set, as well as the metric distance between them, recorded as the comparison distance, and arrange all the differences in order from small to large according to the comparison distance to form a relative feature data sequence of any node.

4. The model data format conversion method for a three-dimensional engine according to claim 1, characterized in that: The relative difference value is the average of the metric distances between each node and all nodes in the set of its neighboring nodes relative to the feature data sequence.

5. The model data format conversion method for a three-dimensional engine according to claim 1, characterized in that: The partition association value is determined by the obtained metric distance between any two nodes and the obtained similarity of the association structure feature vectors between any two nodes.

6. The model data format conversion method for a three-dimensional engine according to claim 5, characterized in that: The sum of the similarity and a preset value greater than 0 is calculated, and the ratio of the metric distance between the arbitrary two nodes and the sum is used as the partition association value between the arbitrary two nodes.

7. The model data format conversion method for a three-dimensional engine according to claim 1, characterized in that: The segmentation process is as follows: For each cluster, obtain the minimum and maximum coordinate values ​​of all nodes in the cluster in the X, Y, and Z directions to obtain the spatial range of the cluster. , , ,in, , , are the minimum coordinate values ​​of the cluster in the X, Y, and Z directions, respectively. , , are the maximum coordinate values ​​of the cluster in the X, Y, and Z directions respectively; The axis-aligned bounding box of the target model is constructed according to the GLB data, and the axis-aligned bounding box of the target model is segmented according to the spatial range of each cluster.

8. The model data format conversion method for a three-dimensional engine according to claim 7, characterized in that: The classification method is: counting the number of nodes in each subspace obtained by segmenting the axis-aligned bounding box, and classifying the subspaces in descending order according to the number of nodes.

9. A model data format conversion system for a three-dimensional engine, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the model data format conversion method for a three-dimensional engine as described in any one of claims 1-8 are implemented.

Images