A modeling method, system, device and storage medium based on dynamic 3D target

Abstract

The application relates to the technical field of three-dimensional modeling, in particular to a modeling method and system based on dynamic 3D targets, an electronic device and a storage medium, the method comprising: generating a three-dimensional modeling operation instruction, wherein the operation instruction at least contains an operation type, a target object identifier and a space scope parameter; constructing an operation semantic tree based on the operation instruction; detecting a space scope overlapping area by comparing a local operation semantic tree with a received remote operation semantic tree; applying a space-time dimension priority merging rule when the overlapping area is detected; executing the merged operation instruction and generating a binary log containing an operation semantic tree hash value; and performing a local rollback based on a differential CRC check value when an operation conflict is detected, wherein the differential CRC value is only calculated for the overlapping area data block. The application effectively solves the problems of semantic understanding loss, logical inconsistency and low recovery efficiency during three-dimensional space operation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of three-dimensional modeling, in particular to a modeling method and system based on dynamic 3D targets, electronic equipment and storage medium. BACKGROUND

[0002] Dynamic 3D targets may involve real-time updates, multi-user collaboration, data synchronization, network optimization, etc. Improvements in communication logic may include data transmission protocols, state synchronization mechanisms, compression algorithms, distributed processing, etc.

[0003] In a three-dimensional modeling collaboration system, the problem of data conflict caused by multi-user concurrent operation has long existed. Traditional solutions such as operation transformation (OT) algorithm rely on linear timestamp sequence for conflict resolution, but have inherent defects when dealing with three-dimensional space operations: such as semantic understanding missing (simplifying three-dimensional space operations to text editing operations, unable to recognize the spatial semantic correlation of modeling operations such as rotation and stretching), logical inconsistency (merging simply based on operation time sequence, resulting in space operation results violating physical laws, such as object penetration), low recovery efficiency (using full-quantity log rollback mechanism, producing up to GB level of redundant data in complex model scenarios).

[0004] Therefore, there is an urgent need for a client-side solution to the problems of semantic understanding missing, logical inconsistency and low recovery efficiency in three-dimensional space operations. SUMMARY

[0005] To solve at least one of the above technical problems, the present application provides a modeling method and system based on dynamic 3D targets, electronic equipment and storage medium.

[0006] In a first aspect, the present application provides a modeling method based on dynamic 3D targets, which adopts the following technical solution:

[0007] Generating a three-dimensional modeling operation instruction, the operation instruction at least containing an operation type, a target object identifier and a spatial scope parameter;

[0008] Building an operation semantic tree based on the operation instruction, the operation semantic tree nodes containing the following hierarchical structure,

[0009] Parent node, three-dimensional space topological relationship of target object,

[0010] Child node, spatial scope weight value of user operation;

[0011] By comparing the local operation semantic tree with the received remote operation semantic tree, detecting the spatial scope overlap area;

[0012] When detecting the overlap area, applying the space-time dimension priority merging rule, specifically as follows,

[0013] time dimension priority condition, when the operation initiation time difference is less than a threshold value Δt, the spatial dimension determination is activated;

[0014] spatial dimension priority condition, priority coefficients are assigned according to operation types, wherein the priority of a structure modification operation P1> the priority of a geometric deformation operation P2> the priority of a material editing operation P3;

[0015] the merged operation instruction is executed, and a binary log containing the operation semantic tree hash value is generated;

[0016] when an operation conflict is detected, a local rollback is performed based on a differential CRC check value, and the differential CRC value is only calculated for data blocks in the overlapping area.

[0017] By the above scheme, the operation semantic tree is constructed, the spatial context relationship graph of three-dimensional operations is constructed by using parent node topology coding, child node scope quantization and semantic conflict detection, so that the system can understand the semantic association of the composite operation, and the semantic recognition accuracy is improved compared with the traditional method. At the same time, the time-space dimension priority merging rule is used, the time condition trigger, the physical constraint injection and the conflict resolution algorithm are used, and the physical abnormality rate in a specific scene is greatly reduced compared with the traditional OT algorithm, and the correct merging probability that meets the rigid body dynamics is improved. In addition, through the differential CRC local rollback, the spatial locality utilization, data compression optimization and the use of fast recovery protocol are enhanced, the data amount of single rollback is reduced, and the recovery time is shortened. In summary, through the organic integration of the three technical means of three-dimensional space semantic structuring, operation priority quantization and recovery mechanism localization, a real-time collaboration framework with topology perception ability is formed, and the problems of semantic understanding loss, logical inconsistency and low recovery efficiency in three-dimensional space operation are effectively solved.

[0018] In a possible implementation manner, in the step of constructing an operation semantic tree based on the operation instruction, the following steps are included:

[0019] the three-dimensional spatial topology relationship of the parent node is recorded by using a half-edge data structure to record the connection relationship of vertices-edges-surfaces;

[0020] the spatial scope weight value of the child node is calculated by using a Gaussian attenuation function, and the action radius σ is positively correlated with the preset influence range of the operation type, wherein,

[0021] the action radius σ of the structure modification operation is 20% of the diagonal length of the model bounding box;

[0022] the action radius σ of the geometric deformation operation is 150% of the radius of the operation tool;

[0023] the action radius σ of the material editing operation is 30% of the UV unfolding area of the texture map.

[0024] Through the above scheme, in the building model test, the action range recognition accuracy is improved, and the geometric precision of semantic understanding is effectively improved.

[0025] In a possible implementation manner, in the step of applying the space-time dimension priority merging rule when the overlapping area is detected, the priority coefficient is dynamically adjusted according to the model type, and the dynamic adjustment includes the following:

[0026] In the mechanical assembly model, P1:P2:P3=4:3:1;

[0027] In the organic biological model, the weight of P2 is increased by 30%;

[0028] In the building scene model, the priority of the light operation P4 is increased to 0.8*P1;

[0029] The adjustment is based on the classification label in the model file metadata.

[0030] Through the above scheme, the correct merging rate is improved in the part assembly scene, and the logical consistency adaptation ability is enhanced.

[0031] In a possible implementation manner, in the step of performing local rollback based on the differential CRC check value when the operation conflict is detected, the local rollback includes the following:

[0032] Locating the subtree node to be rolled back based on the operation semantic tree;

[0033] Generating an index list V_list of affected vertices, the list being sorted according to a spatial Morden code;

[0034] Only the vertices in V_list are executed for coordinate rollback calculation.

[0035] Through the above scheme, the rollback data volume is further reduced, and the recovery data volume is reduced.

[0036] In a possible implementation manner, before the step of detecting the spatial scope overlapping area by comparing the local operation semantic tree with the received remote operation semantic tree, the method further includes a preloading step:

[0037] According to the current operation tool type of the user, the model area that is likely to be modified is predicted;

[0038] The spatial topology data of the area is pre-pulled from the server;

[0039] The cache data is organized in a quadtree structure, and the cache hit rate is more than 85%.

[0040] Through the above scheme, the operation response time standard deviation is reduced, and the network delay influence is reduced.

[0041] In a possible implementation, the step of applying the spatio-temporal dimension priority merging rule when the overlapping region is detected further includes:

[0042] Collecting historical operation conflict data to train an LSTM prediction model;

[0043] Proactively limiting non-critical operations when a high conflict probability (> 70%) is detected;

[0044] Dynamically adjusting the Δt threshold value: Δt = base value 50 ms x (1 - conflict probability).

[0045] Through the above scheme, the proactive conflict avoidance rate is improved, the post-processing pressure is reduced, and preventive conflict avoidance is achieved.

[0046] In a possible implementation, in the step of executing the merged operation instruction and generating a binary log containing an operation semantic tree hash value, the binary log adopts differential encoding compression, including:

[0047] Dictionary encoding is established for the operation type field, and 1 byte replaces the original 4-byte enumeration value;

[0048] The spatial coordinates are stored in the form of Δ values relative to the previous operation;

[0049] The DEFLATE algorithm is used for streaming compression;

[0050] The compressed log format is,

[0051] | header identification (0xAE) | compression flag | dictionary encoding | Δ coordinate value | CRC check |.

[0052] Through the above scheme, the log volume is further reduced, and the storage and transmission overhead is reduced.

[0053] In a second aspect, the application provides a modeling system based on dynamic 3D targets, including:

[0054] An operation instruction generation module configured to receive user input and generate three-dimensional modeling operation instructions, the instructions including at least an operation type field, a target object unique identifier, and a three-dimensional space scope parameter set;

[0055] A semantic tree construction module connected to the operation instruction generation module, including,

[0056] A topology relationship analysis unit that extracts the three-dimensional spatial topology relationship of the target object based on a half-edge data structure (Half-edge) and generates parent node data;

[0057] A scope calculation unit that calculates the weight distribution matrix of the child nodes according to the spatial influence algorithm corresponding to the operation type;

[0058] Conflict detection module: bidirectionally connected with the local storage module and the network communication module, including,

[0059] Semantic tree comparator: performs spatial scope overlap detection on the local operation semantic tree and the received remote semantic tree;

[0060] Priority arbiter: built-in spatiotemporal dimension judgment logic circuit, which triggers arbitration process when overlap is detected;

[0061] Operation merging and executing module: connected with the output end of the conflict detection module, including,

[0062] Rule base: stores a spatiotemporal priority coefficient reference table, wherein structure modification P1, geometric deformation P2, and material editing P3 satisfy P1>P2>P3;

[0063] Difference calculation unit: generates an operation delta package containing only vertex changes in the overlap region;

[0064] Log management module: communicatively connected with all modules, including,

[0065] Binary encoder: converts operation records into compressed logs containing semantic tree hash values;

[0066] Local rollback controller: in response to a conflict signal, performs coordinate rollback of affected vertices based on a differential CRC check value;

[0067] Network communication module: adopts a double-channel design, including,

[0068] Operation instruction transmission channel: encapsulates semantic tree data packets;

[0069] Log synchronization channel: transmits compressed binary difference logs;

[0070] The semantic tree construction module, the conflict detection module, and the operation merging and executing module are interconnected through a shared memory bus at the hardware layer to form a real-time processing pipeline architecture.

[0071] In a third aspect, the present application provides an electronic device, which comprises a memory and a processor, wherein the memory is configured to store computer program code, and the processor is configured to execute the computer program code stored in the memory to implement the method in the first aspect and any one of the possible implementation manners of the first aspect, or the method in the second aspect and any one of the possible implementation manners of the second aspect.

[0072] In a fourth aspect, the present application provides a computer readable storage medium, which stores a computer program or instructions, when the computer program or instructions are executed, the method in the first aspect and any possible implementation of the first aspect, or the second aspect and any possible implementation of the second aspect is implemented.

[0073] To sum up, the present application includes at least one of the following beneficial technical effects:

[0074] 1. Conflict resolution accuracy is improved, spatial semantic analysis is realized by operating semantic tree, conflict misjudgment rate is effectively reduced compared with traditional method, and correct merging probability meeting physical law is greatly improved by using time and space dimension priority rules;

[0075] 2. System performance is optimized, local rollback mechanism reduces the amount of recovered data to 1 / 2 of full rollback, binary log compression rate is obviously improved compared with JSON format, and storage overhead is reduced;

[0076] 3. Real-time performance is guaranteed, end-side processing delay is effectively controlled, operation continuity can still be maintained in network fluctuation scenarios (RTT 100-500 ms), and user perception of lag rate is obviously reduced. BRIEF DESCRIPTION OF DRAWINGS

[0077] Figure 1 A flowchart of a modeling method based on a dynamic 3D target provided by an embodiment of the present application.

[0078] Figure 2 A structural schematic diagram of a modeling system based on a dynamic 3D target provided by an embodiment of the present application.

[0079] Figure 3 A structural schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0080] The technical solutions in the present application will be described below with reference to all the accompanying drawings. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments.

[0081] In the description of the embodiments of the present application, unless otherwise specified, " / " represents the meaning of "or", for example, A / B can represent A or B; "and / or" in this document is only a description of the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which means that there are three cases of A alone, A and B together, and B alone. In addition, in the description of the embodiments of the present application, "plurality" or "multiple" means two or more than two.

[0082] Hereinafter, the terms "first", "second", "third", etc. are used only for the purpose of description, and should not be understood as indicating or implying relative importance or implying a specific number of the technical features indicated. Thus, the features defined with "first", "second", etc. can explicitly or implicitly include one or more of the features. In the description of the embodiments, the meaning of "a plurality of" is two or more, unless otherwise specified.

[0083] The terms used in the following embodiments are only for the purpose of describing specific embodiments, and are not intended to be limiting on the present application. As used in the specification and the appended claims of the present application, the singular forms "a", "an" and "the" are intended to include, for example, the expression "one or more", unless the context clearly indicates otherwise. It should also be understood that "at least one", "one or more" in the following embodiments of the present application means one, two or more.

[0084] In the present specification, the reference to "one embodiment" or "some embodiments" and the like means that a particular feature, structure or characteristic described in connection with the embodiment is included in at least one embodiment of the present application. Thus, the appearance of the phrases "one embodiment", "some embodiments", "another embodiment", "additional embodiments" and the like in various places in the specification is not necessarily all referring to the same embodiment, but means "one or more but not all embodiments", unless otherwise specifically noted. The terms "comprise", "include", "have" and their conjugates mean "including but not limited to", unless otherwise specifically noted.

[0085] The embodiments of the present application provide a method for modeling based on dynamic 3D targets, which is executed by an electronic device. The electronic device can be a standalone physical electronic device, an electronic device cluster or distributed system composed of multiple physical electronic devices, or a cloud electronic device providing cloud computing services. The embodiments of the present application do not limit here, as shown in the following figure, the method comprises: Figure 1

[0086] S1, generating a three-dimensional modeling operation instruction.

[0087] Specifically, the operation instruction at least contains operation type, target object identifier and spatial scope parameter.

[0088] Further, the operation type is defined as three types of basic operation codes preset by the system, for example: 0x01 = structure modification, such as cutting, drilling, corresponding to high priority operation; 0x02 = geometric deformation, such as stretching, bending, corresponding to medium priority operation; 0x03 = material editing, such as mapping, coloring, corresponding to low priority operation. ​

[0089] Among them, the operation type is automatically mapped by the tool type, such as selecting "cutting tool" to be mapped as "structure modification".

[0090] Further, the target object identification adopts a globally unique identifier (UUIDv4) to mark the model components, and a globally unique identification code is generated for each model component, using a 36-bit hybrid code (containing numbers, letters and separators).

[0091] Among them, the identification code contains a generation timestamp, a device fingerprint and a random entropy value, ensuring global uniqueness.

[0092] Further, the spatial scope parameter collection is performed by a three-dimensional picking algorithm to accurately locate the operation center point coordinates (x, y, z), and the action radius is automatically calculated according to the tool attributes, such as brush radius and stretching range.

[0093] Among them, the action radius of the structural tool (such as a drill bit) = physical diameter x 1.2 times safety factor; the action radius of the surface treatment tool (such as a spray gun) = preset parameter x operation pressure value.

[0094] In the embodiments of the present application, the user operation is converted into a machine-understandable standardized instruction, and each operation is given a unique identity and spatial positioning information, like creating a "digital identity card" for modeling operation, achieving a significant effect of improving instruction generation efficiency. In a standard workstation environment, the single operation instruction generation time is not more than 3 milliseconds, and the instruction generation time is significantly reduced compared with the traditional scheme.

[0095] S2, construct an operation semantic tree based on the operation instruction.

[0096] Specifically, the operation semantic tree node includes the following hierarchical structure: parent node, three-dimensional spatial topological relationship of the target object; child node, spatial scope weight value of the user operation.

[0097] Further, the implementation steps of the parent node construction (topological relationship) are as follows:

[0098] Extract the three-dimensional mesh topological data of the target component from the model database, record the spatial connection relationship using a "vertex-hemisphere-face" circular linked list structure, construct a local topological subgraph, and only retain elements within 1.5 times the operation influence radius.

[0099] Further, the implementation steps of the child node calculation (scope weight) are as follows:

[0100] Calculate the degree of influence of each vertex based on the distance attenuation principle. Among them, the operation center is taken as the origin, the closer the vertex, the higher the weight value, the attenuation curve follows the Gaussian distribution law, and the weight value interval is [0, 1].

[0101] The attenuation rate parameter is dynamically adjusted to adapt to different operating characteristics. Among them, the structure modification adopts fast attenuation (the influence range is concentrated), and the material editing adopts slow attenuation (the influence range is diffused).

[0102] In the embodiments of the present application, the influence range and model structure of the operation are encoded into a "spatial relationship map" to enable the computer to understand the spatial semantics of the operation.

[0103] S3, by comparing the local operation semantic tree with the received remote operation semantic tree, the spatial scope overlap area is detected.

[0104] Specifically, the implementation steps include fast spatial index construction and overlap determination process.

[0105] Further, the fast spatial index construction includes creating an axis-aligned bounding box (AABB) for each semantic tree, using an octree structure to accelerate spatial queries, and dividing the three-dimensional space into a hierarchical grid.

[0106] Further, the overlap determination process includes,

[0107] First stage: coarse detection - compare the intersection of bounding boxes (time-consuming <1ms);

[0108] Second stage: fine detection - calculate the overlap rate of the weight heat map (time-consuming 5-15ms);

[0109] Determination threshold setting: when the overlap vertex ratio exceeds 70%, trigger conflict processing.

[0110] In the embodiments of the present application, the operation conflict area is identified through spatial relationship comparison, similar to the collision detection mechanism of radar scanning, a multi-level cache mechanism is realized, the cache reuse rate of the latest comparison result is improved, and the repeated calculation overhead is greatly reduced.

[0111] S4, when the overlap area is detected, the space-time dimension priority merging rule is applied.

[0112] Specifically, the space-time dimension priority merging rule is as follows:

[0113] Time dimension priority condition: when the operation initiation time difference is less than the threshold Δt, the spatial dimension determination is activated.

[0114] Spatial dimension priority condition: according to the operation type, a priority coefficient is assigned, wherein the structure modification operation priority P1> geometric deformation operation priority P2> material editing operation priority P3.

[0115] Further, the time dimension determination step is to set a dynamic time window threshold Δt, which is adaptively adjusted according to network delay, specifically Δt = 50 ms in a 5G network environment and Δt = 100 ms in a general broadband environment. When the operation time difference is less than Δt, the spatial priority arbitration is activated.

[0116] Further, the spatial priority arbitration step is to set a basic priority coefficient, the “structure modification” priority coefficient is 3, and the physical constraint strength is rigid constraint; the “geometry deformation” priority coefficient is 2, and the physical constraint strength is elastic constraint; the “material editing” priority coefficient is 1, and the physical constraint strength is no constraint. The scene adaptation is enhanced, the structure modification coefficient is × 1.5 times in a mechanical assembly scene, and the geometry deformation coefficient is × 1.2 times in a character modeling scene.

[0117] In the embodiments of the present application, by establishing a multi-dimensional decision model, the operation instructions of the time and space conflict are intelligently coordinated, a two-level arbitration mechanism is formed, first, operations with significant time difference are excluded, and then physical rationality evaluation is performed on operations with close time and space, to ensure the integrity of the model structure.

[0118] S5, execute the merged operation instructions, and generate a binary log containing the operation semantic tree hash value.

[0119] Specifically, the implementation steps of generating the binary log are to design a log structure and a storage optimization strategy.

[0120] Further, the log structure design includes,

[0121] Header identification: 2-byte magic number (0xAE01 identifies operation log);

[0122] Semantic tree fingerprint: 32-byte SHA-256 hash value;

[0123] Operation parameter segment: variable-length byte storage coordinate change amount;

[0124] Check code: 4-byte CRC32 cyclic redundancy check.

[0125] Further, the storage optimization strategy includes,

[0126] Incremental record: only store the difference value (Δ value) of the vertex coordinates;

[0127] Compression encoding: using streaming compression algorithm, real-time compression rate ≥ 80%;

[0128] Block storage: 1024 operation records are packaged into one data block.

[0129] In the embodiments of the present application, a high-efficiency access log format is designed, the whole life cycle management of operation history is realized, and the log writing speed reaches 120,000 pieces per second, and the reading delay is stabilized within 5 milliseconds.

[0130] S6, when detecting operation conflict, performing local rollback based on differential CRC check value.

[0131] Specifically, the differential CRC value is only calculated for the overlapping area data block.

[0132] Specifically, the implementation steps of the local rollback mechanism include locating the affected area, calculating the difference value, and performing data recovery.

[0133] Further, the step of locating the affected area includes reversing the operation influence path through the semantic tree, generating a list of affected vertex indexes, and sorting by spatial Morton code.

[0134] Further, the step of calculating the difference value includes recording the vertex coordinate snapshot before rollback, calculating the cyclic redundancy check code (CRC) of the coordinate value, and generating the difference check code.

[0135] Further, the step of performing data recovery includes locating the vertex to be recovered according to the index list, applying the ΔCRC difference value to restore the original coordinate value, locally redrawing the model, and avoiding global refresh.

[0136] In the embodiments of the present application, the data recovery is accurate to the vertex level, the rollback influence range is minimized, the double check mode is adopted, and the data recovery accuracy reaches 99.9999%.

[0137] In some embodiments, the step S2 further includes:

[0138] The three-dimensional spatial topological relationship of the parent node records the connection relationship of vertex-edge-face by using half-edge data structure.

[0139] Specifically, the topological connection relationship of the model grid is recorded by using the half-edge data structure (Half-edge) in computer graphics, and the core elements include that the “vertex” stores three-dimensional coordinates (x, y, z); the “half-edge” records the connection relationship between adjacent faces, forming a bidirectional circular linked list; and the “face” is composed of closed polygons by connecting the first and last half-edges.

[0140] Further, the implementation step of extracting the topological subgraph is to calculate the model enclosing sphere with the operation center as the origin, extract the local topological structure with a radius covering 1.5 times the operation influence range, and only retain the vertex, edge and face connection relationship within the range when constructing the parent node data.

[0141] Further, the spatial scope weight value of the child node is calculated by a Gaussian decay function, and the action radius σ is positively correlated with the preset influence range of the operation type.

[0142] Specifically, the weight calculation formula of the Gaussian decay function is

[0143]

[0144] wherein d is the Euclidean distance from the vertex to the operation center, and the action radius σ is dynamically adjusted according to the operation type.

[0145] Specifically, the structure modification operation σ = 20% of the diagonal line length of the model bounding box. The structure modification needs to cover the whole component, such as cutting the engine cylinder body needs to ensure the integrity of the cross section.

[0146] The geometric deformation operation σ = 150% of the operation tool radius. The physical diffusion effect of the tool force is considered, such as the influence range of a 10 cm diameter carving knife is extended to 15 cm.

[0147] The material editing operation σ = 30% of the UV unfolding area of the texture map. The material operation is unfolded in the UV space, and needs to match the texture distribution density.

[0148] In the embodiments of the present application, through the dual design of structured data coding and spatial influence quantization, the abstract modeling operation is converted into a spatial semantic relationship network that can be accurately calculated, like creating a “spatial DNA map” for the modeling process. The memory occupation is effectively reduced compared with the traditional adjacency matrix, and the topological query speed is improved.

[0149] In some embodiments, in the S4 step, the priority coefficient is dynamically adjusted according to the model type, including:

[0150] In the mechanical assembly model, the priority ratio P1:P2:P3=4:3:1 is set. Among them, the structure modification (such as tolerance adjustment) directly affects the assembly function and needs the highest priority to ensure. For example, in the test of the gearbox model, the correctness rate of solving the gear meshing conflict is effectively improved.

[0151] In the organic biological model, the P2 weight is increased by 30%, such as the original P2=2→after adjustment P2=2.6. Among them, the biological form modeling takes the surface deformation operation as the core (such as muscle line shaping). For example, in the human modeling scene, the surface smoothness compliance rate is effectively improved.

[0152] In the building scene model, the priority of the light operation P4=0.8×P1 is increased. Among them, the light effect and the space structure have a strong coupling relationship (such as the window position affects the lighting). For example, when the user simultaneously modifies the wall structure (P1) and adjusts the angle of the spotlight (P4), the system prioritizes the wall modification, but retains 80% of the light operation weight.

[0153] Further, the adjustment is according to a classification label from the model file metadata.

[0154] Specifically, the implementation step of model type identification is to parse the classification label from the model file metadata.

[0155] Among them, the model type supports three basic classifications: mechanical assembly, industrial components containing precise structure matching relationship (such as engine, gearbox); organic biological model, biological form with complex surface (such as human organs, animal body); architectural scene, spatial structure containing lighting system (such as interior design, urban planning).

[0156] In the embodiments of the present application, a scene adaptive priority model is established according to the physical characteristics and operation requirements of different modeling scenes, so that the merging rule is intelligently adapted to the model type.

[0157] In some embodiments, in the S6 step, the local rollback includes:

[0158] Based on the operation semantic tree, the subtree node to be rolled back is located.

[0159] Specifically, first, conflict impact analysis is performed, such as obtaining the spatial scope parameters of the overlapping area from the conflict detection engine, and matching the corresponding subtree node in the operation semantic tree, which covers 1.2 times the conflict area.

[0160] Further, the subtree retrieval mechanism uses depth-first search (DFS) to traverse the semantic tree nodes, and uses a bounding box fast matching algorithm to locate the target subtree. The center of the subtree bounding box is ≤σ (operation influence radius) × 0.8 away from the center of the conflict area, so that the retrieval time is controlled within 5ms.

[0161] Also included is generating an index list V_list of affected vertices, which is sorted by spatial Morton code.

[0162] Specifically, the principle of Morton code encoding is to convert three-dimensional coordinates (x, y, z) into linear one-dimensional encoding, where the encoding rule is to alternately take the binary bits of each coordinate axis to form a Z-shaped space-filling curve.

[0163] Specifically, the vertex list generation step is to extract all vertex coordinates under the subtree node, calculate the 64-bit Morton code of each vertex, arrange the vertices in ascending order of Morton code, and generate the index list V_list. The vertex data is stored continuously in memory according to spatial proximity, and the cache miss rate when the coordinates are backed up is reduced compared to traditional random access.

[0164] Also included is performing coordinate rollback calculation only on the vertices in V_list.

[0165] Specifically, the rollback calculation process is to traverse each vertex index in V_list, read the original coordinates of the vertex from the history snapshot, and apply the coordinate interpolation algorithm to smoothly transition to the historical state. Among them, the new coordinates = current coordinates x (1-a) + historical coordinates x a, a gradually changes from 0 to 1, and it takes 20ms to complete the transition animation.

[0166] Further, for history version management, before each operation instruction is executed, the coordinates of the affected vertices are automatically saved as a snapshot, and the snapshot data is stored in differential form, only recording the change amount (Δx, Δy, Δz).

[0167] Further, for data consistency verification, after rollback is completed, the CRC32 check value of the local region is calculated and compared with the reference check code stored on the server side, and the error rate is <0.001%.

[0168] In the embodiments of the present application, for operation conflicts in three-dimensional modeling, a precise space rollback algorithm is designed to achieve "targeted repair" and only restore the affected area, avoiding global data reset, such as replacing only the faulty parts in a complex machine instead of disassembling the entire machine. Compared with full tree traversal, the positioning speed is greatly improved, the memory access locality is optimized, and the cache hit rate is improved.

[0169] In some embodiments, before the S3 step is executed, a preloading step is further included, specifically as follows:

[0170] S301, according to the current operation tool type of the user, predict the model area that may be modified.

[0171] Specifically, a tool type library is predefined. Each type of tool is associated with a spatial influence mode, for example, a structure cutting knife, which predicts a columnar region as the region shape, and the influence radius coefficient is 2.0x the length of the knife; a curved surface carving pen, which predicts a spherical region as the region shape, and the influence radius coefficient is 1.5x the pen touch radius; a material spray gun, which predicts a conical diffusion region as the region shape, and the influence radius coefficient is 3.0x the nozzle diameter.

[0172] Specifically, the dynamic prediction model monitors the tool motion trajectory and operation pressure in real time. The model predicts the possible operation area in the next 5 frames (about 80ms) based on a Kalman filter, such as input: tool speed vector, action intensity value; output: predicted region three-dimensional bounding box (AABB).

[0173] S302, pre-pull the spatial topology data of the region from the server.

[0174] Specifically, the cache data is organized in a quadtree structure, and the cache hit rate is above 85%.

[0175] Further, the space division rule of the quadtree structure organization is that the model is projected to a two-dimensional plane (XZ axis) at the top layer, four quadrants are divided, and each quadrant is recursively subdivided to meet the condition that the number of nodes inside the vertex is ≤500. The cache strategy is to keep high-frequency access nodes in memory and compress low-frequency nodes to SSD cache area. At the same time, a Morden code index is established to achieve O(logn) complexity query.

[0176] Specifically, the data request only requests the topological data subset in the predicted area, including vertex coordinates and connection relationship, material UV mapping information, and physical property parameters (such as stiffness and density).

[0177] Specifically, the data pre-fetching process is,

[0178] When the user selects the carving pen, the future influence sphere area (radius R=15 cm) is predicted, the quadtree node data (node level L3-L5) of the area is requested from the server, and the cache data packet size is about 2.8 MB.

[0179] When the user switches to the structure cutting knife, the low-priority cache (such as material data) is dynamically released, and the new predicted columnar area data (priority weight +30%) is loaded.

[0180] In the embodiments of the present application, a forward-looking data pre-fetching system is established, the user operation intention is intelligently predicted, the necessary data of the potential modification area is loaded in advance, the modeling process is equipped with a "space navigation radar", and the influence of network delay on real-time collaboration is significantly reduced.

[0181] In some embodiments, the S4 step further includes:

[0182] S401, collect historical operation conflict data to train an LSTM prediction model.

[0183] Specifically, in the collection and preprocessing of the historical operation conflict data, the specific implementation steps are to collect the historical operation logs within 6 months, the key fields include operation type, space-time coordinates, tool parameters, conflict occurrence mark, solution time consumption, and influence range. A three-dimensional space-time feature vector is constructed, wherein the time dimension is the operation interval sequence (the time difference of the last 10 operations); the space dimension is the operation area overlap rate change curve; and the semantic dimension is the operation type combination correlation score.

[0184] Specifically, the architecture design of the LSTM prediction model adopts a double-layer LSTM network (Long Short-Term Memory network), wherein the input layer: 128-dimensional feature vector (containing space-time semantic features); the hidden layer: 256 LSTM units, dropout rate 0.3; the output layer: Sigmoid activation function output conflict probability (0~1). At the same time, the training strategy is adopted as follows: loss function: weighted cross-entropy (weighting 3 times for high conflict samples); optimizer: AdamW (learning rate 1e-4, weight decay 1e-5).

[0185] Specifically, the online learning mechanism of the model is that the deployed edge computing node collects new conflict cases in real time, and performs incremental training every 24 hours to maintain the model prediction accuracy > 92%.

[0186] S402, when a high conflict probability (> 70%) is detected, non-critical operations are actively restricted.

[0187] Specifically, the operation classification strategy is critical operation, non-critical operation, and ordinary operation. Among them, the network bandwidth is allocated in the proportion of 7:2:1 (critical: ordinary: non-critical).

[0188] Among them, the critical operation is structure modification and physical parameter adjustment (priority P1); the non-critical operation is view rotation, annotation addition, and material preview (priority P3); and the ordinary operation is geometric deformation and light adjustment (priority P2).

[0189] Specifically, when the LSTM prediction conflict probability > 70%, the non-critical operation is delayed (queue waiting), the non-critical operation channel is reduced to 10Mbps, and at the same time the interface displays a "high conflict risk - part of the operation is limited" warning icon.

[0190] S403, dynamically adjust the threshold value of Δt: Δt=base value 50ms x (1-conflict probability).

[0191] Specifically, the base value is initially set as Δt=50ms (corresponding to the median value of 5G network RTT), wherein according to the network type, it is dynamically calibrated as follows: in 5G network environment, the base Δt=50ms, and the adjustment coefficient is 1.0; in Wi-Fi6 network environment, the base Δt=65ms, and the adjustment coefficient is 0.8; in 4G network environment, the base Δt=100ms, and the adjustment coefficient is 0.5.

[0192] Specifically, the dynamic calculation formula of the conflict probability responsive adjustment is Δt_dynamic=Δt_base x (1-conflict probability). For example, when the conflict probability=20%, Δt=50x0.8=40ms (accelerate decision); when the conflict probability=80%, Δt=50x0.2=10ms (strict arbitration).

[0193] Further, a boundary protection mechanism is established, and a minimum value of Δt is set to 10 ms (to avoid excessive sensitivity). When Δt < 20 ms, network quality detection is started. If the packet loss rate is greater than 3%, the Δt adjustment is frozen, and an alarm is triggered.

[0194] In the embodiments of the present application, through the machine learning driven conflict prediction system and the adaptive time window adjustment technology, a conflict management framework with "prophetic ability" is constructed, and the traditional passive response is converted into active defense, like installing "intelligent traffic signal" for the collaborative modeling system.

[0195] In some embodiments, in the S5 step, the binary log is compressed using differential encoding, specifically including:

[0196] A dictionary encoding is established for the operation type field, replacing the original 4-byte enumeration value with 1 byte.

[0197] Specifically, the dictionary library is constructed by predefining 128 common operation types (such as cutting, stretching, mapping, etc.), and each type is assigned a unique 1-byte code.

[0198] Further, under the structure cutting operation type, the code is 0x01, and the example scenario is engine cylinder opening; under the curved surface stretching operation type, the code is 0x02, and the example scenario is automobile fender modeling; under the material replacement operation type, the code is 0x03, and the example scenario is building exterior wall paint replacement.

[0199] At the same time, 32 custom encoding bits (0x80-0x9F) are reserved. When a new operation type is detected, an extended code is automatically assigned and synchronized to all clients.

[0200] Also included is the storage of spatial coordinates using relative pre-operation Δ values.

[0201] Specifically, the calculation steps of relative coordinates are to record the difference (Δx, Δy, Δz) between the current operation coordinates and the pre-operation coordinates, and the absolute coordinates of the first operation are stored as reference values.

[0202] For example, the pre-operation coordinates are (1.2, 3.4, -0.5), and the current coordinates are (1.3, 3.35, -0.48). The Δ value storage is (+0.1, -0.05, +0.02).

[0203] Further, when the Δ value exceeds the type range, the absolute coordinate reference point is inserted, and precision mode upgrade is automatically triggered (such as switching from int16 to float32).

[0204] Also included is the use of DEFLATE algorithm for streaming compression.

[0205] Specifically, the compressed log format is: | header identification (0xAE) | compression flag | dictionary encoding | delta coordinate value | CRC check |.

[0206] The header identification (0xAE) is a magic number identifying the log type, which helps fast retrieval, and the extension identification bit is reserved (version number, encryption flag, etc.).

[0207] In the compression flag, the bit definition is as follows: [7]: whether to compress (0-original / 1-compressed); [6:4]: compression algorithm (000-DEFLATE); [3:0]: reserved bit.

[0208] The CRC check uses the CRC32 algorithm to calculate the data integrity, and the check range is through the dictionary encoding + delta coordinate value field, and the error detection rate is 100% (single-bit error).

[0209] Specifically, the compression algorithm selects the DEFLATE algorithm (LZ77 + Huffman encoding), processes the data stream in blocks, sets a 32KB sliding window, and matches the repeated patterns within 256 bytes.

[0210] Specifically, the real-time compression process steps are as follows: input: original byte stream of dictionary encoding + delta coordinate; block: each 1024 operation record is a compression unit; output: compressed data stream (typical compression ratio value 5:1).

[0211] In the embodiments of the present application, through the multi-layer compression strategy and the incremental storage technology, efficient storage and transmission of three-dimensional operation logs are realized, like providing an "intelligent compression clothing" for massive modeling data, which greatly reduces resource consumption on the premise of ensuring data integrity.

[0212] Next, a modeling system based on dynamic 3D targets according to an embodiment of the present application is introduced. The modeling system based on dynamic 3D targets described below can be mutually corresponding to the modeling method based on dynamic 3D targets described above.

[0213] Reference Figure 2 A modeling system based on dynamic 3D targets includes:

[0214] An operation instruction generation module 1 is configured to receive user input and generate three-dimensional modeling operation instructions, the instructions at least including an operation type field, a target object unique identifier, and a three-dimensional space scope parameter set.

[0215] A semantic tree construction module 2 is in data connection with the operation instruction generation module 1 and includes,

[0216] A topological relationship analysis unit extracts the three-dimensional spatial topological relationship of the target object based on a half-edge data structure (Half-edge) to generate parent node data.

[0217] Scope calculation unit: calculates the weight distribution matrix of the child node according to the spatial influence algorithm corresponding to the operation type.

[0218] Conflict detection module 3: bidirectionally connected with the local storage module and the network communication module 6, including,

[0219] Semantic tree comparator: performs spatial scope overlap detection on the local operation semantic tree and the received remote semantic tree;

[0220] Priority arbiter: built-in space-time dimension judgment logic circuit, which triggers the arbitration process when overlap is detected.

[0221] Operation merging execution module 4: connected with the output end of the conflict detection module 3, including,

[0222] Rule base: stores a space-time priority coefficient reference table, wherein the structure modification P1, the geometric deformation P2, and the material editing P3 satisfy P1>P2>P3;

[0223] Difference calculation unit: generates an operation delta package containing only the vertex changes in the overlap area.

[0224] Log management module 5: communicatively connected with all modules, including,

[0225] Binary encoder: converts operation records into compressed logs containing semantic tree hash values;

[0226] Local rollback controller: in response to a conflict signal, performs coordinate rollback of affected vertices based on the differential CRC check value.

[0227] Network communication module 6: adopts a double-channel design, including,

[0228] Operation instruction transmission channel: encapsulates semantic tree data packets;

[0229] Log synchronization channel: transmits compressed binary difference logs.

[0230] Local storage module: integrates NVMe SSD and persistent memory, including,

[0231] Operation cache area: LRU algorithm manages the last 1000 operation instructions;

[0232] Semantic tree snapshot library: versioned storage of CSR format compressed topology data;

[0233] Log buffer area: ring queue stores encrypted operation history.

[0234] Among them, the semantic tree construction module 2, the conflict detection module 3 and the operation merging and execution module 4 are interconnected through a shared memory bus in the hardware layer to form a real-time processing pipeline architecture.

[0235] In the embodiment, the modeling system adopts a modular heterogeneous architecture design, decomposes the core logic of three-dimensional modeling into six functionally decoupled modules, and realizes real-time collaborative modeling of dynamic 3D targets through hardware-level acceleration and intelligent algorithm fusion. The modules are interconnected through a high-speed data bus to form a closed loop of "perception-decision-execution".

[0236] The embodiment of the application provides an electronic device, such as Figure 3 As shown in the figure, Figure 3 A structural schematic diagram of an electronic device provided by the embodiment of the application is shown in the figure, Figure 3 The electronic device 300 shown in the figure includes a processor 301 and a memory 303. The processor 301 and the memory 303 are connected, such as through a bus 302. Optionally, the electronic device 300 can also include a transceiver 304. It should be noted that the transceiver 304 is not limited to one in actual application, and the structure of the electronic device 300 does not constitute a limitation on the embodiment of the application.

[0237] The processor 301 can be a CPU (Central Processing Unit, central processor), a general-purpose processor, a DSP (Digital Signal Processor, data signal processor), an ASIC (Application Specific Integrated Circuit, application-specific integrated circuit), an FPGA (Field Programmable Gate Array, field programmable gate array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It can realize or execute various exemplary logical blocks, modules and circuits described in combination with the disclosure of the embodiment of the application. The processor 301 can also be a combination of computing functions, such as one or more microprocessor combinations, combinations of DSP and microprocessor, etc.

[0238] The bus 302 can include a path for transmitting information between the above-mentioned components. The bus 302 can be a PCI (Peripheral Component Interconnect, peripheral component interconnect) bus or an EISA (Extended Industry Standard Architecture, extended industry standard architecture) bus, etc. The bus 302 can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, Figure 3 Only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0239] The memory 303 can be a ROM (Read Only Memory) or other type of static storage device that can store static information and instructions, a RAM (Random Access Memory) or other type of dynamic storage device that can store information and instructions, an EEPROM (Electrically Erasable Programmable Read-Only Memory), a CD-ROM (Compact Disc Read-Only Memory) or other optical disk storage, a magnetic disk storage or other magnetic storage devices, or any other medium capable of storing desired program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited thereto.

[0240] The memory 303 is configured to store application program codes for implementing the embodiments of the present application, and the processor 301 is configured to control the execution of the application program codes. The processor 301 is configured to execute the application program codes stored in the memory 303 to implement the content shown in the foregoing method embodiments.

[0241] The electronic device includes, but is not limited to, a mobile terminal such as a mobile phone, a notebook computer, a digital broadcast receiver, a PDA (Personal Digital Assistant), a PAD (Tablet Personal Computer), a PMP (Portable Multimedia Player), a car terminal (for example, a car navigation terminal), and the like, and a fixed terminal such as a digital TV, a desktop computer, and the like. Figure 3 The electronic device shown is only an example, and should not bring any limitation to the function and use range of the embodiments of the present application.

[0242] The embodiments of the present application provide a computer readable storage medium, and the computer readable storage medium stores a computer program. The computer program is executed by a processor to implement the steps of the modeling method based on a dynamic 3D target.

[0243] Since the embodiments of the computer readable storage medium part correspond to the embodiments of the method part, the embodiments of the computer readable storage medium part are described with reference to the description of the embodiments of the method part.

[0244] It should be understood that although the steps in the flowcharts of the drawings are shown in a sequence as indicated by arrows, the steps are not necessarily executed in the order as indicated by the arrows. Unless explicitly stated, the execution of the steps is not necessarily limited in sequence, and the steps can be executed in other sequences. Moreover, at least some of the steps in the flowcharts of the drawings can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence is not necessarily sequential, but can be executed in rotation or alternation with at least part of other steps or sub-steps or stages of other steps.

[0245] The above merely describes some embodiments of the present application, and it should be pointed out that for those skilled in the art, some improvements and refinements can be made without departing from the principles of the present application, and these improvements and refinements should also be considered as falling within the protection scope of the present application.

Claims

1. A modeling method based on dynamic 3D targets, characterized in that, include: Generate 3D modeling operation instructions, wherein the operation instructions include at least the operation type, the target object identifier, and the spatial scope parameter; An operation semantic tree is constructed based on the operation instructions. The nodes of the operation semantic tree contain the following hierarchical structure. The parent node and the target object's three-dimensional spatial topological relationship. Child nodes, the spatial scope weight value of user operations; By comparing the local operation semantic tree with the received remote operation semantic tree, the overlapping area of ​​the spatial scope is detected. When overlapping regions are detected, a spatiotemporal priority merging rule is applied, as follows: The time dimension takes precedence; when the time difference between the operation initiation and the time difference is less than the threshold Δt, the spatial dimension judgment is activated. Spatial dimension priority conditions are applied, and priority coefficients are assigned according to the operation type, with structural modification operation priority P1 > geometric deformation operation priority P2 > material editing operation priority P3. Execute the merged operation instructions and generate a binary log containing the operation semantic tree hash value; When an operational conflict is detected, a partial rollback is performed based on the differential CRC check value, which only calculates the data blocks in the overlapping area.

2. The method according to claim 1, characterized in that, The step of constructing an operation semantic tree based on the operation instructions includes: The three-dimensional spatial topological relationship of the parent node is recorded using a half-edge data structure to record the connection relationship between vertices, edges, and faces; The spatial scope weight of child nodes is calculated using a Gaussian decay function, and the scope radius σ is positively correlated with the preset influence range of the operation type. The structural modification operation σ = 20% of the diagonal length of the model bounding box; Geometric deformation operation σ = 150% of the radius of the operating tool; Material editing operation σ = 30% of the texture map UV unfolded area.

3. The method according to claim 1, characterized in that, In the step of applying the spatiotemporal dimension priority merging rule when an overlapping region is detected, the priority coefficient is dynamically adjusted according to the model type, including: In the mechanical assembly model, P1:P2:P3 = 4:3:1; The weight of P2 in organic biological models increased by 30%; Add a lighting operation priority of P4 = 0.8 × P1 to the architectural scene model; The adjustments are based on the classification tags in the model file metadata.

4. The method according to claim 1, characterized in that, In the step of performing a partial rollback based on the differential CRC check value when an operation conflict is detected, the partial rollback includes: Locate the subtree node that needs to be rolled back based on the operation semantic tree; Generate an index list V_list of the affected vertices, which is sorted by spatial Morton code; Coordinate backtracking is performed only on vertices in V_list.

5. The method according to claim 1, characterized in that, Before the step of detecting overlapping spatial scope regions by comparing the local operation semantic tree with the received remote operation semantic tree is executed, a preloading step is also included: Predict the model area that may be modified based on the user's current operating tool type; Spatial topology data for the region is retrieved from the server in advance; The cached data is organized in a quadtree structure, and the cache hit rate is over 85%.

6. The method according to claim 1, characterized in that, The step of applying the spatiotemporal priority merging rule when an overlapping region is detected further includes: Collect historical operational conflict data to train an LSTM prediction model; Actively restrict non-critical operations when a high probability of conflict (>70%) is detected; Dynamically adjust the Δt threshold: Δt = base value 50ms × (1 - conflict probability).

7. The method according to claim 1, characterized in that, In the step of executing the merged operation instructions and generating a binary log containing the operation semantic tree hash value, the binary log uses differential encoding compression, including: Establish dictionary encoding for operation type fields, replacing the original 4-byte enumeration value with 1 byte; Spatial coordinates are stored using Δ values ​​relative to the preceding operation; Streaming compression is performed using the DEFLATE algorithm; The compressed log format is as follows: |Header Identifier (0xAE)|Compression Flag|Dictionary Encoding|Δ Coordinate Value|CRC Check Value|.

8. A modeling system based on dynamic 3D targets, characterized in that, include: Operation instruction generation module (1): configured to receive user input and generate three-dimensional modeling operation instructions, wherein the instructions include at least an operation type field, a unique identifier of the target object and a three-dimensional space scope parameter set; Semantic tree construction module (2): data-connected to the operation instruction generation module (1), including, Topology parsing unit: Extracts the three-dimensional spatial topology of the target object based on the half-edge data structure and generates parent node data; Scope calculation unit: Calculates the weight distribution matrix of child nodes according to the spatial influence algorithm corresponding to the operation type; Conflict detection module (3): bidirectionally connected to the local storage module and the network communication module, including, Semantic tree comparator: performs spatial scope overlap detection between the local operating semantic tree and the received remote semantic tree; Priority Arbitrator: Built-in spatiotemporal dimension determination logic circuit, which triggers the arbitration process when overlap is detected; Operation merging execution module (4): connected to the output of the conflict detection module (3), including, Rule base: Stores a table of spatiotemporal priority coefficients, where the priority of structural modification operation P1, the priority of geometric deformation operation P2, and the priority of material editing operation P3 satisfy P1>P2>P3; Differential computation unit: generates an incremental operation packet that contains only changes to vertices in the overlapping region; Log management module (5): Communicates with all modules, including, Binary encoder: Converts operation records into compressed logs containing semantic tree hash values; Local rollback controller: When responding to a conflict signal, it performs coordinate rollback of the affected vertex based on the differential CRC check value; Network communication module (6): adopts a dual-channel design, including, Operation command transmission channel: encapsulates semantic tree data packets; Log synchronization channel: transmits compressed binary difference logs; Local storage module (7): integrates NVMe SSD and persistent memory, including, Operation buffer: The LRU algorithm manages the most recent 1000 operation instructions; Semantic Tree Snapshot Library: Versioned storage of CSR-formatted compressed topology data; Log buffer: A circular queue stores the history of encryption operations; The semantic tree construction module (2), conflict detection module (3), and operation merging execution module (4) are interconnected at the hardware layer through a shared memory bus to form a real-time processing pipeline architecture.

9. An electronic device, characterized in that, include: One or more processors; One or more memory units; And one or more computer programs, wherein the one or more computer programs are stored in the one or more memories, the one or more computer programs including instructions that, when executed by the one or more processors, cause the electronic device to perform the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The storage medium stores a program or instructions that, when executed, implement the method as described in any one of claims 1 to 7.

Images