Method for updating appearance of three-dimensional model and electronic device

Abstract

The appearance updating method of a three-dimensional model and an electronic device provided by the present application relate to the technical field of automobile 3D modeling. The method comprises the following steps: receiving an appearance updating instruction for a target three-dimensional model, and obtaining a corresponding regular texture mapping diagram, which is a two-dimensional image generated by unfolding a surface to be updated of the target three-dimensional model; analyzing the appearance updating instruction to obtain appearance updating semantic information; generating one or more new regular texture mapping diagrams based on the regular texture mapping diagram and the appearance updating semantic information; and in response to a selection operation of the user on the one or more new regular texture mapping diagrams, fitting the selected new regular texture mapping diagram to the corresponding surface of the target three-dimensional model to update the appearance of the target three-dimensional model. The present application can enable the user to quickly obtain a car skin effect in line with his / her own preferences through a simple instruction, while supporting targeted customization of different surface regions of the three-dimensional model, and realizing personalized and differentiated customization of the appearance of the three-dimensional model.

Description

Technical Field

[0001] This application relates to the field of automotive 3D modeling technology, and more specifically, to a method and electronic device for updating the appearance of a 3D model. Background Technology

[0002] With the popularization and application of 3D digitization technology, 3D models have become a core carrier of information presentation and human-computer interaction in many fields, such as product display, virtual simulation, games and entertainment, and digital twins. Users are no longer satisfied with static and fixed 3D model appearances, but want to personalize and update and customize the appearance of 3D models in real time according to their personal preferences or scenario requirements.

[0003] Currently, updating the appearance of 3D models mainly relies on preset resource replacement methods. For example, developers or content creators need to pre-create multiple sets of fixed texture maps (such as colors, patterns, and materials), and users can only switch between them in a limited number of preset options.

[0004] However, while existing technologies can achieve basic appearance changes, the limited number of preset styles fails to meet users' personalized needs, easily leading to aesthetic fatigue. Furthermore, existing technologies lack an intelligent solution that can understand users' natural intentions (such as "Give me a cyberpunk-style paint job") and automatically generate texture maps that perfectly match the surface geometry of a specific 3D model. This prevents users' personalized ideas from being directly applied to 3D model appearance updates, severely limiting the richness of the interactive experience and user engagement. Summary of the Invention

[0005] The method and electronic device for updating the appearance of a 3D model provided in this application allow users to quickly obtain a car wrap effect that suits their preferences with just simple commands. At the same time, it supports targeted customization of different surface areas of the 3D model, realizing personalized and differentiated customization of the appearance of the 3D model.

[0006] Firstly, a method for updating the appearance of a 3D model is provided. The method includes: receiving an appearance update instruction for a target 3D model and obtaining a corresponding regular texture map, wherein the regular texture map is a two-dimensional image generated by unfolding the surface to be updated of the target 3D model; parsing the appearance update instruction to obtain appearance update semantic information; generating one or more new regular texture maps based on the regular texture map and the appearance update semantic information; and responding to a user's selection operation of one or more new regular texture maps, applying the selected new regular texture map to the corresponding surface of the target 3D model to update the appearance of the target 3D model.

[0007] In the above technical solution, the appearance update command for the target 3D model is received first, and a regular texture mapping map generated by unfolding the surface of the target 3D model to be updated is obtained. This lays the foundation for matching the model's geometric structure for subsequent texture generation and bonding, avoiding texture bonding deformation and misalignment problems. Then, the appearance update command is segmented, denoised, and semantically parsed to obtain structured appearance update semantic information, realizing the understanding of the user's natural language intent. This transforms everyday, non-standardized appearance update needs into AI-recognizable execution criteria, solving the deficiency of existing technologies in understanding the user's natural intent. Subsequently, based on the regular texture mapping map and the appearance update semantic information, a preset AI large model sequentially completes three steps: AI calling editing tools, targeted editing processing, and new image generation to generate one or more new regular texture mapping maps. The AI ​​large model automatically calls a dedicated appearance texture editor to identify the size and contour parameters of the original regular texture mapping map without damaging the model. The model adapts parameters and uses the original regular texture map as a carrier, modifying only the texture content such as color, pattern, and symmetry while retaining the original image size and outline shape. This process outputs one or more new images, overcoming the limitations of the limited number of preset styles in existing technologies, satisfying diverse and personalized user needs, avoiding aesthetic fatigue, and ensuring perfect adaptation of the new image to the surface geometry of the target 3D model. This allows users' personalized ideas to be transformed into directly applicable appearance textures. Finally, in response to the user's selection of one or more new regular texture maps, the completed 3D model is adapted, and the new regular texture map, without secondary adjustments, is applied to the corresponding surface of the target 3D model, completing the instant update of the target 3D model's appearance. The updated appearance can be fully integrated in all scenes displaying the 3D model, fundamentally solving the problems of fixed 3D model appearances and the inability to implement users' personalized ideas in existing technologies, significantly improving the richness of the interactive experience and user engagement.

[0008] Secondly, a device for updating the appearance of a three-dimensional model is provided. The device includes: a receiving module for receiving an appearance update instruction for a target three-dimensional model and obtaining a corresponding regular texture map, wherein the regular texture map is a two-dimensional image generated by unfolding the surface to be updated of the target three-dimensional model; a parsing module for parsing the appearance update instruction to obtain appearance update semantic information; a generating module for generating one or more new regular texture maps based on the regular texture map and the appearance update semantic information; and a bonding module for bonding the selected new regular texture map to the surface corresponding to the target three-dimensional model in response to the user's selection operation of one or more new regular texture maps, so as to update the appearance of the target three-dimensional model.

[0009] Thirdly, an electronic device is provided, including a memory and a processor. The memory is used to store executable program code, and the processor is used to call and run the executable program code from the memory, causing the electronic device to perform the methods of the first aspect or any possible implementation thereof.

[0010] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of this application more easily understood, specific embodiments of this application are given below. Attached Figure Description

[0011] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application; Figure 2 A flowchart illustrating a method for updating the appearance of a 3D model provided in this application embodiment. Figure 1 ; Figure 3 A flowchart illustrating a method for updating the appearance of a 3D model provided in this application embodiment. Figure 2 ; Figure 4 A flowchart illustrating a method for updating the appearance of a 3D model provided in this application embodiment. Figure 3 ; Figure 5 A flowchart illustrating a method for updating the appearance of a 3D model provided in this application embodiment. Figure 4 ; Figure 6 This is a schematic diagram of the structure of a three-dimensional model appearance updating device provided in an embodiment of this application. Detailed Implementation

[0012] The technical solutions in this application will be clearly and thoroughly described below with reference to the accompanying drawings. In the description of the embodiments of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. "And / or" in the text is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Furthermore, in the description of the embodiments of this application, "multiple" refers to two or more than two.

[0013] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature.

[0014] With the widespread application of 3D models in various fields, users' demand for personalized, real-time updates to their appearance is increasing. However, current technologies mainly rely on pre-made fixed texture maps for replacement, offering limited style options and failing to meet users' deep and diverse creative needs. More importantly, there is a lack of an intelligent solution that can understand users' natural language intent and automatically generate texture maps that match the surface geometry of a specific 3D model. This makes it difficult to efficiently and accurately translate personalized ideas into directly applicable appearance updates, severely limiting the interactive experience and user engagement. To address these issues, this application proposes a 3D model appearance update scheme based on the linkage between a preset AI (Artificial Intelligence) large model and a 3D model's UV (Ultro Violet) map. The following detailed description, in conjunction with the accompanying drawings, through multiple embodiments, illustrates the 3D model appearance update method and electronic device of this application. The preset AI large model can be selected according to actual conditions.

[0015] Figure 1 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 1 As shown, the electronic device 100 may include a processor 110 and a memory 120.

[0016] The memory 120 stores machine-executable instructions that can be executed by the processor 110. When the electronic device 100 is running, these machine-readable instructions are executed. The processor 110 and the memory 120 communicate via a bus. The processor 110 can execute these machine-executable instructions to implement a method for updating the appearance of a 3D model.

[0017] The memory 120, processor 110, and bus components are electrically connected directly or indirectly to enable data transmission or interaction. For example, these components can be electrically connected via one or more communication buses or signal lines. The memory 120 includes at least one software functional module, which can be stored or embedded in the operating system (OS) of the electronic device in the form of software or firmware. This software functional module can be an executable module. The processor 110 is used to execute the executable module stored in the memory 120, such as the software functional modules and computer programs included in the appearance update method of a 3D model.

[0018] The memory 120 may be, but is not limited to, random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc.

[0019] The electronic device 100 can be selected according to the actual situation. Furthermore, the electronic device 100 has software capable of executing methods for updating the appearance of a 3D model.

[0020] The method for updating the appearance of a 3D model provided in this application embodiment can be executed by a processor in the electronic device 100. The method for updating the appearance of a 3D model provided in this application embodiment will be explained further below. Figure 2 A flowchart illustrating a method for updating the appearance of a 3D model provided in this application embodiment. Figure 1 .like Figure 2 As shown, the method may include: S210: Receive the appearance update command for the target 3D model and obtain the corresponding regular texture mapping map.

[0021] The target 3D model can be selected according to the actual situation. For example, the target 3D model can be any type of 3D model such as daily consumer goods, industrial components, virtual characters, or home appliances. It should be noted that, for ease of understanding, the following content will use a vehicle 3D model as an example for explanation, and this example does not constitute any limitation on this application.

[0022] The regular texture mapping map (i.e., UV map) is a two-dimensional image generated by unfolding the surface to be updated of the target 3D model. The surface to be updated of the target 3D model (such as a vehicle 3D model) is the surface of the 3D model created by the 3D engine (such as Unity or Kanzi) module inside the terminal device (such as an in-vehicle infotainment system), and its structure and the parts to be updated are completely consistent with the corresponding entity (if any) of the target 3D model; if the target 3D model is a virtual model (without a corresponding entity), its surface to be updated is completely matched with the preset appearance parts of the model design.

[0023] In one possible implementation, the target user first needs to access the pre-set 3D car model modification module on the terminal device. The target user then issues a customizable appearance texture command via any pre-set interaction method, such as voice, touch, or text input. This command can be any expression that meets the target user's needs, such as "Generate a symmetrical yellow bear car cover for me" or "I want to put a nice car cover on my car model." Upon receiving the command, the terminal device uses existing corresponding interaction processing technologies (e.g., word segmentation, noise reduction, and accidental touch filtering for voice commands, and validity verification for text / touch commands) to standardize the command. Then, using the terminal device's pre-set fuzzy matching domain determination mechanism, it determines whether the command belongs to the AI-generated appearance texture domain. If the command successfully falls into the AI-generated appearance texture domain, the terminal device confirms receipt of the appearance update command (i.e., receives the target user's appearance update command) and notifies the terminal device's main control module of the domain determination result, triggering subsequent actions. If the domain determination fails, the user is notified according to existing interaction processing technology logic (e.g., "Command cannot be recognized, please re-enter"), and no further operations are performed.

[0024] The preset fuzzy matching domain mechanism is used to solve the problem of approximate matching between the model surface and UV coordinates / texture elements, so as to ensure the accuracy and fault tolerance of command domain placement.

[0025] After the terminal device creates a target 3D model that is consistent with the target entity (if any) or conforms to the preset design using its own preset 3D engine module, it extracts the 2D planar information of all surfaces to be updated on the 3D car model. Each independent surface to be updated corresponds to an independent 2D image, which records the shape, size, and position information of the corresponding surface to be updated on the 3D model. Then, it calls the UV editing tool built into the 3D engine module, using the preset baseline of the 3D model (such as the bottom center line, the center line of symmetry, etc., which can be preset according to the model type) as a reference, and unfolds all surfaces to be updated along the baseline in the corresponding direction (such as the left and right sides, the front and back directions, etc., to adapt to the model structure). The surfaces to be updated with perspective view are planarized, and finally a regularized UV map (i.e., regular texture mapping map) that matches the part to be updated of the target 3D model 1:1 is generated. Different types and models of target 3D models generate exclusive regular texture mapping maps. All regular texture mapping maps are stored in the terminal device's preset UV map database according to model type and model.

[0026] Once the terminal device successfully receives the appearance update command, the main control module of the terminal device retrieves the corresponding regular texture mapping map from the UV map database based on the type and model information of the current target 3D model (i.e., the 3D car model that the user is currently operating and that is consistent with the target entity (if any) or conforms to the preset design). The retrieved regular texture mapping map is in PNG format, and the outline and size of the part to be updated on the target 3D model are completely matched, providing a stable carrier for subsequent AI large model editing of appearance texture.

[0027] S220. Parse the appearance update instruction to obtain appearance update semantic information.

[0028] In one possible implementation, the terminal device first receives a user's appearance update command in natural language (e.g., "Help me generate a symmetrical appearance texture for a yellow bear"). Since user appearance update commands are often informal and non-standardized, and the terminal device's existing voice system already covers multiple functional areas, the appearance update command is first segmented, breaking down the coherent natural language into independent semantic units (e.g., breaking down "Help me generate a symmetrical appearance texture for a yellow bear" into "yellow," "bear," "symmetrical," and "appearance texture"). Then, meaningless interjections and redundant words (e.g., auxiliary words like "help me" and "generate") are filtered out, retaining the core information directly related to the appearance update. Finally, the terminal device's preset semantic parsing algorithm identifies and categorizes the segmented and denoised core semantic units to determine whether the appearance update command falls within the domain of AI-generated appearance textures. This avoids accidentally triggering other functions of the terminal device (e.g., device control, entertainment), ensuring that the personalized update process for the 3D model appearance can be accurately initiated. The semantic information for appearance updates obtained after parsing is structured data, not the original natural language instructions. This structured data is classified according to different dimensions, such as color dimension (e.g., "yellow"), pattern dimension (e.g., "bear"), and appearance texture type dimension (e.g., "left and right symmetrical"). This structured data processing can solve the problem that non-standardized user instructions cannot be recognized by large AI models.

[0029] For example, when raw instructions are directly transmitted to a pre-defined AI model, the AI ​​cannot clearly define the specific update requirements. However, structured semantic information provides a clear and explicit execution basis for the AI ​​model's appearance texture editing process, enabling it to match the corresponding appearance texture generation rules. For instance, when the semantic information "left and right are inconsistent" is parsed, the pre-defined AI model will perform the operation of "searching for two different images and matching them to the left and right areas of the UV map respectively"; when the semantic information "front, back, left, and right are inconsistent" is parsed, it will match "matching the four images to the UV map respectively". Figure 4The operation involves "each region". Simultaneously, this appearance update semantic information also serves to search and match user-preferred resources using the AI ​​large model, ensuring that the generated appearance textures meet the user's personalized needs. Furthermore, this appearance update semantic information provides a foundation for subsequent processes, including the AI ​​large model generating multiple appearance texture schemes for the user to choose from, and secondary editing after the user issues fine-tuning commands, ensuring the continuity and integrity of the entire 3D model appearance personalization update process.

[0030] S230. Based on the regular texture mapping map and appearance update semantic information, generate one or more new regular texture mapping maps.

[0031] In one possible implementation, appearance update semantic information and regular texture mapping are used. Figure 1 After being transferred to the preset AI large model, the system automatically retrieves its associated dedicated appearance texture editor. This dedicated appearance texture editor is a tool specifically designed for editing the appearance texture of 3D models. It can recognize the size and contour parameters of regular texture maps, enabling targeted modifications to the original texture maps without disrupting the original model's adaptation parameters. Then, using the retrieved regular texture map as a base, the preset AI large model strictly adheres to the user's personalized needs contained in the appearance update semantic information, performing targeted editing on the original regular texture map. This means that the size and contour of the original regular texture map are preserved throughout the editing process, with only the texture content (such as color, pattern, symmetry, etc.) modified to ensure consistent compatibility between the new and original maps. Finally, after editing is complete, the preset AI large model automatically outputs one or more new regular texture maps. The option of one or more maps is designed to accommodate diverse user needs, allowing users to choose the optimal solution later.

[0032] It should be noted that, because the size and contour parameters of the original regular texture mapping map are strictly preserved during the editing process, it can ensure that the subsequent terminal 3D engine can adapt the new map to the target 3D model. This effectively solves the problem of easy deformation and misalignment when custom textures are attached to 3D models in existing technologies. This is also an important added value of the generation action. It not only completes the core goal of generating new maps, but also provides accuracy assurance for the subsequent appearance fitting steps, ensuring the continuity and accuracy of the entire process of customizing the appearance of 3D models.

[0033] S240, in response to the user's selection of one or more new regular texture maps, the selected new regular texture map is applied to the surface corresponding to the target 3D model to update the appearance of the target 3D model.

[0034] In one possible implementation, the user first enters the 3D model modification module of the terminal device. At this time, the terminal device interface will automatically display preview images of the appearance effects corresponding to all the new rule texture maps returned by the preset AI large model. The user can select one or more new rule texture maps that they are satisfied with from the preview images displayed on the terminal device interface through terminal touch operation or voice command confirmation. The terminal device will receive and recognize the user's selection operation command in real time, and use it as the basis for triggering the subsequent appearance texture bonding action. The entire selection process does not require the user to have professional 3D editing or image processing capabilities, which greatly reduces the operation threshold. Once the user completes the selection and confirmation, the main control module of the terminal device will first transmit the selected PNG format new rule texture map to the terminal's built-in 3D engine module in real time. Then, the 3D engine module calls its own dedicated UV map bonding and rendering function to bond the new rule texture map to the specified surface of the target 3D model according to the surface shape and position coordinate information of the target 3D model pre-recorded in the new rule texture map. Moreover, the new rule texture map has been pre-adapted to the size and contour of the target 3D model by the preset AI large model, so there is no need for the terminal device to make secondary precision adjustments or modifications, and the bonding can be directly achieved without misalignment or deformation.

[0035] Once the 3D engine module completes the bonding and rendering of the new rule texture map, it will feed back the rendered 3D model image to the terminal device interface in real time, realizing the instant visual update of the target 3D model's appearance. At the same time, the updated personalized appearance can be fully integrated with all scenes displaying 3D models on the terminal device without any scene adaptation issues. Users can view the updated effect on all interfaces displaying 3D models on the terminal device, ultimately completing the personalized customization of the target 3D model's appearance. This fundamentally solves the shortcomings of existing technologies where the appearance of 3D models is fixed and cannot be personalized, improving the user's interactive experience and participation in the personalized operation of 3D models.

[0036] The 3D model appearance update method provided in this application first receives an appearance update instruction for the target 3D model and obtains a regular texture mapping map generated from the unfolded surface of the target 3D model to be updated. This lays the foundation for matching the model's geometry for subsequent texture generation and bonding, avoiding texture bonding deformation and misalignment problems. Next, the appearance update instruction is segmented, denoised, and semantically parsed to obtain structured appearance update semantic information, enabling the understanding of the user's natural language intent. This transforms everyday, non-standardized appearance update needs into AI-recognizable execution criteria, addressing the shortcomings of existing technologies that cannot understand users' natural intent. Subsequently, based on the regular texture mapping map and the appearance update semantic information, a pre-set AI large model sequentially completes three stages: AI invoking editing tools, targeted editing processing, and new image generation to generate one or more new regular texture mapping maps. The AI ​​large model automatically invokes a dedicated appearance texture editor to identify the size and contour parameters of the original regular texture mapping map. Furthermore, without disrupting the model's adaptation parameters, the original regular texture mapping map is used as a carrier. Only the texture content, such as color, pattern, and symmetry, is modified while retaining the original image's size and outline. Finally, one or more new images are output. This breaks through the limitations of the limited number of preset styles in existing technologies, meeting diverse and personalized user needs and avoiding aesthetic fatigue. It also ensures that the new images perfectly adapt to the surface geometry of the target 3D model, allowing users' personalized ideas to be transformed into directly applicable appearance textures. Finally, in response to the user's selection of one or more new regular texture mapping maps, the completed 3D model is adapted, and the new regular texture mapping map, which does not require secondary adjustments, is applied to the corresponding surface of the target 3D model. This completes the instant update of the target 3D model's appearance, and the updated appearance can be fully integrated in all scenes displaying the 3D model. This fundamentally solves the problem of fixed 3D model appearances and the inability to realize users' personalized ideas in existing technologies, greatly improving the richness of the interactive experience and user participation.

[0037] Figure 3 A flowchart illustrating a method for updating the appearance of a 3D model provided in this application embodiment. Figure 2 .like Figure 3 As shown, the above method generates one or more new regular texture maps based on the regular texture map and appearance update semantic information, including: S310. Based on the regular texture mapping map, determine the planar regions corresponding to each surface to be updated of the target 3D model.

[0038] In one possible implementation, firstly, the regular texture map itself is a two-dimensional image generated by unfolding all the surfaces to be updated of the target 3D model using the UV editing tools of a 3D engine. Its preparation process synchronously records the shape, size, and position information of each surface to be updated on the target 3D model. Secondly, by tracing the preparation logic of the regular texture map, independent two-dimensional regions in the regular texture map that correspond one-to-one with each surface to be updated on the target 3D model are identified and divided. That is, each independent planar region in the regular texture map uniquely corresponds to a specific surface to be updated on the target 3D model, and this correspondence is synchronously stored during the preparation of the regular texture map and can be directly called upon later without secondary identification. Finally, after clarifying the correspondence through this step, the newly generated regular texture map (personalized texture) can be matched to the specified surface to be updated on the target 3D model, avoiding problems such as texture misalignment and overlay errors.

[0039] For example, in a scenario involving updating the appearance of a 3D vehicle model, since the in-vehicle infotainment system has already created a 3D model of the vehicle (i.e., the target 3D model) that is consistent with the real vehicle using a 3D engine (such as Unity or Kanzi), during the creation of the 3D model, the planar effects of all surfaces with paint (i.e., surfaces to be updated) on the 3D model have been extracted simultaneously. Each paint surface is an independent 2D image, and this 2D image has recorded in detail the shape, size, and specific position information of the corresponding paint surface on the 3D model. Subsequently, the in-vehicle infotainment system calls the UV editing tool of the 3D engine module, using the bottom center line of the 3D model as a reference, to unfold all paint surfaces along the center line to the left and right sides and forward and backward, while simultaneously performing planar editing on paint surfaces with perspective deviations. The process involves chemical correction to ultimately create a regular texture mapping map. Then, based on the regular texture mapping map, the corresponding planar regions for each paint surface (i.e., the surface to be updated) are determined. Essentially, through the above-mentioned preparation process of the regular texture mapping map, the one-to-one correspondence between each independent two-dimensional region in the regular texture mapping map and each specific paint surface on the 3D car model is clarified. That is, each planar region in the regular texture mapping map corresponds to a specific paint surface on the 3D car model, and this correspondence has been recorded and stored synchronously during the preparation of the UV map (regular texture mapping map). It can be directly called when applying the car wrap (applying the new regular texture mapping map) in the future, providing a positional reference for the subsequent correspondence between the car wrap and the paint surface, ensuring that the car wrap can be applied to the designated paint surface of the 3D car model, and avoiding problems such as car wrap misalignment and overflow.

[0040] S320. Based on the semantic information of appearance update, determine the target visual style and target mapping method.

[0041] The target visual style is used to characterize the appearance pattern, color, texture, and overall visual effect of the target 3D model to be updated. The target mapping method is used to characterize the bonding rules, distribution pattern, and region correspondence of the appearance image to be generated on the regular texture map (UV map) corresponding to the target 3D model.

[0042] In one possible approach, a pre-set AI model is used to analyze the semantic information of appearance updates, extract the target visual style desired by the user, and determine the corresponding target texture method based on the requirements for model appearance symmetry, partitioned display, etc. in the semantic information of appearance updates. Taking a 3D in-vehicle model as an example, the semantic information of appearance updates is analyzed by a pre-set AI model to extract the visual effect of the car cover (i.e., the target visual style) desired by the user. For example, "yellow and bear" in the semantic information of appearance updates corresponds to the target visual style of "yellow background with bear pattern". "Colorful and good-looking" in the semantic information of appearance updates corresponds to the target visual style of "rich and beautiful color". This process is completed by the pre-set AI model in combination with semantic keywords to search and match the corresponding visual style features. Then, the pre-set AI model determines the bonding rules of the candidate image content on the regular texture mapping map (UV map) according to the requirements of symmetry of the car cover in the semantic information of appearance updates. For example, if the semantic information of appearance updates specifies a symmetrical car cover, it corresponds to "symmetrical mapping method". If the semantic information of appearance updates specifies an inconsistent car cover, it corresponds to "different mapping method for left and right partitions". If the semantic information of appearance updates specifies an inconsistent car cover in all directions, it corresponds to "different mapping method for front, back and left and right partitions".

[0043] It should be noted that, in addition to the methods listed above, the target texture method may also include other texture methods to match the user's personalized requirements for different areas of the car cover.

[0044] S330: Based on the target visual style, generate candidate image content that matches the size of the regular texture map.

[0045] In one possible implementation, for the target 3D model, candidate image content is generated based on preset target visual style requirements. The size and outline of the candidate image content are adapted to the regular texture map corresponding to the target 3D model, ensuring that the candidate image content matches the regular texture map in terms of specifications, providing a suitable basic material for subsequent texture mapping. Taking an in-vehicle 3D model as an example, after the in-vehicle infotainment system successfully executes the voice command, it uploads the regular texture map (UV map) of the 3D model to the preset AI large model in PNG format. Simultaneously, it transmits the parsed semantic information for appearance updates, containing keywords such as the target visual style. After transmission, the in-vehicle infotainment system interface displays a "AI is editing the car cover" waiting prompt. Upon receiving the UV map and semantic information, the preset AI large model invokes its built-in AI UV car cover editor. Based on the keywords corresponding to the target visual style (such as yellow, bear, colorful, etc.), it searches for image resources matching the visual style. The searched image resources are then adjusted in size and outline to perfectly match the regular texture map (UV map), preventing subsequent texture deformation and misalignment, ultimately yielding the candidate image content. In general, the preset AI model can generate multiple (e.g., 2-10) candidate images at a time for users to select from, according to the preset configuration of the 3D car model modification module.

[0046] S340. Based on the target mapping method, the candidate image content is adapted to one or more planar regions to obtain one or more new regular texture maps.

[0047] In one possible implementation, according to preset target mapping rules, the candidate image content is adapted to one or more independent planar regions in the regular texture map obtained by unfolding the target 3D model. After fitting, alignment, and contour clipping, a new regular texture map matching the surface of the target 3D model is obtained. Taking a 3D car model as an example, the regular texture map is the UV map formed by unfolding the paint part of the 3D car model; the planar regions correspond to the paint surface regions in the UV map; the preset AI large model can call the AI ​​UV car wrap editor according to target mapping methods such as left-right symmetry, left-right inconsistency, and front-rear-left-right inconsistency, and call the AI ​​UV car wrap editor to fit the generated candidate image content to one or more corresponding paint planar regions (i.e., the planar regions corresponding to each paint surface) in the regular texture map (UV map) according to the specific rules of different mapping methods.

[0048] For example, if the mapping method is symmetrical, the default AI model copies a candidate image content into a symmetrical style and applies it to the two largest paint plane areas on the left and right sides of the UV map, while simultaneously cropping along the outline of the UV map. If the mapping method is inconsistent, the default AI model selects two different candidate images, obtains the center points of the two images, aligns the two center points with the center points of the paint plane areas on the left and right sides of the UV map, and then applies them. After applying them, it is cropped along the outline of the UV map to ensure that the content of the left and right car wraps is inconsistent. If the mapping method is inconsistent in all directions, the default AI model selects four different candidate images, obtains the center point of each image, aligns the four center points with the center points of the four paint plane areas on the front, back, left, and right sides of the UV map, and then applies them. Finally, it is cropped along the outline of the UV map.

[0049] It should be noted that after the candidate image content is adapted to the corresponding planar area and cropped, the original regular texture mapping map (UV map) is edited into a UV map with the new car owner effect. This new UV map is the "new regular texture mapping map". At the same time, the preset AI large model will generate one or more new regular texture mapping maps (consistent with the number of candidate image contents, usually 2-10) according to the number of candidate image contents. All new regular texture mapping maps are in PNG format and completely match the original regular texture mapping map and the paint parts of the 3D car model. After generation, the preset AI large model will send these new regular texture mapping maps back to the car model modification module in PNG format through the background for users to select and use. Subsequently, the car model 3D engine can directly apply the new regular texture mapping map to the paint parts of the 3D car model to complete the car cover customization.

[0050] The 3D model appearance update method provided in this application first determines the planar regions corresponding to each surface to be updated of the target 3D model based on a regular texture mapping map, clarifying the adaptation boundaries of the texture content and laying the foundation for subsequent texture content fitting, effectively avoiding problems such as pattern misalignment and stretching deformation during texture application; then, based on the appearance update semantic information input by the user, it determines the target visual style and target mapping method that meet the user's needs, thereby anchoring the appearance customization direction and meeting the diverse personalized customization needs of users; subsequently, based on the determined target visual style, it generates candidate image content that matches the size of the regular texture mapping map, ensuring that the outline and size of the candidate image content are highly consistent with the regular texture mapping map, allowing for direct adaptation without secondary size adjustment, greatly improving the efficiency of appearance texture generation; finally, based on the preset target mapping method, it adapts the candidate image content to the planar regions of one or more corresponding surfaces to be updated, generating one or more new regular texture mapping maps, which not only ensures the matching of texture content with the surface to be updated of the target 3D model, but also flexibly realizes the rapid generation of different personalized appearance effects.

[0051] Optionally, the target texture method is a mirrored mode. This mirrored mode can be selected as a left-right symmetrical texture logic or a front-back symmetrical texture logic. For ease of description, this mirrored mode is a left-right symmetrical mode to adapt to the symmetrical structure of the 3D car model body (such as left and right doors, front, rear, etc.), meet the user's personalized needs for symmetrical and aesthetically pleasing car covers, and solve the problem of cluttered car cover patterns and mismatch with the car model structure caused by traditional asymmetrical textures.

[0052] The above method, based on target mapping, adapts candidate image content to one or more planar regions, including: Based on the mirror mode, the position of the axis of symmetry, as well as the first and second planar regions, are determined in the regular texture mapping map.

[0053] In one possible implementation, based on the mirror pattern and using mirror symmetry as the processing principle, the corresponding axis of symmetry is first located on the regular texture map according to the inherent symmetry features of the target 3D model itself. Then, using the axis of symmetry as the dividing line, the entire regular texture map is divided into two sets of planar regions with mirror symmetry, which are denoted as the first planar region and the second planar region, respectively, thereby providing standardized partitions for subsequent mapping of symmetrical textures and symmetrical patterns.

[0054] Taking a 3D in-vehicle model as an example, based on the mirror mode, the vehicle's 3D engine module performs geometric analysis on the regular texture mapping map. Combining the actual symmetry parameters of the 3D model's body (such as the longitudinal centerline of the model, i.e., the vertical centerline from the front to the rear of the car), the axis that completely corresponds to the symmetry centerline of the model's body in the regular texture mapping map is located as the axis of symmetry in the mirror mode. The position of the axis of symmetry matches the symmetrical structure of the 3D model, ensuring that the subsequent filled pattern is symmetrical and consistent with the actual appearance of the model. Then, using the determined axis of symmetry as a clear dividing line, the regular texture mapping map is evenly divided into two completely symmetrical planar regions. The UV map region corresponding to the paint part of the left side of the 3D model's body is the first planar region, and the UV map region corresponding to the paint part of the right side of the 3D model's body is the second planar region. The size, shape, and geometric parameters of these two regions are completely consistent.

[0055] Based on the content of the candidate images, the basic patterns in the candidate image content are determined.

[0056] Among them, in the texture customization scenario for general 3D models, the candidate image content mainly includes two categories: one is the image content automatically generated by a preset AI large model according to the semantic analysis result of the user's demand instructions; the other is the user's own uploaded image resources. Taking a 3D vehicle model as an example, one is the image content generated by a preset AI large model according to the semantic analysis result of the user's voice instructions. For example, when the user says "want a simple striped car wrap", the preset AI large model will generate candidate image content containing striped elements; the other is the user's own uploaded image resources, such as personal photos, personalized patterns, etc.

[0057] In a possible implementation manner, the candidate image content is parsed and processed through a preset AI large model or a supporting image processing module, that is, first, the redundant information irrelevant to the surface texture of the 3D model in the image content is removed, and the effective pattern elements that meet the user's needs and can be used for regular laying on the model surface are retained; then, standard processing such as scale adjustment and clarity optimization is performed on the effective pattern elements to make them adapt to the regional size of the regular UV unfolding diagram of the target 3D model; finally, a basic pattern with the characteristics of being replicable, mirror-invertible, and adaptable to the UV region is obtained.

[0058] It should be noted that if the candidate image itself is already a single and pure target pattern, such as a single logo uploaded by the user or a single striped pattern generated by AI, etc., it can be directly determined as the basic pattern.

[0059] Taking a 3D vehicle model as an example, the candidate image content is parsed through a preset AI large model (or an image processing module supporting the in-vehicle system), that is, first, the redundant elements irrelevant to the car wrap pattern in the candidate image are filtered out, such as the background and irrelevant decorations in the user-uploaded photo, and the redundant textures in the AI-generated image, and only the pattern elements that can reflect the user's needs and can be used for symmetric filling are retained; then, standard processing is performed on the selected pattern elements, such as adjusting the pattern scale and optimizing the pattern clarity, to ensure adaptation to the regional size of the regular UV map, and finally, this pattern that has been screened and standardized is determined as the basic pattern. This basic pattern is the material for subsequent symmetric filling and must have the characteristics of being replicable, mirror-invertible, and adaptable to the UV map region.

[0060] Based on the position of the symmetry axis, the basic pattern is filled into the first plane region and the second plane region respectively.

[0061] In one possible implementation, a symmetry axis and a first planar region and a second planar region separated by the symmetry axis are pre-defined in the texture mapping map. Using the symmetry axis as a symmetry reference, the determined basic patterns are respectively placed on the two planar regions, so that the basic patterns form a symmetrical layout in the texture mapping map. This ensures that when the basic patterns are subsequently mapped to the surface of the three-dimensional model, they can match the symmetrical structure of the three-dimensional model without misalignment or deformation, thus achieving a standardized and symmetrical fit between the basic patterns and the surface of the three-dimensional model.

[0062] Taking a 3D in-vehicle model as an example, the 3D engine module of the in-vehicle infotainment system adjusts the proportions and sizes of the base pattern according to the actual dimensions of the first and second planar regions. This ensures that the size of the base pattern perfectly matches the two planar regions, avoiding issues such as pattern stretching, compression, and misalignment, thus achieving adaptation between the car wrap pattern and the UV map. Then, using a defined axis of symmetry as a mirror reference, the pre-processed base pattern is first filled into the first planar region, ensuring that the base pattern completely covers the first planar region without omissions or offsets, and that the orientation of the base pattern is consistent with the orientation of the paint area on the left side of the 3D car model. Subsequently, using the axis of symmetry as the center, the base pattern is further... The basic pattern is mirrored to ensure that it is symmetrical about the axis of symmetry with the original pattern, meeting the requirements of the mirror mode. Then, the mirrored basic pattern is filled into the second plane area, ensuring that it completely covers the second plane area without omissions or offsets, and is symmetrical about the axis of symmetry with the pattern in the first plane area. Finally, after filling, the regular texture map is rendered and verified in real time by the 3D engine module to confirm that the filling position of the basic pattern in the two plane areas is consistent, symmetrical, and without deformation or offset. This ensures that the UV map after filling can fit the left and right paint parts of the 3D car model, laying the foundation for subsequent application of the car cover pattern to the 3D car model and completion of the car cover customization.

[0063] The 3D model appearance update method provided in this application adopts a mirror mode as the target texture mapping method, which can determine the symmetry axis position for the regular texture mapping map corresponding to the 3D model, and divide the first plane region and the second plane region corresponding to the two sides of the 3D model, clarifying the symmetry reference and region boundary, avoiding pattern misalignment and deformation caused by blurred region division; the candidate image content is analyzed to extract the basic pattern that meets the user's needs, ensuring that the filled basic pattern is highly consistent with the user's expected visual effect; finally, based on the determined symmetry axis position, the basic pattern is filled into the corresponding first plane region and the second plane region respectively, realizing the symmetrical presentation of the basic pattern on the left and right sides of the 3D model, improving the overall visual coordination and aesthetics of the texture, while ensuring the adaptation of the basic pattern and the regular texture mapping map, laying the foundation for the rapid texture bonding of the subsequent 3D model.

[0064] Optionally, the target texture mapping method is the first non-mirror mode; the first non-mirror mode is a mode that mirrors the different surfaces to be updated. Here, the first non-mirror mode can be understood as a left-right asymmetric mode. Taking a 3D car model as an example, the patterns of the car wraps applied to these two sets of symmetrical mirrored car paint surfaces must be different, and mirrored or completely identical patterns are not allowed.

[0065] The above method, based on target mapping, adapts candidate image content to one or more planar regions, including: Based on the first non-mirror mode, the first planar region and the second planar region are determined.

[0066] In one possible implementation, under the non-mirror symmetric layout rules of the target 3D model, the surface of the target 3D model is divided and independently defined into symmetrical regions. Two mutually symmetrical and independent planar regions on the target 3D model are identified and defined as the first planar region and the second planar region, respectively, providing a basis for the subsequent loading of different texture patterns on the two regions.

[0067] Taking a 3D in-vehicle model as an example, the 3D engine module of the in-vehicle infotainment system is activated to perform a comprehensive scan of the 3D model's paint surface. The mirrored paint surfaces, symmetrical on the left and right sides of the vehicle, are identified and divided into two independent planar regions using the vehicle's central axis as the dividing line: the first planar region and the second planar region. These two regions are then positioned and marked, clearly defining their boundaries. The first planar region corresponds only to all paint surfaces on the left side of the vehicle, and the second planar region corresponds only to all paint surfaces on the right side of the vehicle. This ensures that the two regions do not overlap or intersect, and that they perfectly match the actual size and shape of the model's paint surface, preventing misalignment or omissions during subsequent pattern filling and preparing for the filling of different patterns.

[0068] Based on the content of the candidate images, the first pattern and the second pattern are determined.

[0069] The first pattern is different from the second pattern.

[0070] In one possible approach, to address the asymmetric texture customization needs of the target 3D model, based on the acquired candidate image materials, a first pattern and a second pattern are determined to fit different sections of the 3D model. It is ensured that the two patterns have significant differences in content, style, color, and detail, and that they are not identical or mirror copies. Taking an in-vehicle 3D model as an example, based on the requirement of the first non-mirror mode (i.e., different paint patterns on the left and right mirrored surfaces), the user can manually select two different patterns (i.e., the first and second patterns) from the candidate image content, or the in-vehicle infotainment system can automatically recommend different patterns based on user preferences. Simultaneously, it is ensured that the determined first and second patterns are sized to fit the left and right planar areas of the 3D model, avoiding stretching and deformation during subsequent filling, ultimately satisfying the user's personalized asymmetric customization needs for the 3D model's appearance.

[0071] The first pattern is filled into the first planar area, and the second pattern is filled into the second planar area.

[0072] In one possible implementation, for different independent planar regions pre-divided in the target 3D model, the first and second patterns are precisely matched and filled into their respective planar regions to achieve differentiated pattern layout for different planar regions of the 3D model. Taking a 3D in-vehicle model as an example, the first and second patterns are imported into the texture tool of the 3D engine module of the in-vehicle infotainment system. The size, proportion, and angle of the two patterns are adjusted using the texture tool to ensure they perfectly match the actual size and shape of the previously determined first and second planar areas, avoiding problems such as stretching, compression, and misalignment after pattern filling. Next, according to the requirements of the first non-mirror mode, the texture filling command is initiated, and the 3D engine module fills the first pattern into the marked first planar area. During the filling process, it is ensured that the pattern completely covers all the paint surfaces of the first planar area, without omissions or gaps, and that the pattern conforms to the curvature and texture of the paint surface, presenting a realistic car wrap effect. At the same time, the 3D engine module fills the second pattern into the marked second planar area, with the filling standard consistent with the first pattern, ensuring that the second pattern completely covers the second planar area and presents different visual effects on the left and right paint surfaces compared to the first pattern. Finally, after the filling is completed, the 3D engine module performs real-time rendering to confirm that the pattern filling is complete and without abnormalities, thus completing the entire car wrap texture operation in the first non-mirror mode, ultimately achieving personalized customization of different patterns on the left and right paint surfaces of the 3D car model.

[0073] The 3D model appearance update method provided in this application adopts a first non-mirror mode as the target texture mapping method. That is, the first non-mirror mode uses a texture mapping mode with different patterns for each mirror surface to be updated of the 3D model, which is different from the conventional texture mapping method with traditional mirror symmetry and homogeneous surface patterns. Based on the first non-mirror mode, the first plane region and the second plane region of the 3D model are determined, and the first pattern and the second pattern are determined according to the content characteristics of the candidate image. The first pattern is filled into the first plane region and the second pattern is filled into the second plane region. This can effectively break the problem of single, repetitive and homogeneous patterns on each surface of the 3D model caused by the traditional mirror texture mapping mode, realize the differentiated and personalized customization of the mirror plane region of the 3D model, significantly improve the uniqueness, visual hierarchy and diversity of the 3D model appearance, and meet the user's needs for refined, personalized and diversified customization of the 3D model appearance.

[0074] Optionally, the target texture mapping method is the second non-mirror mode. The second non-mirror mode is a mode where each surface to be updated is different. It is used for asymmetric, non-mirror, and differentiated appearance customization of the target 3D model. In this mode, different areas of the 3D model are not required to use the same or mirrored pattern. Taking a 3D car model as an example, the second non-mirror mode is a mode where each paint surface (such as the hood, roof, front doors, rear doors, trunk lid, fenders, and all other textureable independent paint surfaces) is different. In this mode, the car cover pattern applied to each independent paint surface is unique and non-mirrored; each paint surface corresponds to a unique personalized pattern. Here, the second non-mirror mode can be understood as a mode where the front, rear, left, and right sides are all different.

[0075] The above method, based on target mapping, adapts candidate image content to one or more planar regions, including: Based on the second non-mirror mode, multiple planar regions are determined.

[0076] In one possible implementation, based on the second non-mirror mode, multiple physical regions on the surface of the target 3D model that can be independently textured are transformed into multiple 2D planar carriers that are different from each other in terms of contour, size, UV coordinate range, and corresponding model parts through 3D to 2D mapping processing. This allows each independent planar region to be allocated a dedicated texture space, enabling different patterns to be loaded on different planar regions, ultimately achieving a non-mirror, multi-regional differentiated appearance customization effect on the surface of the 3D model.

[0077] Taking a 3D in-vehicle model as an example, after the in-vehicle infotainment system is set to the second non-mirror mode, the system automatically calls the modeling structure data of the 3D model, extracts all textured independent paint surfaces, clarifies the spatial position, outline size, and boundary range of each paint surface, and excludes non-texturable areas (such as windows, headlights, wheels, and other non-painted areas). Then, each independent paint surface is planarized. Since the paint surfaces of the 3D model are mostly curved, they cannot be directly filled with patterns. Therefore, based on the previously extracted regularized UV map, each independent paint surface is transformed into a corresponding planar region. That is, through UV mapping technology, the coordinates of the curved paint surface are transformed into... Planar coordinates are used to eliminate the influence of curved surface curvature on pattern filling, ensuring that the pattern is not stretched or deformed after filling. Finally, the converted planar areas are identified and distinguished, with each planar area assigned a unique identifier (e.g., planar area 1 corresponds to the hood, planar area 2 corresponds to the roof, etc.). At the same time, parameters such as the size, scale, and UV coordinate range of each planar area are recorded to ensure that each planar area corresponds one-to-one with an independent paint surface of the 3D car model, and that multiple planar areas do not overlap and have clear boundaries. This lays the foundation for the subsequent filling logic of one pattern corresponding to one planar area. Essentially, it is to transform multiple independent paint surfaces of the 3D car model into multiple planar carriers that can be individually textured.

[0078] Based on the content of the candidate images, multiple third patterns corresponding to each planar region are determined.

[0079] In one possible implementation, based on the acquired candidate images, multiple third patterns are matched and determined one-to-one with each independent planar region divided on the target 3D model, ensuring that each third pattern fits its corresponding planar region. First, the candidate image content is processed according to the number, position, and size ratio of the planar regions divided in the target 3D model. If the candidate image is a single image, multiple different patterns matching the number of planar regions and their size ratios are extracted from it through cropping, scaling, and segmentation. Each pattern is then mapped to a planar region to form multiple third patterns corresponding to each planar region. If there are multiple candidate images, multiple patterns matching the number of planar regions, having a unified style, or meeting user requirements are selected. These selected patterns are then fine-tuned using a preset AI model to ensure that the pattern size and ratio match the parameters of the corresponding planar regions, thus obtaining multiple third patterns one-to-one with each planar region.

[0080] Fill each of the third patterns into the corresponding planar areas.

[0081] In one possible implementation, for multiple independent planar regions divided from the target 3D model, a one-to-one correspondence is established between each third pattern and each planar region. Each third pattern is independently applied to the single planar region that matches it, so that each planar region is completely covered by its own third pattern, thereby achieving a texture effect that presents different patterns in different regions of the target 3D model. Taking a 3D in-vehicle model as an example, the planar area corresponding to each third pattern can be determined by either manual allocation by the user (e.g., the planar area displayed on the vehicle interface is manually associated with the third pattern) or automatic allocation by the in-vehicle system (e.g., automatically matching the position and size of the planar area with the style and size of the third pattern). Subsequently, the texture mapping function of the 3D engine module is called to align the pixel data of each third pattern with the UV coordinates of the corresponding planar area, and adjust the scaling ratio and rotation angle of the third pattern to ensure that the third pattern completely fits the planar area after filling, without stretching, offset, or boundary overflow. At the same time, based on the rules of the second non-mirror mode, the in-vehicle system automatically verifies the filling results to ensure that the third pattern filled in each planar area is different, without repetition or mirroring, ultimately achieving a personalized car wrap texture effect with different paint surfaces on the 3D car model, completing the car wrap customization process.

[0082] The 3D model appearance update method provided in this application adopts a second non-mirror mode as the target texture mapping method. This second non-mirror mode is different for each surface to be updated. This allows the 3D model to be textured by first dividing and defining multiple independent planar regions based on this second non-mirror mode, then determining a corresponding third pattern for each planar region according to the candidate image content, and finally filling each third pattern into the corresponding planar region one-to-one. Therefore, this application avoids the problems of pattern repetition and monotonous styles caused by traditional mirrored or uniform texture mapping, enabling each surface to be updated in the 3D model to present independent and different pattern effects. Without relying on specific equipment or vehicle models, it achieves refined, differentiated, and personalized customization of the 3D model's appearance, effectively improving the visual richness and overall expressive effect of the 3D model's appearance.

[0083] Optionally, after adapting the candidate image content to one or more planar regions based on the target mapping method, the above method further includes: The candidate image content is adapted to one or more planar regions to obtain the initial appearance image.

[0084] In one possible implementation, based on the surface unfolding structure of the target 3D model, one or more planar regions for carrying the appearance texture are first determined. Then, candidate image content is matched and adjusted with these planar regions in terms of size, scale, and position, so that the image content can correspondingly cover the corresponding planar regions, thus obtaining a preliminary appearance image. Taking a 3D vehicle model as an example, the size, scale, and other parameters of the candidate image content can be obtained first. Then, combined with the size and contour parameters of each selected planar region, image processing operations such as scaling, stretching, translation, and alignment are used to adjust the candidate image content to perfectly match the size and scale of each target planar region, ensuring that the candidate image content can completely and closely cover the corresponding planar region. If multiple planar regions are to be adapted, the above adaptation operation is performed on each planar region separately. The same candidate image content can be adapted to multiple planar regions, or different candidate image contents can be adapted to different planar regions. After adaptation, all adapted planar region images are integrated to form the initial appearance image. The initial appearance image already has personalized content and a planar shape of the car model's paint area, but it has not been contoured and still contains redundant image parts that exceed the actual contour of the car model's paint area, providing a basis for subsequent cropping steps.

[0085] Based on the outline of the regular texture map, the initial appearance image is cropped to generate one or more new regular texture maps.

[0086] In one possible implementation, the generated initial appearance image is aligned with the outline of a pre-constructed regular texture map of the target 3D model as the standard boundary. Then, the initial appearance image is precisely cropped according to the outline range to remove redundant image content that exceeds the outline range. This ensures that the cropped appearance image completely matches the regular texture map in shape, size, and boundary, ultimately forming one or more standardized new regular texture maps that can be directly attached to the surface of the 3D model.

[0087] Taking a 3D car model as an example, the outline of the regular texture mapping map is completely consistent with the actual shape outline of the paint part of the 3D car model. It includes all detailed outline information such as the boundary lines, corner curvature, and hollow areas (if any) of each paint part. Moreover, the outline parameters (such as outline coordinate points, boundary length, curvature value, etc.) have been calibrated to ensure that they are completely matched with the actual size and shape of the paint part of the 3D car model. Next, the complete contour data of the regular texture map is extracted and imported into the image cropping module. Then, the previously generated initial appearance image is aligned with the contour data (e.g., the alignment reference is the position coordinates of the planar area to ensure that the position of each planar area in the initial appearance image completely coincides with the contour position of the corresponding paint part in the regular texture map). Then, the initial appearance image is cropped with the contour of the regular texture map as the boundary, removing the excess image parts that exceed the contour boundary in the initial appearance image. At the same time, the edges of the cropped image are smoothed to correct edge jaggedness, misalignment and other problems that may occur during the cropping process, ensuring that the contour of the cropped image is completely consistent with the contour of the regular texture map and fits the actual shape of the paint part of the 3D car model. Finally, one or more new regular texture maps are generated. The specific number generated is consistent with the correspondence between the number of planar regions adapted in the previous step and the regular texture maps: if multiple planar regions were adapted in the previous step, and each planar region corresponds to an independent regular texture map outline, then multiple new regular texture maps are generated after cropping, each corresponding to a paint part; if multiple planar regions were adapted in the previous step, but these planar regions correspond to the same regular texture map outline (such as multiple paint parts integrated into a single planar region), then one new regular texture map containing all adapted planar regions is generated after cropping. This new regular texture map is a standardized final texture map that can be directly used for 3D car model car wrap application. It can be directly transferred to the 3D engine module and applied to the corresponding paint parts of the 3D car model to complete the generation of a personalized car wrap.

[0088] The 3D model appearance update method provided in this application obtains an initial appearance image by selectively adapting candidate image content to one or more preset planar regions, thereby matching the candidate image with the geometric shape and size parameters of the corresponding planar region and reducing distortion problems such as image stretching and distortion. Then, based on the contour boundary of the regular texture map, the initial appearance image is cropped to generate one or more new regular texture maps, so that the contour of the new regular texture map completely matches the target texture region, improving the regularity and consistency of the 3D model's appearance texture, and providing high-precision basic materials for subsequent texture mapping and fitting on the 3D model.

[0089] Optionally, after generating one or more new regular texture maps in the above method, the method further includes: Receive user instructions to modify a selected new rule texture map.

[0090] In one possible implementation, after generating multiple regular texture maps that adapt to the target 3D model, the user selects one of them as the object to be adjusted. The host computer or interactive system obtains the user's adjustment requirements for the texture map through the interactive interface, so as to achieve fine customization of the texture effect and make the final presentation effect more in line with the user's expectations.

[0091] Taking a 3D in-vehicle model as an example, the in-vehicle infotainment system can connect to the user interface (such as a car cover texture design platform or a visual editing interface) through its built-in interactive receiving module. This allows the system to capture user input signals in real time, ensuring that modification commands are not lost or deviated. Then, the target user selects one of several new rule texture maps displayed by the in-vehicle infotainment system as the target for modification through interface interactive elements (such as clicking, checking, or dragging). The system records the unique identifier of this map (such as file ID or texture number) using a selection recognition module, ensuring that subsequent modification commands correspond to the target map. Finally, the target user submits specific adjustment requests for the selected new rule texture map. The modification commands can be implemented in various ways (such as interface parameter input, brush editing, texture attribute modification, text command input, etc.). For example, the target user can modify commands such as "adjust the texture spacing to 2cm," "change the texture color from black to dark gray," or "add texture symmetry." The interactive receiving module will perform preliminary analysis of these modification commands, extracting information such as the modification type, parameters, and scope, providing data support for subsequent texture map regeneration or optimization.

[0092] Based on the modification instructions, the regular texture map, and the semantic information of appearance update, one or more updated new regular texture maps are generated.

[0093] In one possible implementation, the user's modification instructions are used as the basis for adjustment, the original regular texture map is used as the texture structure and layout, and the semantic information of appearance update is used as the constraint. Under the premise of keeping the texture mapping rules unchanged, the two-dimensional regular texture map corresponding to the target three-dimensional model is modified and optimized in a targeted manner, and finally one or more updated regular texture maps that meet the user's needs and can be directly attached to the surface of the three-dimensional model are output.

[0094] Taking a 3D in-vehicle model as an example, the modification instruction is the user's specific adjustment requirements received and parsed, serving as the guiding basis for the generation action and clarifying "what to modify and how to modify it"; the rule texture mapping map is used to ensure that the modification process does not deviate from the rule attributes of the original texture; the appearance update semantic information refers to the constraint and guidance semantic content related to the generation of the car cover (such as the vehicle model parameters that the car cover is compatible with, the body surface dimensions, texture splicing rules, the texture display requirements corresponding to the car cover material, texture compliance standards, etc.). Its function is to ensure that the generated update map can adapt to the actual generation requirements of the subsequent car cover, and to avoid the modified texture being unable to be applied to the production of the car cover. The in-vehicle infotainment system will load this semantic information in advance and incorporate it as a constraint into the generation process.

[0095] The in-vehicle infotainment system has a built-in texture mapping generation module. This module first integrates the input elements of modification instructions, regular texture mapping maps, and appearance update semantic information. It then performs secondary parsing of the modification instructions and, combined with the appearance update semantic information, determines whether the modification instructions conform to the car cover generation specifications (e.g., whether the texture spacing in the user's instructions is adapted to the vehicle body size; if not, it makes reasonable corrections). Based on the parameters of the original regular texture mapping map, it executes specific modification operations (e.g., adjusting texture pixel coordinates, modifying color channels, optimizing parameters at texture splicing points, etc.). Finally, if the user's modification instructions are for a single, clear requirement (e.g., only adjusting texture color), the texture mapping generation module outputs only one updated map that meets all requirements. If the user's modification instructions include multiple options (e.g., providing three alternative colors when adjusting texture color), or if the in-vehicle infotainment system generates multiple optimized solutions that meet the requirements based on the modification instructions and appearance update semantic information, it will output multiple updated maps for the target user to further select from.

[0096] The 3D model appearance update method provided in this application, based on the modification command issued by the target user for the selected new rule texture map, and then based on the modification command, the original new rule texture map, and appearance update semantic information, performs targeted texture parameter adjustments and pattern reconstruction to generate one or more updated new rule texture maps. Thus, this application can address the user's personalized modification needs, ensuring that the updated map not only matches the user's customization preferences but also maintains a high degree of consistency with the appearance update semantics, while guaranteeing accurate adaptation between the texture and the 3D model. This effectively improves the flexibility and efficiency of texture customization, reduces repetitive design costs, and provides reliable support for the efficient and accurate updating of the 3D model's appearance.

[0097] Figure 4 A flowchart illustrating a method for updating the appearance of a 3D model provided in this application embodiment. Figure 3 .like Figure 4 As shown, the above method parses the appearance update instruction to obtain appearance update semantic information, including: S410: Perform parsing on the appearance update command to obtain the corresponding customization intent and attribute information.

[0098] In one possible approach, the appearance update command text is first preprocessed. This preprocessing includes at least word segmentation (breaking long sentences into core terms such as "TPU material," "Tesla Model Y," "matte," "scratch-resistant," and "starry sky pattern"), noise reduction (removing redundant expressions like "make one for me" and "want"), and standardization (unifying colloquial terms like "frosted surface" into the industry-standard term "matte material"). Then, based on a pre-trained Natural Language Understanding (NLU) model and a professional corpus of 3D model appearance customization, deep semantic analysis of the appearance update command is performed, laying the foundation for subsequent intent recognition and attribute extraction. The next step is to construct a pre-defined 3D model appearance customization intent classification system. This system covers core categories such as decorative appearance, enhanced protection, personalized theme customization, and 3D model-specific adaptation. Finally, an intent recognition model fine-tuned from the 3D model appearance customization corpus (such as BERT (Bidirectional Encoder Representations from...)) is used. The Transformers (bidirectional Transformer encoder representation) class model is used to perform similarity matching between the parsed appearance update command text and the intent categories in the appearance customization intent classification system of the 3D model. The category with the highest similarity is selected as the user's core customization intent. For example, if the user's appearance update command is "scratch-resistant matte black car wrap with racing stripes", it will be matched as "combined intent of protection and appearance decoration". Finally, a Named Entity Recognition (NER) model is trained by annotating the appearance customization domain of the 3D model. This enables it to recognize and extract key entity words in the appearance update command text. These words correspond to the core attribute dimensions of the 3D model's appearance, which may include: the 3D model's adaptation attributes (such as the compatible car model and year), material attributes (such as TPU (Thermoplastic Urethane) and PVC (Polyvinyl)). Chloride (polyvinyl chloride), matte / glossy finish, color attributes (such as black, gradient pink, starry blue), pattern attributes (such as carbon fiber, racing stripes, cartoon patterns), and functional attributes (such as scratch resistance, UV protection, self-healing). Simultaneously, for ambiguous expressions in appearance update instructions (such as "new energy vehicle models"), reasonable supplementation is performed using a library of common compatible 3D models for appearance customization, ultimately forming a structured set of customization intent and attribute information.

[0099] S420. Generate appearance update semantic information based on customized intent and attribute information.

[0100] The semantic information for appearance updates includes at least: visual style information for controlling image content and texture information for controlling image layout. Visual style information transforms the visually relevant dimensions of attribute information into quantifiable visual parameters recognizable by the image generation model, thereby controlling the presentation of the 3D model's appearance. For example, for material attributes (such as matte, glossy, and carbon fiber textures), it is transformed into gloss parameters (e.g., matte corresponds to a gloss range of 0.2-0.4, gloss to 0.8-1.0), and texture density and direction parameters (e.g., carbon fiber texture corresponds to a density of 15-20 lines / cm, with the texture direction along the length of the vehicle body). For color attributes (such as gradient pink and starry blue), it is transformed into color space parameters (e.g., RGB value range and color gradient transition parameters). Simultaneously, the visual parameter tendencies are adjusted in conjunction with the customization intent. For example, under a protective function intent, the visual performance parameters of material thickness are increased, and the color saturation of fancy patterns is weakened; the opposite is true for personalized customization intent. Ultimately, this forms a set of style information that can control the visual presentation of the 3D model's appearance.

[0101] Among them, the texture information of the image layout is based on the vehicle model adaptation, pattern position and other content in the attribute information to construct the layout control parameters of the appearance image of the 3D model, thereby controlling the element layout effect of the appearance image of the 3D model. For example, first, the baseline layout template of the corresponding 3D model is retrieved based on the adaptation attributes of the 3D model. This baseline layout template is built based on the CAD drawings of the 3D model. Taking a 3D vehicle model as an example, the CAD drawings of the 3D vehicle at least include the precise coordinates and size scale of the body areas such as the hood, roof, doors, and side skirts. Then, combined with the pattern and functional area requirements in the attribute information, the layout parameters of each texture element are determined. The layout parameters include at least the texture position (such as the coordinate range of the side skirt area corresponding to "carbon fiber side skirt"), texture size (such as a 1:1 scale adaptation with the actual area of ​​the 3D model), texture layer level (such as pattern textures located on the basic color of the 3D model's appearance, and functional logo textures located on top of pattern textures), and texture alignment rules (such as horizontal alignment along the waistline of the 3D model's appearance). At the same time, layout constraints are set, such as the textures not covering the body functional areas such as windows and door handles. Finally, texture information that can control the layout of the appearance image elements of the 3D model is formed.

[0102] In one possible implementation, a pre-built mapping rule library of "customization intent - attribute information - appearance update semantic information" is used. This mapping rule library presets the generation priority of appearance update semantic information for different customization intent types. For example, personalized decoration intent prioritizes strengthening the uniqueness of visual style, while functional protection intent prioritizes highlighting the functional visual features of materials. During the generation stage, the structured customization intent and attribute information obtained in the preceding stage are input into the mapping rule library to trigger the corresponding generation logic, thereby outputting appearance update semantic information containing visual style information and texture information, ensuring that the appearance update semantic information can meet the user's customization needs.

[0103] The 3D model appearance update method provided in this application first performs parsing processing on appearance update commands to extract the hidden customization intent and core attribute information, thereby improving the accuracy and completeness of appearance update command parsing and effectively avoiding problems such as misinterpretation of intent and omission of attributes. On this basis, it further generates semantic information specific to the appearance of the 3D model. This semantic information must at least cover visual style information used to control the presentation effect of image content and texture information used to standardize the layout logic of the image. The visual style information can ensure that the visual presentation of the 3D model's appearance is highly consistent with the user's customization needs, enhancing the personalized expression effect, while the texture information can anchor the key nodes of the image layout, improving the standardization and controllability of the 3D model's appearance image generation, and providing reliable semantic guidance for the efficient generation of subsequent 3D model appearance images.

[0104] Figure 5 A flowchart illustrating a method for updating the appearance of a 3D model provided in this application embodiment. Figure 4 .like Figure 5 As shown, the above method also includes: S510. Obtain the three-dimensional model data of the target three-dimensional model.

[0105] In one possible approach, if a ready-made 3D model file (such as OBJ, FBX, STL, or other common 3D model formats) already exists for the target 3D model, it can be directly imported using professional 3D data reading software (such as Blender, 3ds Max, Catia, etc.) to extract the 3D model data contained within. If no ready-made 3D model file exists, a 3D scanning device (such as a laser 3D scanner or a structured light 3D scanner) can be used to perform a full-range scan of the real target 3D model, collecting 3D spatial coordinate information, geometric contour information, etc., of the 3D model surface, and then generating the corresponding 3D model data.

[0106] It should be noted that, regardless of which method is used, the acquired 3D model data must completely cover all components of the target 3D model. For example, a 3D car model should at least include the geometric information of paint, glass, tires, headlights, interior, etc. This geometric information should at least include the vertex coordinates, face elements (such as triangles or quadrilaterals), vertex normals, topological relationships, etc.

[0107] S520. Based on the 3D model data, extract the mesh data of each surface to be updated.

[0108] In one possible approach, all components of the 3D model are distinguished based on the attribute labels, material information, or geometric features of each component in the 3D model data. For example, since each surface to be updated (such as car paint) usually has specific material properties (such as smoothness and reflectivity), components with the material type "surface paint to be updated" in the 3D model data can be filtered out, or non-painted components such as vehicle glass, tires, headlights, and door handles can be excluded based on the preset geometric regions corresponding to the surfaces to be updated (such as car body shells, doors, hoods, trunk lids, etc., which are areas covered by car paint). Secondly, after distinguishing all surface components to be updated, the complete mesh data corresponding to these components is extracted. The extracted mesh data of the surface to be updated must at least include: the vertex spatial coordinates of the surface to be updated, the connection relationship of all triangles (or quadrilaterals) that make up the surface to be updated, the normal direction of the facets, and the topological structure of the mesh. This data is the core foundation for subsequent texture mapping generation. It is necessary to ensure that the extracted mesh data is complete, undamaged, and contains only mesh information related to the surface to be updated, without mixing in mesh data of other components.

[0109] S530. Unfold the mesh data onto a two-dimensional plane to obtain the initial texture map.

[0110] In one possible implementation, a 3D mesh parameterization (also known as UV unwrapping) technique is used to unwrap the 3D mesh of the surface to be updated. This involves establishing a one-to-one correspondence between each vertex of the 3D mesh data of the surface to be updated and a coordinate point on the 2D plane (i.e., UV coordinates), eliminating the spatial curvature of the 3D mesh data and tiling it onto the 2D plane. During the unwrapping process, it is necessary to avoid problems such as excessive stretching, folding, and distortion of the mesh data. This can be achieved by dividing the mesh (e.g., setting dividing lines at the turning points and seams of the vehicle body) and adjusting the UV coordinate allocation to reduce deformation after unwrapping, ensuring that the original outline and details of the mesh of the surface to be updated are well preserved on the 2D plane. After unwrapping, an initial texture mapping map can be generated.

[0111] It should be noted that the initial texture map is such that each region on the two-dimensional plane corresponds one-to-one with a certain mesh region on the three-dimensional surface to be updated. However, the initial texture map is usually irregular (i.e. its shape is consistent with the unfolded shape of the three-dimensional mesh of the surface to be updated), and the planar regions corresponding to each surface to be updated are arranged irregularly. It only completes the goal of converting three-dimensional to two-dimensional without normalization.

[0112] S540. Normalize the initial texture map to obtain a regular texture map.

[0113] The standardization process involves ensuring that the planar regions corresponding to each surface to be updated are arranged regularly in a two-dimensional space along a preset direction. This preset direction can be selected based on the specific circumstances; for example, it can be uniformly arranged horizontally, uniformly arranged vertically, or along a preset grid line. Regular arrangement means that the two-dimensional regions corresponding to each surface to be updated must be arranged neatly and orderly, without any disorder, overlap, or misalignment. This can be achieved through methods such as equal spacing, aligned arrangement (e.g., left-aligned, top-aligned), or uniform distribution to ensure the regularity of the arrangement.

[0114] In one possible implementation, the two-dimensional regions corresponding to each surface to be updated in the initial texture mapping map are first identified and segmented to clarify the position of the three-dimensional surface to be updated corresponding to each planar region (such as the two-dimensional regions corresponding to the hood, doors, and trunk of a vehicle's three-dimensional model). Then, according to preset normalization rules, each segmented two-dimensional region is adjusted. This adjustment includes at least: adjusting the size, angle, and position of the region, eliminating overlaps and disordered arrangements between regions, to ensure that all two-dimensional regions corresponding to the surfaces to be updated are arranged in an orderly manner along a preset direction (such as horizontal or vertical). During the adjustment process, the one-to-one correspondence between each two-dimensional region and the corresponding three-dimensional paint surface mesh must remain unchanged to ensure that subsequent texture drawing and editing based on the regular texture mapping map can accurately map back to the three-dimensional surface mesh to be updated. After all adjustments are completed, a regular texture mapping map is formed. The characteristic of this regular texture mapping map is that the two-dimensional regions corresponding to each surface to be updated are arranged neatly and regularly, facilitating subsequent texture design, modification, and reuse operations, improving the efficiency and accuracy of texture mapping. The preset normalization rules can be selected according to the specific situation, and will not be elaborated here.

[0115] The 3D model appearance update method provided in this application first acquires the 3D model data of the target 3D model, laying the data foundation for subsequent mesh data extraction and texture mapping of the surfaces to be updated. Then, it extracts the mesh data of all surfaces to be updated, effectively eliminating interference from non-painted structures on the surfaces. Next, it unfolds the mesh data of the surfaces to be updated onto a 2D plane to generate an initial texture mapping map, completing the spatial transformation from a 3D surface to a 2D plane, providing an intuitive carrier for texture editing and optimization. Finally, it performs normalization processing on the initial texture mapping map, arranging the planar regions corresponding to all surfaces to be updated in a regular manner within the 2D plane according to a preset direction, forming a regular texture mapping map. This improves the orderliness and consistency of the regular texture mapping map, facilitating subsequent texture drawing, texture reuse, and batch adjustment, thereby improving the efficiency and standardization of 3D model texture production.

[0116] Based on the same inventive concept, this application also provides a three-dimensional model appearance update device. Since the principle of the device in this application to solve the problem is similar to the three-dimensional model appearance update method described above in this application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0117] Figure 6 This is a schematic diagram of a three-dimensional model appearance updating device provided in an embodiment of this application. Figure 6 As shown, the appearance update device 600 for the 3D model may include: The receiving module 601 is used to receive the appearance update instruction for the target 3D model and obtain the corresponding regular texture mapping map, wherein the regular texture mapping map is a two-dimensional image generated by unfolding the surface to be updated of the target 3D model. The parsing module 602 is used to parse the appearance update command to obtain appearance update semantic information; The generation module 603 is used to generate one or more new regular texture maps based on the regular texture map and appearance update semantic information; The bonding module 604 is used to bond the selected new regular texture map to the surface corresponding to the target 3D model in response to the user's selection operation of one or more new regular texture maps, so as to update the appearance of the target 3D model.

[0118] In one optional implementation, the generation module 603 is specifically used for: determining the planar regions corresponding to each surface to be updated of the target 3D model based on the regular texture map; determining the target visual style and target mapping method based on appearance update semantic information; generating candidate image content that matches the size of the regular texture map based on the target visual style; and adapting the candidate image content to one or more planar regions based on the target mapping method to obtain one or more new regular texture maps.

[0119] In one optional implementation, the target texture mapping method is a mirror mode; the generation module 603 is specifically used to: determine the position of the axis of symmetry and the first planar region and the second planar region in the regular texture mapping based on the mirror mode; determine the basic pattern in the candidate image content based on the candidate image content; and fill the first planar region and the second planar region respectively based on the position of the axis of symmetry.

[0120] In one optional implementation, the target mapping method is a first non-mirror mode; the first non-mirror mode is a different mode of mirroring the surface to be updated; the generation module 603 is specifically used for: determining a first planar region and a second planar region based on the first non-mirror mode; determining a first pattern and a second pattern based on the candidate image content, wherein the first pattern and the second pattern are different; filling the first pattern into the first planar region, and filling the second pattern into the second planar region.

[0121] In one optional implementation, the target mapping method is a second non-mirror mode; the second non-mirror mode is a mode in which the paint surfaces of multiple vehicles are different; the generation module 603 is specifically used to: determine multiple different planar regions based on the second non-mirror mode; determine multiple third patterns corresponding to each planar region based on the candidate image content; and fill each third pattern into each corresponding planar region.

[0122] In an optional implementation, the generation module 603 is further configured to: adapt the candidate image content to one or more planar regions to obtain an initial appearance image; and crop the initial appearance image based on the contour of the regular texture map to generate one or more new regular texture maps.

[0123] In one optional implementation, the generation module 603 is further configured to: receive a user's modification instruction for a selected new regular texture map; and generate one or more updated new regular texture maps based on the modification instruction, the regular texture map, and appearance update semantic information.

[0124] In one optional implementation, the parsing module 602 is specifically used to: perform parsing processing on the appearance update instruction to obtain the corresponding customized intent and attribute information; and generate appearance update semantic information based on the customized intent and attribute information; wherein the appearance update semantic information includes at least: visual style information for controlling image content and texture information for controlling image layout.

[0125] In one optional implementation, the appearance update device 600 for the three-dimensional model is further configured to acquire three-dimensional model data of the target three-dimensional model; extract mesh data of each surface to be updated based on the three-dimensional model data; unfold the mesh data to a two-dimensional plane to obtain an initial texture mapping map; and perform normalization processing on the initial texture mapping map to obtain a regular texture mapping map; wherein, the normalization processing is to make the planar regions corresponding to each surface to be updated regularly arranged in a preset direction in the two-dimensional plane.

[0126] It should be noted that for details not disclosed in the appearance updating device of the three-dimensional model in the embodiments of this application, please refer to the details disclosed in the appearance updating method of the three-dimensional model in the embodiments of this application, which will not be repeated here.

[0127] These modules can be one or more integrated circuits configured to implement the above methods, such as one or more Application Specific Integrated Circuits (ASICs), one or more microprocessors, or one or more Field Programmable Gate Arrays (FPGAs). Alternatively, when a module is implemented using processing element scheduler code, the processing element can be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. Furthermore, these modules can be integrated together as a system-on-a-chip (SOC).

[0128] Optionally, embodiments of this application also provide a computer-readable storage medium storing a computer program. When the computer program is run by a processor, the processor executes the steps of the appearance update method for the 3D model of the removable storage medium described in the above embodiments. The specific implementation and technical effects are similar and will not be repeated here.

[0129] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the functional units in the various embodiments of this application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The integrated unit described above can be implemented in hardware or in the form of hardware plus software functional units.

[0130] Optionally, this embodiment also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned related steps to implement the appearance update method for a three-dimensional model provided in the above embodiment.

[0131] In this embodiment, the device, computer-readable storage medium, computer program product, or chip are all used to execute the corresponding methods provided above. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0132] Through the above description of the embodiments, those skilled in the art will understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0133] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another apparatus, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0134] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for updating the appearance of a three-dimensional model, characterized in that, include: Receive an appearance update instruction for a target 3D model and obtain the corresponding regular texture mapping map, wherein the regular texture mapping map is a 2D image generated by unfolding the surface to be updated of the target 3D model; The appearance update instruction is parsed to obtain appearance update semantic information; Based on the regular texture mapping map and the appearance update semantic information, generate one or more new regular texture mapping maps; In response to the user's selection of one or more new regular texture maps, the selected new regular texture map is applied to the surface corresponding to the target 3D model to update the appearance of the target 3D model.

2. The method according to claim 1, characterized in that, The step of generating one or more new regular texture maps based on the regular texture map and the appearance update semantic information includes: Based on the regular texture mapping map, the planar regions corresponding to each surface to be updated of the target 3D model are determined; Based on the appearance update semantic information, determine the target visual style and target mapping method; Based on the target visual style, candidate image content that matches the size of the regular texture map is generated; Based on the target mapping method, the candidate image content is adapted to one or more planar regions to obtain one or more new regular texture mapping maps.

3. The method according to claim 2, characterized in that, The target texture is in mirror mode; The step of adapting the candidate image content to one or more planar regions based on the target mapping method includes: Based on the mirror pattern, the position of the axis of symmetry, the first planar region, and the second planar region in the regular texture mapping map are determined. Based on the candidate image content, determine the basic pattern in the candidate image content; Based on the position of the axis of symmetry, the basic pattern is filled into the first planar region and the second planar region respectively.

4. The method according to claim 2, characterized in that, The target mapping method is a first non-mirror mode; the first non-mirror mode is a different mode from the mirrored surface to be updated. The step of adapting the candidate image content to one or more planar regions based on the target mapping method includes: Based on the first non-mirror mode, a first planar region and a second planar region are determined; Based on the content of the candidate images, a first pattern and a second pattern are determined, wherein the first pattern and the second pattern are different; The first pattern is filled into the first planar area, and the second pattern is filled into the second planar area.

5. The method according to claim 2, characterized in that, The target mapping method is the second non-mirror mode; the second non-mirror mode is a mode in which each surface to be updated is different. The step of adapting the candidate image content to one or more planar regions based on the target mapping method includes: Based on the second non-mirror mode, multiple different planar regions are determined; Based on the content of the candidate images, multiple third patterns corresponding to each planar region are determined; Fill each of the third patterns into the corresponding planar areas.

6. The method according to any one of claims 2 to 5, characterized in that, After adapting the candidate image content to one or more planar regions based on the target mapping method, the method further includes: The candidate image content is adapted to one or more planar regions to obtain an initial appearance image; Based on the outline of the regular texture map, the initial appearance image is cropped to generate one or more new regular texture maps.

7. The method according to claim 2, characterized in that, After generating one or more new regular texture maps, the method further includes: Receive the user's instruction to modify a selected new rule texture map; Based on the modification instructions, the rule texture mapping, and the appearance update semantic information, one or more updated new rule texture mappings are generated.

8. The method according to claim 1, characterized in that, The parsing of the appearance update instruction to obtain appearance update semantic information includes: The appearance update command is parsed to obtain the corresponding customization intent and attribute information; Based on the customization intent and the attribute information, the appearance update semantic information is generated; The appearance update semantic information includes at least: visual style information for controlling image content and texture information for controlling image layout.

9. The method according to claim 1, characterized in that, The method further includes: Obtain the three-dimensional model data of the target three-dimensional model; Based on the three-dimensional model data, extract the mesh data of each surface to be updated; The mesh data is unfolded onto a two-dimensional plane to obtain an initial texture map; The initial texture map is normalized to obtain the regular texture map; The normalization process involves arranging the planar regions corresponding to each surface to be updated in a regular manner along a preset direction within the two-dimensional plane.

10. An electronic device, characterized in that, The electronic device includes: Memory, used to store executable program code; A processor for calling and running the executable program code from the memory, causing the electronic device to perform the method as described in any one of claims 1 to 9.

Images