A method for producing three-dimensional urban assets, electronic devices, storage media, and products.
By implementing a process for generating 3D city assets, including original artwork processing, UV mapping, and material instance allocation, the project addresses the issues of time-consuming production and stylistic consistency in 3D city building asset creation. This improves production efficiency and artistic consistency while meeting the performance requirements of real-time rendering.
Patent Information
- Application Number
- CN202611199709.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-07
- Publication Date
- 2026-11-17
AI Technical Summary
Existing technologies suffer from time-consuming production of 3D city building assets, high labor costs, difficulty in unifying the styles of multiple assets, and low efficiency in assembling large-scale scenes.
By acquiring the original artwork from the target game, we generate 3D models, UV mapping data, and textures for each building asset, merge and group them, and finally import the asset groups into the game engine. We then use intelligent algorithms and modeling tools to optimize the generation process of 3D models and material instances.
It significantly shortened the early-stage scene construction cycle of the project, improved the artistic consistency and visual expressiveness of the game city scenes, reduced the production costs of manual image retouching and resource adaptation, and improved the overall efficiency of game scene asset production and implementation.
Smart Images

Figure CN123118979A_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to the field of computer graphics technology, and more specifically, to a method for producing three-dimensional city assets, an electronic device, a computer-readable storage medium, and a computer program product. Background Technology
[0002] With the rapid development of the gaming industry and virtual reality technology, the demand for 3D city scene assets in strategy games, historical games, and virtual cultural tourism is increasing. The construction of large-scale 3D city scenes requires a large number of architectural model assets. These assets not only need high visual fidelity but also need to maintain stylistic consistency and meet the performance requirements of real-time rendering. In related technologies, the production of 3D city architectural assets mainly relies on artists using 3D modeling software for manual modeling. This process includes multiple stages such as reference image analysis, model sculpting, topology optimization, UV unwrapping, and texture painting. The production cycle for a single detailed building typically takes several days, and the production of assets for large-scale city scenes often requires a workload of dozens of person-months.
[0003] In actual project development, the following specific requirements exist: First, a large number of architectural assets need to be generated rapidly within a limited development cycle; second, the generated architectural assets need to maintain a high degree of consistency in historical style, material characteristics, and color tone; third, the rendering performance of the assets needs to be optimized while ensuring visual quality to meet the performance requirements of real-time rendering engines. Therefore, a technical solution is needed that can improve the efficiency of batch production of 3D city assets while ensuring the geometric accuracy, style consistency, and rendering performance of the generated assets. Summary of the Invention
[0004] One objective of this disclosure is to provide a technical solution for producing 3D city assets, in order to solve the technical problems existing in related technologies, such as long production time, high labor costs, difficulty in unifying the styles of multiple assets, and low efficiency in assembling large-scale scenes.
[0005] According to a first aspect of this disclosure, a method for producing three-dimensional urban assets is provided, the method comprising: Obtain the original artwork of the target city in the target game; Generate a 3D model of each building asset in the target city based on the original artwork; Based on the 3D model of each building asset, generate UV mapping data for each building asset; A texture map is generated for each building asset based on the original artwork and the UV mapping data; The building assets within the target city are merged and grouped to obtain multiple asset groups, and a material instance referencing the corresponding texture is assigned to each asset group; wherein, each asset group includes at least one style of building asset; Build a 3D model exchange file for each asset group based on the 3D model, material instances, UV mapping data, and textures of each asset group; Import the 3D model exchange files of each asset group into the game engine of the target game.
[0006] Optionally, generating a 3D model of each building asset in the target city based on the original artwork includes: performing image cleaning processing on the original artwork for each building asset to obtain a segmented image of the corresponding building asset; wherein each segmented image contains only the corresponding building asset; and generating a 3D model of the corresponding building asset based on the segmented image of each building asset.
[0007] Optionally, generating a 3D model of the corresponding building asset based on the segmented image of each building asset includes: generating a multi-view view of the corresponding building asset based on the segmented image of each building asset, wherein the multi-view view includes a front view, a back view, a left view, and a right view; and generating a 3D model of the corresponding building asset based on the multi-view view of each building asset.
[0008] Optionally, generating a multi-view view of the corresponding building asset based on the segmented image of each building asset includes: constructing a three-dimensional frame model of each building asset using basic geometry; wherein the basic geometry includes at least one of cube, cuboid, and cylinder; and generating a multi-view view of the corresponding building asset based on the segmented image and the three-dimensional frame model of each building asset.
[0009] Optionally, assigning a material instance referencing the corresponding texture to each asset group includes: classifying multiple asset groups according to the importance of the building assets in the urban scene; and assigning a material instance referencing the corresponding texture to each asset group according to the classification results.
[0010] Optionally, the step of assigning a material instance referencing the corresponding texture to each asset group according to the classification result includes: obtaining a first asset group among the plurality of asset groups according to the classification result; and assigning a material instance referencing the corresponding texture to each building asset in the first asset group according to the identifier of the UV mapping data corresponding to each building asset in the first asset group.
[0011] Optionally, assigning a material instance referencing the corresponding texture to each asset group based on the classification result includes: obtaining a second asset group from multiple asset groups based on the classification result; scaling the UV mapping data corresponding to each building asset in the second asset group to a UV mapping layout; recombining the textures of each building asset in the second asset group into a merged texture according to the UV mapping layout; and assigning a material instance referencing the merged texture to the 3D model of the second asset group.
[0012] Optionally, the step of merging and grouping the building assets within the target city to obtain multiple asset groups includes: merging building assets of the same asset category according to their asset class; and grouping the building assets to obtain multiple asset groups.
[0013] Optionally, the method further includes: performing integrated verification of the 3D models of each asset group in the game engine.
[0014] Optionally, the 3D models of each asset group are integrated and verified in the game engine, including at least one of the following: importing the 3D model exchange files of each asset group into the game engine; performing size verification on the 3D models of each building asset and detecting collision bodies of each building asset in the game engine; configuring the material instances of each asset group in the game engine; and adjusting the lighting and shadows of the city scene in the game engine.
[0015] Optionally, the method further includes: performing super-resolution enhancement processing on textures with resolutions below a threshold.
[0016] According to a second aspect of this disclosure, an electronic device is also provided, including a memory and a processor, the memory being used to store a computer program; the processor being used to execute the computer program to implement the method according to a first aspect of this disclosure.
[0017] According to a third aspect of this disclosure, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements the method described in accordance with a first aspect of this disclosure.
[0018] According to a fourth aspect of this disclosure, a computer program product is also provided, including a computer program that, when executed by a processor, implements the method described in the first aspect of this disclosure.
[0019] This embodiment establishes a complete three-stage production process from concept art to game engine-runnable assets, including the first stage of asset production, the second stage of material texture optimization, and the third stage of engine integration and effect verification. This significantly shortens the early-stage scene construction cycle of the project, effectively improves the overall artistic consistency, visual expressiveness, and scene running smoothness of the game city scene, greatly reduces the production costs of manual image retouching and resource adaptation, and improves the overall efficiency of game scene asset production and implementation.
[0020] Other features and advantages of the embodiments of this disclosure will become clear from the following detailed description of exemplary embodiments with reference to the accompanying drawings. Attached Figure Description
[0021] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments of the present disclosure and, together with the description, serve to explain the principles of the disclosure.
[0022] Figure 1 This is a schematic diagram of the structure of an electronic device to which the methods of the embodiments of this disclosure can be applied.
[0023] Figure 2 This is a flowchart illustrating a method for producing three-dimensional urban assets according to some embodiments.
[0024] Figure 3 This is a block schematic diagram of an electronic device according to one embodiment. Detailed Implementation
[0025] Various exemplary embodiments of the present disclosure will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values of the components and steps set forth in these embodiments do not limit the scope of the present disclosure.
[0026] The following description of at least one exemplary embodiment is merely illustrative and is not intended to limit the scope of this disclosure or its application or use.
[0027] Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and equipment should be considered part of the specification.
[0028] In all the examples shown and discussed herein, any specific values should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.
[0029] It should be noted that similar labels and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be discussed further in subsequent figures.
[0030] This disclosure relates to a three-dimensional urban asset production scheme. Figure 1 This is a schematic diagram of the structure of an electronic device capable of applying the three-dimensional urban asset production method of the embodiments of this disclosure. Figure 1 As shown, users can interact with electronic devices to produce 3D city assets.
[0031] Electronic device 1100 can be a mobile device, tablet computer, laptop computer, desktop computer, wearable computer, or other electronic product. Of course, this is only an illustrative example and does not mean that this disclosure is limited to this.
[0032] like Figure 1 As shown, the electronic device 1100 may include a processor 1101, a memory 1102, an interface device 1103, a communication device 1104, an output device 1105, an input device 1106, etc. Figure 1 The hardware configuration shown is illustrative only and is not intended to limit this disclosure, its application, or its use.
[0033] The processor 1101 executes computer programs, which can be written using instruction sets of architectures such as x86, Arm, RISC, MIPS, and SSE. The memory 1102 includes, for example, ROM (Read-Only Memory), RAM (Random Access Memory), and non-volatile memory such as a hard disk. The interface device 1103 includes, for example, a USB interface, a network cable interface, and a headphone jack. The communication device 1104 is capable of wired or wireless communication. The communication device 1104 may include at least one short-range communication module, such as any module for short-range wireless communication based on short-range wireless communication protocols such as Hilink, WiFi (IEEE 802.11), Mesh, Bluetooth, ZigBee, Thread, Z-Wave, NFC, UWB, and LiFi. The communication device 1104 may also include a long-range communication module, such as any module for WLAN, GPRS, or 2G / 3G / 4G / 5G long-range communication. The output device 1105 may include, for example, an LCD screen or touch screen, and a speaker. Input device 1106 may include, for example, a touch screen, a keyboard, a microphone, various sensors, etc.
[0034] In this embodiment, the memory 1102 of the electronic device 1100 is used to store a computer program that controls the processor 1101 to operate in order to execute the steps of the three-dimensional urban asset production method performed by the electronic device according to any embodiment of the present disclosure.
[0035] Electronic device 1100 acquires the concept art of a target city in a target game; generates a 3D model of each building asset in the target city based on the concept art; generates UV mapping data for each building asset based on the 3D model of each building asset; merges and groups the building assets in the target city to obtain multiple asset groups; generates textures for each building asset based on the concept art and the UV mapping data, and assigns a material instance referencing the corresponding texture to each asset group; wherein each asset group includes at least one style of building asset; constructs a 3D model exchange file for the corresponding asset group based on the 3D model, material instance, UV mapping data, and texture of each asset group; and imports the 3D model exchange files of each asset group into the game engine of the target game.
[0036] The following is combined Figure 1 The electronic device shown illustrates the method for producing three-dimensional urban assets provided in this disclosure. It should be noted that... Figure 1 This is merely one application scenario of the 3D city asset production method provided in this disclosure embodiment, and does not mean that the 3D city asset production method can only be applied to... Figure 1 The application scenarios shown.
[0037] <First Embodiment> Figure 2 This is a flowchart illustrating a method for producing 3D city assets according to some embodiments. This method can be implemented by electronic devices; specifically, it can be implemented by... Figure 1 The illustrated electronic device 1100 is implemented. For example... Figure 2 As shown, the three-dimensional city asset production method of this embodiment may include the following steps S210 to S270.
[0038] Step S210: Obtain the original artwork of the target city in the target game.
[0039] The original artwork of the target city is output by the art design team. It is the art concept design drawing of the target city in the game. It is a standardized two-dimensional reference image material that has been pre-designed and finalized by the game art team.
[0040] Specifically, the original artwork includes all design specifications information such as the overall city layout outline of the target city, the shape and structure of various buildings, component styles, size proportions, color styles, era and materials, and decorative details.
[0041] Step S220: Generate a 3D model of each building asset in the target city based on the original artwork.
[0042] In this embodiment, building assets are reusable and standardized building resource units within the target city scene of the game, serving as the basic building components for a complete city scene. A 3D model is the digital 3D geometric carrier corresponding to a single building asset, consisting of a large number of vertices, edges, and triangular faces forming a complete closed mesh, carrying information about the building's shape and the spatial structure of its components.
[0043] In some embodiments, the target concept art can be imported into a modeling tool to obtain a 3D model of each building asset.
[0044] In some embodiments, generating a 3D model of each building asset in the target city based on the original artwork includes: performing image cleaning processing on the original artwork for each building asset to obtain a segmented image of the corresponding building asset; wherein each segmented image contains only the corresponding building asset; and generating a 3D model of the corresponding building asset based on the segmented image of each building asset.
[0045] In this embodiment, the original artwork can be divided into regions using contour recognition and semantic segmentation algorithms to identify the independent contour boundaries of each building asset within the original artwork. The original artwork image is then cropped along the identified building contour boundaries to obtain segmented images that match each building asset. Each segmented image retains only the complete appearance structure of a single building asset, without other buildings or redundant background elements, and each building asset corresponds to a unique segmented image. The contour recognition and semantic segmentation algorithms can be, for example, U-Net or Mask R-CNN.
[0046] In these examples, image cleaning and segmentation cropping are used to decompose the complete original artwork containing multiple buildings and redundant backgrounds into segmented images containing only a single building, eliminating the interference of irrelevant elements on 3D reconstruction. Using the segmented image of a clean single building asset as modeling input can reduce the probability of structural confusion and shape distortion during modeling, ensuring that the 3D model of each building asset accurately matches the single building appearance standard of the original artwork design.
[0047] In some embodiments, the segmented images of each building asset can be imported into a modeling tool to obtain a three-dimensional model of each building asset.
[0048] In some embodiments, generating a 3D model of a corresponding building asset based on a segmented image of each building asset includes: generating a multi-view view of the corresponding building asset based on a segmented image of each building asset, wherein the multi-view view includes a front view, a rear view, a left view, and a right view; and generating a 3D model of the corresponding building asset based on the multi-view view of each building asset.
[0049] In this embodiment, the two-dimensional contour of the segmented image is used as the appearance constraint, and four sets of multi-view views without perspective distortion are generated according to the orthogonal projection rules. These views are, in order, the front view, the back view, the left view, and the right view. The resolution and color parameters of each view are uniform, and the component size ratio of all view views of the same building asset is consistent. The pixel positions of the same component correspond to each other under different views.
[0050] Specifically, the segmented image of a single building asset is used as the sole original input. This segmented image is free from background interference and occlusion by other buildings, and fully preserves the exterior color, texture, component layout, and outline features of the target building. First, a global feature analysis is performed on the segmented image using an intelligent view inference algorithm to accurately extract core two-dimensional feature information such as the overall outer contour dimensions, wall proportions, roof structure, door and window component arrangement, and decorative positions of the building asset. Combined with prior rules of urban building structures, the overall spatial structural attributes, symmetry relationships, and component thickness parameters of the building asset are inferred.
[0051] Subsequently, based on the obtained two-dimensional features and prior information about the spatial structure, multi-view pose transformation and pixel reconstruction are performed. According to the principle of orthogonal perspective projection, the camera's shooting perspectives from four orthogonal directions—front, back, left, and right—are simulated to complete adaptive inference and rendering of the building's outline, component positions, and occlusion relationships under different perspectives. For visual blind spots and missing pixel areas in the back and side views, the missing texture details and outline information are intelligently supplemented by combining the building's structural symmetry and facade texture consistency features, ensuring that each generated view has a reasonable structure, coherent texture, and unified style.
[0052] Finally, the four generated multi-view images are standardized by preprocessing to unify the resolution, pixel ratio, lighting and color parameters of each view, and eliminate uneven brightness and color differences caused by perspective rendering deviations, resulting in four multi-view images of the front, back, left and right sides with uniform specifications, complete structure and regular details.
[0053] The four multi-view images of the building asset are all sent to the modeling tool. The modeling tool reads the building outline and component distribution information from different perspectives, uses the spatial relationship between the multiple perspectives to complete the missing depth information of a single two-dimensional image, automatically fits the subdivided geometric structures such as walls, roofs, beams, columns, and window frames, and generates a closed and continuous triangular mesh to obtain the three-dimensional model corresponding to the building asset.
[0054] The modeling tool in this embodiment can be either HY 3D or Tripo. Developers can choose the modeling tool according to their own needs.
[0055] Relying solely on a single segmented image can only obtain information from a single plane, easily leading to defects such as missing depth and distorted side structures in the 3D model. This embodiment generates orthogonal multi-view perspectives in four directions (front, back, left, and right) to supplement the appearance constraints of the building asset from various angles, providing complete two-dimensional reference information for 3D reconstruction and effectively improving the structural integrity and dimensional accuracy of the 3D model.
[0056] In some embodiments, generating a multi-view view of a building asset based on a segmented image of each building asset includes: constructing a three-dimensional frame model of each building asset using basic geometry; wherein the basic geometry includes at least one of a cube, cuboid, and cylinder; and generating a multi-view view of the building asset based on the segmented image and the three-dimensional frame model of each building asset.
[0057] In this embodiment, two-dimensional dimensional constraint parameters such as the overall length-width-height ratio of the building, the wall span, the roof height, and the relative spacing of components are extracted from the segmented image.
[0058] Select one or more basic geometric shapes from cubes, cuboids, and cylinders, and combine them in a 3D editing environment to build a 3D frame model that fits the building's outer contour. Using 2D dimensional constraint parameters as a standard, scale, translate, and rotate each basic geometric shape to make the overall outer contour of the 3D frame model and the volume ratio of each module perfectly match the building's external dimensions presented in the segmented image.
[0059] The segmented image is used as a style and detail reference texture and applied to the corresponding facade of the 3D frame model. The spatial pose of the frame model is locked as an orthogonal, non-perspective state. The image is rendered along the front, back, left, and right orthogonal projection directions of the model to obtain the front view, back view, left view, and right view of the corresponding building asset.
[0060] Furthermore, the resolution, color channels, and bit depth of the four multi-view images of each building asset can be unified. The length, width, and relative position of the same building component in each view can be verified to be consistent with the 3D frame model and segmented image. If there is a pixel deviation, the geometric proportions of the 3D frame model can be readjusted and the view can be rendered again.
[0061] Iterate through the segmented images of all building assets, repeatedly perform the operations of building a 3D frame model from the basic geometry and orthogonally rendering multi-view views to obtain the four-way multi-view views associated with all building assets.
[0062] Simply relying on segmented images to generate multi-view perspectives only provides two-dimensional planar information, which can easily lead to inconsistencies in architectural scale and depth logic across different perspectives. This embodiment first constructs a three-dimensional framework model using basic geometry to lock in the overall three-dimensional scale of the building, and then combines segmented images to render orthogonal multi-view perspectives. This ensures the spatial logic consistency of the front, back, left, and right views, eliminates perspective distortion and dimensional misalignment, and provides multi-dimensional reference materials with consistent scale and structural self-consistency for subsequent three-dimensional model construction, reducing the probability of geometric distortion in the three-dimensional model.
[0063] Step S230: Generate UV mapping data for each building asset based on the 3D model of each building asset.
[0064] UV mapping data is a digital dataset that records the coordinates of each vertex of a 3D model in a 2D texture space (UV space, with coordinate values ranging from 0 to 1). It establishes a one-to-one mapping relationship between the surface of the 3D model and the 2D texture, and is the basic data for realizing the surface texture rendering of the 3D model.
[0065] Specifically, this can involve retrieving a complete, closed triangular 3D mesh of a single building asset, pre-removing invalid geometric structures such as hidden faces, tiny fragments, and overlapping faces within the model, and dividing different components such as walls, roofs, beams, columns, doors, and windows into regions based on their material properties; performing automatic UV unwrapping on the geometric surfaces of each marked component, decomposing the 3D surface into a 2D texture space; setting a minimum stretch constraint threshold during the unwrapping process to control the degree of texture stretching and deformation after the surface is unwrapped, avoiding large-area texture distortion; and then performing UV unwrapping on the dispersed UV islands (a set of continuously connected triangular faces of the 3D model). After UV unwrapping, a continuous closed coordinate region (i.e., a UV island) is formed in the two-dimensional texture space. A regular UV arrangement operation is performed, which translates, scales, and rotates each UV island in the two-dimensional texture space to reduce the blank and redundant areas between UV islands, while ensuring the alignment of UV seams of adjacent components and reducing color difference in subsequent texture splicing. After completing the UV arrangement, the complete set of vertex UV coordinates of the building is exported to form standardized UV mapping data. The UV mapping data records the one-to-one correspondence between each vertex of the 3D model and the coordinates of the two-dimensional texture space, and is stored in a format that can be read by 3D editing tools and AI texture generation tools.
[0066] Furthermore, it can also verify the integrity of UV mapping data, checking that all faces of the 3D model have matching UV coordinates, no missing UVs, and no UV islands exceeding the range of the 2D texture space; if the verification is qualified, the UV mapping data is output; if it is unqualified, the UV unfolding and arrangement parameters are readjusted; and the UV mapping data of all building assets is generated in batches by traversing all 3D building models.
[0067] Step S240: Generate textures for each building asset based on the original artwork and UV mapping data.
[0068] Textures are two-dimensional image files that adapt to the 3D model of a building asset and store surface visual information such as color, bump, and reflection. They are mapped onto the surface of the 3D model based on UV mapping data to achieve the rendering of artistic details of the model. They are an indispensable supporting resource for a complete building asset.
[0069] In this embodiment, the visual style, color system, material texture, decorative details, and overall art design of the corresponding building assets in the target city's original artwork can be read first. Simultaneously, the UV mapping data of the building asset can be read to clarify the mapping relationship of each vertex on the 3D model surface in the 2D texture space. Based on the building structure and UV island distribution, each building asset is divided into multiple material regions, such as wall regions, roof regions, door and window regions, beam and column regions, ground step regions, and decorative pattern regions. Each region corresponds to one or more UV islands in the UV mapping data. Key texture semantics of the building asset are extracted from the original artwork, including but not limited to: walls being made of blue bricks, red bricks, adobe, wood, limestone, or stone; roofs being made of blue tiles, glazed tiles, thatched roofs, or metal plates; and doors and windows being made of wooden grilles, carved windows, or stone door frames. These material features are then converted into texture generation conditions. Using the UV mapping data as the basis for expansion, the building materials, colors, patterns, and lighting features in the original artwork are projected into the 2D texture space to generate the basic texture map of the building asset. The texture map should include the color, material, texture direction, and general details of the main parts of the building. The base texture map should be adapted according to the boundaries, stretching direction, and arrangement of the UV islands. For UV areas at risk of stretching, the texture direction, repetition density, and pixel sampling method should be adjusted to avoid obvious stretching, blurring, or misalignment of the texture on the 3D model surface. Further details visible in the original artwork should be added to the base texture map, such as brick seams, corrugated tiles, wood grain, stone cracks, peeling paint, door and window carvings, plaques, eaves decorations, and worn corners. Detailed textures should maintain consistency with the style of the original artwork and conform to the structural logic of the architectural asset. The seam areas between different UV islands should be blended to ensure that adjacent UV islands do not exhibit obvious color bands, breaks, or repeated texture boundaries after being stitched together on the 3D model surface. For continuous decorative patterns spanning UV islands, seam alignment and color transitions should be implemented.
[0070] Depending on the requirements of the target game, texture information can be broken down into multi-channel maps, including any one or more of the following: base color map, normal map, roughness map, metallicity map, height map, and ambient occlusion map. The base color map primarily carries the original artwork's colors and material patterns; the normal map is used to represent the surface textures such as brick seams, corrugations, and wood grain.
[0071] In some embodiments, generating textures for each building asset based on original artwork and UV mapping data may include: original artwork preprocessing and high-precision texture detail extraction; UV mapping data adaptation and calibration; high-precision texture generation based on original artwork and UV data; texture optimization and specification unification; texture association binding and detail verification optimization.
[0072] In the preprocessing of original artwork and the extraction of high-precision texture details, to address issues such as blurred details, simplified textures, and missing subtle decorations in the original artwork, an AI high-precision detail filling model can be introduced to initially enhance the details. Based on a trained game city building texture dataset, the model intelligently identifies weak texture areas such as building wall textures, door and window patterns, roof carvings, and corner decorations, and performs high-precision detail filling to fill in the missing fine structures, transitional textures, and subtle light and shadow layers in the original artwork. Simultaneously, based on image pixel recognition technology, high-precision texture feature information of the building assets in the AI-enhanced original artwork is extracted, including core appearance information such as wall material textures, door and window textures, roof patterns, decorative patterns, color ratios, and light and shadow transitions. This establishes a high-fidelity building appearance texture pixel dataset, providing a high-precision texture base for subsequent texture generation.
[0073] In the UV mapping data adaptation and calibration step, the UV mapping data corresponding to the current building asset is retrieved, and the size specifications, UV block layout, and vertex coordinate correspondence of the UV mapping panel are analyzed. UV adaptation calibration is performed for building models with different structures. For regular cube and cuboid wall and roof models, a regular UV layout is adopted to ensure that the texture is not stretched or distorted. For irregular decorative structures and curved building components, the UV coordinate parameters are fine-tuned to correct the mapping deviation between the 3D curved surface and the 2D texture plane, ensuring that the UV mapping panel can completely and without omission cover all the outer surface areas of the 3D building model, achieving a one-to-one correspondence between the model surface and the UV texture area.
[0074] In the high-precision texture generation step based on original artwork and UV data, the preprocessed AI-enhanced original artwork texture pixel information is bound and matched with the calibrated UV mapping data. Using UV mapping coordinates as a reference, the color, texture, and pattern information of the original artwork are accurately mapped onto the 2D UV texture surface. For texture blank areas, texture stretching and blurring areas, and pixel breakage areas generated after UV unwrapping, an AI high-precision detail filling algorithm is used for intelligent completion and repair. This algorithm can adaptively generate highly matching micro-texture details based on the overall texture style of the architectural asset, the texture features of adjacent areas, and the design logic of the original artwork, avoiding problems such as stiff textures, missing details, and style inconsistencies caused by traditional pixel filling. Simultaneously, through a texture baking algorithm, the surface lighting and structural features of the 3D architectural model are deeply integrated with the AI-optimized original artwork design effect, automatically correcting issues such as texture overlap, pixel misalignment, and edge breakage, generating a high-precision base texture map that is rich in detail, has a unified style, and is adapted to the current 3D architectural asset model.
[0075] In the texture optimization and specification unification steps, the generated high-precision base textures undergo refined optimization. Based on the resource adaptation standards of the target game engine, texture resolution, pixel format, and channel parameters are unified, and redundant pixel information is removed. During this process, AI-powered detail-preserving compression technology is used to lightweight the textures, accurately retaining core high-precision information such as micro-textures, lighting transitions, and decorative details filled by the AI while compressing the texture file size, avoiding the blurring and distortion caused by traditional compression. Simultaneously, considering the lighting and shadow performance requirements of architectural assets, an AI material detail transfer model intelligently generates corresponding normal maps, specular maps, and roughness maps, automatically matching the physical detail characteristics of different architectural materials. This enriches the material performance effects of architectural assets, enabling buildings to achieve realistic light and shadow reflections and changes in brightness within the game scene.
[0076] In the steps of texture association and binding, and detail verification and optimization, the optimized high-precision custom textures are associated and bound with the UV mapping data and 3D model of the corresponding building assets. This ensures that the textures can accurately fit the surface of the 3D model when material balls are called and model files are imported into the engine. An AI visual inspection model is introduced to perform full-domain verification of the texture mapping effect, intelligently identifying issues such as texture stretching, blurring, misalignment, color difference, and AI-filled detail inconsistencies and uneven texture connections in areas such as building walls, doors, windows, and roofs. If detail deviations are detected, the AI can automatically fine-tune the UV mapping coordinates and texture fill parameters in reverse, iteratively optimizing the detail effect until the overall texture, fill details, lighting and shadow performance of the texture are highly consistent with the original concept art design style, achieving accurate implementation of high-precision details.
[0077] In some embodiments, after performing step S240, the method may further include: performing super-resolution enhancement processing on textures with resolutions below a threshold.
[0078] Specifically, a preset texture resolution threshold is established, based on the target game engine's rendering standards, scene resource loading limits, and the visual display requirements of the urban scene. Resolution pixel statistics and accuracy checks are performed on the generated and optimized textures of each building asset, filtering out low-resolution textures with resolutions below the preset threshold. For low-resolution textures, an AI super-resolution enhancement algorithm is used for iterative upscaling and detail reconstruction. Based on a high-resolution building texture training model, the algorithm intelligently fills in missing high-frequency texture details, repairs blurred textures, and optimizes edge contour clarity while increasing the texture resolution by a factor of 1, avoiding problems such as texture blurring, jagged edges, and loss of detail caused by traditional interpolation upscaling. After super-resolution enhancement, the texture UV mapping arrangement is re-aligned to ensure that the enhanced texture coordinates perfectly match the original UV mapping data, preventing texture shifts and stretching errors. Ultimately, all building asset texture resolutions meet the engine's rendering standards, further enhancing the overall high-definition display effect of the 3D urban scene.
[0079] Step S250: Merge and group the building assets in the target city to obtain multiple asset groups, and assign a material instance referencing the corresponding texture to each asset group; wherein, each asset group includes building assets of at least one style.
[0080] An asset group is a collection of building assets gathered according to preset rules. Each collection contains one or more building assets. Building assets within the same asset group share a set of UV layout and material configuration logic and are uniformly packaged into a 3D model exchange file for batch reuse in the game engine.
[0081] Material instances are reusable rendering units derived from base material templates and bound to corresponding texture resources. Different material instances differ only in texture and UV mapping parameters, while the underlying rendering pipeline remains consistent. The game engine controls the visual appearance of building assets, such as surface color, bump, and reflection, through material instances.
[0082] When merging and grouping building assets within the target city to obtain multiple asset groups, the UV mapping data corresponding to each building asset within the asset group is not merged.
[0083] In some embodiments, building assets within a target city are merged and grouped to obtain multiple asset groups, including: merging building assets of the same asset category according to their asset class; and grouping the merged building assets to obtain multiple asset groups.
[0084] In this embodiment, the names of building assets adopt a segmented structured naming format, which can be: Scene Identifier_Building Function Category_Sub-style Number_Detail Suffix, with each segment separated by an underscore, and each segment carrying fixed semantic information. The Scene Identifier is a unified code for the target city, and this field is completely consistent for all building assets in the same city; the Building Function Category is used to distinguish the building usage types such as government offices, residences, shops, city defenses, and gate towers; the Sub-style Number is used to distinguish different exterior structural versions within the same category and is the core field for distinguishing styles; the Detail Suffix is used to indicate the precision and draft status of the version and does not affect the style determination.
[0085] The names of building assets can be set according to the asset category, so as to determine whether two building assets belong to the same asset category based on their names. For two building assets to be determined to belong to the same asset category, they must simultaneously meet the following conditions: the scene identifier, building function category, and sub-style number in the name must be completely identical; the final precision and draft version suffix are not included in the style matching verification.
[0086] For building assets with the same name and style, perform instance merging, retaining only one complete 3D model, UV mapping data and texture resources. Other assets of the same style will uniformly reference this set of basic resources, and duplicate geometric and texture data will be removed.
[0087] After merging building assets of the same style, the building assets are then grouped. Specifically, building assets can be grouped according to a set quantity and the final precision in the detail suffix, so that the number of building assets in each asset group does not exceed the set quantity, and the final precision of building assets in the same asset group is consistent.
[0088] When multiple asset groups are obtained, an independent index file is created for each asset group. The index records the names of all building asset styles, corresponding 3D model paths, UV mapping data files, and texture file relationships within the group, completing the structured storage of grouped data, which is then used in subsequent material instance allocation and 3D model exchange file construction processes.
[0089] In these examples, identical building assets are first merged by name matching to eliminate the storage space waste caused by storing duplicate geometry and texture resources; then, asset groups are divided based on style uniformity to achieve standardized classification and management of building assets, which facilitates the subsequent unified configuration of UV layout schemes, allocation of material instances, and batch packaging and export of 3D model exchange files by asset group, which can simplify the batch production of 3D assets and engine import process.
[0090] In some embodiments, assigning a material instance referencing a corresponding texture to each asset group includes: classifying multiple asset groups according to the importance of architectural assets in an urban scene; and assigning a material instance referencing a corresponding texture to each asset group according to the classification results.
[0091] In this embodiment, the degree of importance can be determined based on a combination of location salience, game function importance, and view distance. Location salience indicates whether the corresponding building asset is located on either side of a main road; game function importance indicates whether the corresponding building asset is a mission objective building; and view distance indicates whether the closest distance between the player's viewpoint and the corresponding building asset is less than a preset threshold.
[0092] Based on a comparison of the importance of architectural assets within each asset group within the urban scene and their importance thresholds, all asset groups are binary classified according to the comparison results, resulting in a high-precision first asset group and a lightweight, distant-view second asset group. The first asset group consists of architectural assets that are displayed up close in the urban scene and are the focus of the camera, while the second asset group consists of architectural assets that are shown from a distance and are rarely seen in close-ups.
[0093] Material instances are derived from a unified basic material template. The allocation operation involves creating a new material instance that matches the rendering requirements for the asset group and writing a group identifier for subsequent packaging identification.
[0094] Based on the classification results, material instances referencing corresponding textures are assigned to each asset group. Specifically, the scene importance classification of the asset group can be used as the criterion for material instance allocation. Texture referencing rules are configured differently according to the classification to generate material instances that meet the rendering needs of each group and bind corresponding texture resources, thus establishing a complete association index of asset groups, material instances, textures, and UV mapping data.
[0095] In these examples, asset groups are automatically classified based on their importance, enabling automatic separation of two production schemes: high-precision rendering and lightweight resources. This eliminates the need for manual configuration of material rules for each building group, improves the standardization of 3D asset mass production, and balances game graphics quality and performance.
[0096] In some embodiments, assigning a material instance referencing a corresponding texture to each asset group according to the classification result includes: obtaining a first asset group among multiple asset groups according to the classification result; and assigning a material instance referencing a corresponding texture to each building asset in the first asset group according to the identifier of the UV mapping data corresponding to each building asset in the first asset group.
[0097] The first asset group contains one or more building assets that require a high-precision independent texture rendering solution.
[0098] In this embodiment, the identifier of the UV mapping data corresponding to each building asset in the first asset group is retained. Each building asset corresponds to a unique and exclusive UV mapping data identifier, which is associated with the independent sticker of that building asset. Figure 1 One binding.
[0099] Call the unified basic material template and derive material instances for each UV mapping data identifier in the first asset group; during the derivation process, read the texture corresponding to the UV mapping data identifier, load the texture into the texture sampling channel of the material instance, and make the material instance reference the building-specific texture.
[0100] The generated material instances, texture storage paths, UV mapping data identifiers, and corresponding building asset style numbers are associated and recorded to ensure that when the 3D model of the building is rendered, it can match its own UV mapping data to read exclusive textures and corresponding material instances.
[0101] In these examples, for the highly important first asset group, one-to-one material instance allocation is achieved by relying on the unique UV mapping data identifiers of each building. Each building asset uses its own exclusive texture, which fully preserves texture details and prevents texture blurring and pattern disorder when the camera zooms in. By using the UV mapping identifier as a matching link, the binding of textures and material instances is automatically completed, avoiding texture misalignment and UV mismatch problems caused by manual matching, and ensuring high-precision rendering effect of core buildings in urban scenes.
[0102] In some embodiments, assigning a material instance referencing the corresponding texture to each asset group based on the classification result includes: obtaining a second asset group from multiple asset groups based on the classification result; scaling the UV mapping data corresponding to each building asset in the second asset group to a UV mapping layout; recombining the textures of each building asset in the second asset group into a merged texture according to the UV mapping layout; and assigning a material instance referencing the merged texture to the 3D model of the second asset group.
[0103] Specifically, this involves extracting the UV mapping data and corresponding UV island boundary information of each building asset within the second asset group. Within the UV mapping panel, all building UV islands are uniformly scaled, translated, and rotated for arrangement. During this arrangement, the gaps between UV islands are compressed to prevent overlap and interweaving, thus integrating the UV mapping data of multiple building assets into a single UV mapping panel. Using the integrated UV mapping panel as the splicing reference, the original independent textures of all building assets within the group are read. Pixel clipping and alignment are performed according to the position of each building asset's UV islands in the common panel, unifying image resolution and color channels. Pixel fusion smoothing is applied to the UV seam areas of adjacent building assets, ultimately outputting a single merged texture map covering the texture information of all building assets within the second asset group. A common basic material template is retrieved to derive a unique material instance. The spliced merged texture map is loaded into the texture sampling channel of this material instance, ensuring that the material instance references only this one merged texture map.
[0104] Record the second asset group identifier, public UV mapping panel data, merged texture storage path, unique material instance number, and mapping correspondence of each building 3D model within the group to ensure that all building 3D models in the second asset group can be rendered by reading merged textures through the UV mapping panel.
[0105] In these examples, the second asset group consists of low-importance distant buildings. The UVs of multiple buildings in the group are uniformly scaled and arranged onto the same UV mapping panel, recombined to generate a single merged texture and assigned only one material instance. This significantly reduces the number of texture files and the total number of material instances loaded by the game engine, reduces memory usage and rendering batch overhead, and prevents significant image quality loss in distant display scenes, balancing resource loading efficiency and basic visual performance in urban scenes.
[0106] Step S260: Construct the corresponding 3D model exchange file for each asset group based on the 3D model, material instances, UV mapping data, and textures.
[0107] The 3D model exchange file is a standardized file carrier that encapsulates the complete set of rendering resources for a single asset group. It integrates and stores all architectural 3D models, UV mapping data, textures, material instances, and resource association indexes within the asset group. The file has cross-tool and engine universal reading capabilities. After being imported into the game engine, it can directly complete operations such as building size verification, collision generation, and scene rendering. It is the final finished product file for delivering architectural assets to the game engine.
[0108] In one example, the 3D model exchange file can be an FBX file, a glTF file, an OBJ file, or a USD file.
[0109] The complete set of resources for each asset group, including 3D models, UV mapping data, textures, and material instances, is integrated into a single standardized exchange file. Resource relationships are stored internally, eliminating the need to manually match various resources sequentially when importing into the game engine. At the same time, the storage logic for high-precision and lightweight asset groups is differentiated, fully preserving their respective UV and material configuration characteristics. This simplifies the process of importing, managing, and reusing game resources, and improves the efficiency of building city scenes.
[0110] Step S270: Import the 3D model exchange files of each asset group into the game engine of the target game.
[0111] Batch import the 3D model exchange files corresponding to all asset groups into the target game engine. The target game engine automatically parses the 3D models, UV mapping data, textures, and material instances encapsulated in the 3D model exchange files, and automatically unpacks and loads geometric resources, texture resources, and material resources to generate internal resource objects that the engine can call in real time.
[0112] As can be seen from the above steps S210 to S270, this embodiment forms a complete three-stage production process from original artwork to game engine-runnable assets, including the first stage of asset production, the second stage of material texture optimization, and the third stage of engine integration and effect verification. This significantly shortens the early-stage scene building cycle of the project, effectively improves the overall artistic unity, visual expressiveness, and scene running smoothness of the game city scene, greatly reduces the production costs of manual image retouching and resource adaptation, and improves the overall efficiency of game scene asset production and implementation.
[0113] In some embodiments, after performing step S270, the method may further include: performing integrated verification of the 3D models of each asset group in the game engine.
[0114] Integrated verification is a comprehensive verification process conducted from four dimensions—resource integrity, rendering performance, scene performance, and interaction logic—after the packaged 3D model exchange files of the asset group are imported into the game engine in batches. The goal is to confirm that the entire set of architectural asset resources can be parsed, loaded, rendered, and used normally in the engine, without issues such as data gaps, abnormal visuals, performance exceeding limits, or interaction failures.
[0115] By importing the 3D model exchange files of each asset group into the target game's game engine for integration and verification, the correctness and performance of the processed 3D models in the game engine can be ensured, forming a complete closed loop from concept art to game engine-runnable assets.
[0116] In some embodiments, the 3D models of each asset group are integrated and verified in the game engine, including at least one of the following: performing size verification on the 3D models of each building asset in the game engine; performing collision detection on the 3D models of each building asset; configuring materials for material instances of each asset group in the game engine; and adjusting lighting and shadows for the city scene in the game engine.
[0117] Size verification can involve retrieving segmented images of each building asset and the two-dimensional size constraint parameters of the 3D frame model, and comparing them with the 3D model loaded in the target game engine. This verifies whether the overall length, width, height, and proportions of components such as walls, roofs, doors, and windows of the corresponding building asset are consistent with the initial design standards. Building assets with scaling distortion or size deviations are marked as abnormal so that game developers can make modifications.
[0118] Collision detection can automatically generate simple collision bounding bodies based on the 3D model of building assets, and detect whether the collision bodies fit the outer contour of the building, have no large empty areas, and have no internal penetration; at the same time, it verifies that the collision judgment area of interactive buildings is accurate, and that there is no redundant occupation of collision bodies of distant decorative buildings.
[0119] When configuring materials for each asset group in the game engine, for the first asset group, each building's independent material instance within the group is checked to ensure it can correctly read its proprietary texture, that UV mapping data sampling is unstretched, that there are no obvious color differences at seams, and that texture details are complete and clear from close-up views. For the second asset group, the shared single material instance is verified to be able to completely read the merged texture, that all buildings within the group display textures correctly based on the UV mapping panel, and that there are no missing textures or misaligned blocks. Material rendering parameters are uniformly checked, including the effective status of channels such as ambient occlusion, reflection, and height / bump, ensuring that the parameters are consistent with the project's art specifications. Game developers can also adjust the material rendering parameters for any building asset based on the original image.
[0120] All building assets are instantiated and arranged to recreate the complete target city scene, configuring the main light source, auxiliary light source, and ambient light parameters. Specifically, the light intensity, color temperature, and direction of illumination can be adjusted to ensure natural transitions between light and shadow on foreground and background buildings in the main city; shadow rendering can also be enabled to verify that the shadows of highly important core buildings have clear outlines and soft edges, while the shadows of lightweight buildings in the distance have no large areas of noise and no rendering stutters; multiple lighting schemes for daytime, dusk, and nighttime can be switched to check that the colors of various building textures are not distorted under different lighting conditions and that the material reflection performance matches the original artwork settings.
[0121] In some examples, the results of four verification items—size, collision, material, and lighting—can be summarized. If all indicators meet the standards, the integration verification of the asset group is deemed to have passed. If problems such as size errors, collision failures, missing material maps, or abnormal lighting rendering occur, the corresponding asset group number is recorded, and the process is returned to the upstream process to reprocess the 3D model, UVs, textures, or material instances. After correction, the data is imported and verified again. After all asset groups have passed verification, the integration of urban scene building resources is completed.
[0122] <Second Embodiment> This embodiment provides an electronic device. Figure 3 A schematic diagram of the hardware structure of the electronic device is shown.
[0123] like Figure 3 As shown, the electronic device 300 includes a processor 310 and a memory 320. The memory can be used to store computer programs, and the processor can be used to retrieve the computer programs from the memory to execute any method embodiment of this disclosure. The processor can be one or more, and these processors can execute instructions individually or jointly. Similarly, the memory can be one or more, and these memories can store the aforementioned computer programs individually or jointly.
[0124] This disclosure also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the methods in any embodiment of this disclosure.
[0125] Optionally, the computer-readable storage medium may be a non-transitory storage medium, but is not limited thereto; it may also be a temporary storage medium.
[0126] This disclosure also provides a computer program product, including a computer program that, when executed by a processor, implements the methods in any embodiment of this disclosure.
[0127] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.
[0128] This disclosure may be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having a computer-readable program loaded thereon for causing a processor to implement any of the methods in the foregoing embodiments of this disclosure.
[0129] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media may include, for example, electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), static random access memory (SRAM), compact disc-read-only memory (CD-ROM), digital versatile disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any combination thereof. The computer-readable storage medium used herein is not to be interpreted as a transient signal itself, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.
[0130] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include one or more of copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to computer-readable storage media in the respective computing / processing device.
[0131] The computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source or object programs written in any combination of one or more programming languages, including object-oriented programming languages (such as Smalltalk, C++, etc.) and conventional procedural programming languages (such as the "C" language or similar programming languages). The computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network (e.g., a local area network or a wide area network), or it may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays, or programmable logic arrays, can execute computer-readable program instructions to implement various aspects of the embodiments of this disclosure by utilizing state information from the computer-readable program instructions.
[0132] Various aspects of this disclosure are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus, and computer program products according to embodiments of this disclosure. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer-readable program instructions.
[0133] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processor of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner; thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.
[0134] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions that execute on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more boxes of a flowchart and / or block diagram.
[0135] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions. It should be noted that implementation in hardware, implementation in software, and implementation using a combination of software and hardware are all equivalent.
[0136] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, and are not limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein. The scope of this disclosure is defined by the appended claims.
Claims
1. A method of three-dimensional city asset production, characterized by, The method includes: Obtain the original artwork of the target city in the target game; Generate a 3D model of each building asset in the target city based on the original artwork; Based on the 3D model of each building asset, generate UV mapping data for each building asset; A texture map for each building asset is generated based on the original artwork and the UV mapping data; The building assets within the target city are merged and grouped to obtain multiple asset groups, and a material instance referencing the corresponding texture is assigned to each asset group; wherein, each asset group includes building assets of at least one style; Build a 3D model exchange file for each asset group based on the 3D model, material instances, UV mapping data, and textures of each asset group; Import the 3D model exchange files of each asset group into the game engine of the target game.
2. The method of claim 1, wherein, The step of generating a 3D model of each building asset in the target city based on the original artwork includes: The original artwork is cleaned for each building asset to obtain a segmented image of the corresponding building asset; each segmented image contains only the corresponding building asset. Generate a 3D model of each building asset based on its segmented image.
3. The method of claim 2, wherein, The step of generating a 3D model of the corresponding building asset based on the segmented image of each building asset includes: A multi-view view of the corresponding building asset is generated based on the segmented image of each building asset, wherein the multi-view view includes a front view, a rear view, a left view, and a right view; Generate a 3D model of each building asset based on its multi-view perspective.
4. The method of claim 3, wherein, The step of generating a multi-view view of each building asset based on its segmented image includes: A three-dimensional framework model of each building asset is constructed using basic geometry; wherein, the basic geometry includes at least one of cube, cuboid, and cylinder; Based on the segmented image and 3D frame model of each building asset, generate multi-view views of the corresponding building asset.
5. The method of claim 1, wherein, The process of assigning a material instance that references the corresponding texture to each asset group includes: The asset groups are categorized based on the importance of the building assets within the urban context. Based on the classification results, assign a material instance that references the corresponding texture to each asset group.
6. The method of claim 5, wherein, The process of assigning a material instance referencing the corresponding texture to each asset group based on the classification results includes: Based on the classification results, obtain the first asset group among the multiple asset groups; Based on the identifier of the UV mapping data corresponding to each building asset in the first asset group, assign a material instance that references the corresponding texture to each building asset in the first asset group.
7. The method according to claim 5, characterized in that, The process of assigning a material instance referencing the corresponding texture to each asset group based on the classification results includes: Based on the classification results, obtain the second asset group from multiple asset groups; Scaling the UV mapping data corresponding to each building asset in the second asset group to the UV mapping layout; According to the UV mapping layout, the textures of each building asset in the second asset group are recombined into a single merged texture; Assign a material instance that references the merged texture to the 3D model of the second asset group.
8. The method of claim 1, wherein, The building assets within the target city are merged and grouped to obtain multiple asset groups, including: According to the asset category of the building assets, building assets of the same asset category shall be merged; The merged construction assets are grouped into multiple asset groups.
9. The method of claim 1, wherein, The method further includes: The 3D models of each asset group are integrated and validated within the game engine.
10. The method of claim 9, wherein, The integration and verification of the 3D models of each asset group in the game engine includes at least one of the following: The game engine performs dimensional verification and collision detection on the 3D models of each building asset. In the game engine, material instances of each asset group are configured with materials. The lighting and shadows of the city scene are adjusted in the game engine.
11. The method of claim 1, wherein, The method further includes: Super-resolution enhancement is performed on textures with resolutions below a threshold.
12. An electronic device, comprising: It includes a memory and a processor, the memory being used to store a computer program; the processor being used to execute the computer program to implement the method according to any one of claims 1-11.
13. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, which, when executed by a processor, implements the method according to any one of claims 1-11.
14. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method of any one of claims 1-11.
