A method and system for generating a map of a game

Abstract

The application discloses a kind of generation method and system of the map of game, and the application relates to the technical field of the map of game, according to the detection of three preliminary maps to determine corresponding map bias important element, based on multiple map bias important element, the role form and role training requirement of game character determine optimal map, optimal map as the stage image of game map, improve the accuracy of optimal map.According to multiple optimal maps, corresponding stage position and the playing goal of game character in each stage determine the map synthesis mode of multiple optimal maps;Based on the map synthesis mode, the stage position and position order of multiple optimal maps determine corresponding overall map, according to the detection of the overall map to determine multiple map abnormal areas, according to multiple map abnormal areas and the virtual drill of game character to determine corresponding optimization area, improve the accuracy of overall map.

Description

Technical Field

[0001] This invention relates to the technical field of game maps, and more particularly to a method and system for generating game maps. Background Technology

[0002] With the development of technology, games have gradually been applied to people's lives and provided for entertainment. In this game, there are characters and game maps. The characters move on the game map. In the process of constructing the game map, the actual scene image materials are captured based solely on the decomposition of the game map requirements to directly output the corresponding game map. The changes of the map in the initial and optimal stages are not considered, which affects the accuracy of the overall map. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the prior art. This invention provides a method and system for generating game maps.

[0004] This invention provides a method for generating game maps, including:

[0005] The initial map combination was determined based on the game's map requirements and multiple real-world scene image materials;

[0006] Multiple primary maps are determined based on the combination of primary maps, and the three best primary maps are determined based on the matching coefficients of the multiple primary maps and the image materials of each actual scene relative to the map requirements.

[0007] Based on the detection of the three best primary maps, the corresponding map bias elements are determined. Based on multiple map bias elements, the character form of the game character and the character training requirements, the best map is determined. The best map serves as a stage image of the game map.

[0008] When a game character needs to traverse the overall map, mark the stage positions of multiple best maps, and determine the map synthesis mode of multiple best maps based on multiple best maps, their corresponding stage positions, and the game character's play objectives at each stage.

[0009] Based on the map synthesis mode, the stage positions and position order of multiple best maps, the corresponding overall map is determined. Based on the detection of the overall map, multiple abnormal map areas are identified. Based on the multiple abnormal map areas and the virtual simulation of the game characters, the corresponding optimization areas are determined to automatically update the overall map.

[0010] This invention provides a game map generation system, which is applied to the aforementioned game map generation method. The game map generation system includes:

[0011] The basic map combination module is used to determine the basic map combination based on the game's map requirements and multiple real-world scene image materials;

[0012] The primary map module is used to determine multiple primary maps based on a combination of primary maps, and to determine the three best primary maps based on the matching coefficients of multiple primary maps and image materials of each actual scene relative to the map requirements.

[0013] The optimal map module is used to determine the corresponding map bias elements based on the detection of the three better primary maps. The optimal map is determined based on multiple map bias elements, the character form of the game character, and the character training requirements. The optimal map serves as a phase image of the game map.

[0014] The map merging mode module is used to mark the stage positions of multiple best maps when the game character needs to traverse the overall map. Based on the multiple best maps, the corresponding stage positions, and the game character's play objectives at each stage, the map merging mode of multiple best maps is determined.

[0015] The map update module is used to determine the corresponding overall map based on the map synthesis mode, the stage position and position order of multiple best maps, identify multiple abnormal map areas based on the detection of the overall map, and determine the corresponding optimization areas based on the multiple abnormal map areas and the virtual simulation of the game characters, so as to automatically update the overall map.

[0016] Compared with the prior art, the beneficial effects of the present invention are:

[0017] In this embodiment of the invention, the method is used to determine a primary map combination based on the game's map requirements and multiple actual scene image materials; multiple primary maps are determined based on the primary map combination; three optimal primary maps are determined based on the matching coefficients of the multiple primary maps and each actual scene image material relative to the map requirements; corresponding map bias elements are determined based on the detection of the three optimal primary maps; and an optimal map is determined based on multiple map bias elements, the character form of the game character, and the character training requirements. The optimal map serves as a stage image of the game map. By introducing three primary maps, the overall consideration of multiple map bias elements, the character form of the game character, and the character training requirements is taken into account, thereby improving the accuracy of the optimal map and completing further image control of the three primary maps.

[0018] Therefore, when a game character needs to traverse the overall map, the stage positions of multiple optimal maps are marked. Based on the multiple optimal maps, their corresponding stage positions, and the game character's play objectives at each stage, a map synthesis mode for the multiple optimal maps is determined. Based on this map synthesis mode, the stage positions and position order of the multiple optimal maps, the corresponding overall map is determined. Based on the detection of this overall map, multiple map anomaly areas are identified. Based on the multiple map anomaly areas and the game character's virtual practice, corresponding optimization areas are identified to autonomously update the overall map. This introduces a map synthesis mode and further synthesizes multiple optimal maps, improving the accuracy of the overall map. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the method for generating a game map in an embodiment of the present invention;

[0020] Figure 2 This is a flowchart illustrating step S11 of the game map generation method in this embodiment of the invention.

[0021] Figure 3 This is a flowchart illustrating step S12 in the method for generating a game map in an embodiment of the present invention.

[0022] Figure 4 This is a flowchart illustrating step S13 in the method for generating a game map in an embodiment of the present invention.

[0023] Figure 5 This is a flowchart illustrating step S14 of the game map generation method in an embodiment of the present invention.

[0024] Figure 6 This is a flowchart illustrating step S15 of the game map generation method in an embodiment of the present invention.

[0025] Figure 7 This is a schematic diagram of the structural composition of the game map generation system in an embodiment of the present invention. Detailed Implementation

[0026] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0027] Please see Figures 1 to 7 A method for generating game maps, applied to map generation scenarios; the method for generating game maps includes:

[0028] Step S11: Determine the initial map combination based on the game's map requirements and multiple real-world scene image materials;

[0029] Step S12: Determine multiple primary maps based on the combination of primary maps, and determine the three best primary maps according to the matching coefficients of the multiple primary maps and each actual scene image material relative to the map requirements;

[0030] Step S13: Determine the corresponding map bias elements based on the detection of the three better primary maps, and determine the best map based on multiple map bias elements, the character form of the game character and the character training requirements. The best map serves as a stage image of the game map.

[0031] Step S14: When the game character needs to traverse the overall map, mark the stage positions of multiple best maps, and determine the map synthesis mode of multiple best maps based on multiple best maps, corresponding stage positions and the game character's play objectives at each stage.

[0032] Step S15: Based on the map synthesis mode, the stage position and position order of multiple best maps, determine the corresponding overall map, determine multiple map abnormal areas based on the detection of the overall map, determine the corresponding optimization areas based on the multiple map abnormal areas and the virtual simulation of the game character, so as to automatically update the overall map.

[0033] refer to Figure 2 In step S11, a primary map combination is determined based on the game's map requirements and multiple real-world scene image materials.

[0034] In the specific implementation of this invention, the specific steps are as follows:

[0035] S111: Collect the game's map requirements, determine multiple sub-map requirements based on the initial analysis of the game's map requirements, determine the corresponding map target based on the cross-composite of multiple sub-map requirements, and trigger the drone to conduct target inspection in the target site according to the map target in order to collect multiple real-world scene image materials.

[0036] S112: Based on the recognition of multiple real-world scene image materials, the priority of multiple real-world scene image materials is determined, and the multiple real-world scene image materials are filtered in terms of image quality and priority to output multiple candidate real-world scene image materials. According to the matching of multiple candidate real-world scene image materials and multiple sub-map requirements, a corresponding primary map combination is constructed. At this time, the primary map combination records the content that the multiple candidate real-world scene image materials match the game's map requirements.

[0037] In the embodiments of this application, the map requirements of the game are collected, multiple sub-map requirements are determined based on the initial parsing of the game's map requirements, the corresponding map target is determined based on the cross-composite of multiple sub-map requirements, and the UAV is triggered to conduct target inspection in the target site according to the map target to collect multiple real scene image materials. This approach takes into account the overall consideration of cross-composite of multiple sub-map requirements and ensures the accuracy of the corresponding map target.

[0038] At this point, the overall design requirements of the map are obtained from the game design document or requirements document, including but not limited to: terrain type: such as mountains, desert, city, forest, etc.; size: such as 1km×1km, 5km×5km; theme style: such as cyberpunk, fantasy, modern city, etc.; functional areas: such as spawn point, resource point, mission area, boss area, etc.; visual style: such as color tone, lighting, atmosphere, etc.; gameplay requirements: such as whether elevation difference, hiding spots, open areas are needed. These are usually obtained through requirements documents, design meetings, design prototypes, etc., and then organized into structured data.

[0039] The overall map requirements are broken down into multiple sub-modules, each corresponding to a sub-map requirement, facilitating regional design and management. The breakdown is based on: functional zoning (e.g., starting village, main city, wilderness dungeons); terrain zoning (e.g., northern snow mountains, southern plains, eastern swamps); story zoning (e.g., Chapter 1 area, Chapter 2 area); and technical zoning (e.g., areas requiring special lighting, areas requiring physical interaction). Structured analysis, modular design, and requirement decomposition methods can be used to break down large requirements into smaller ones.

[0040] By cross-comparing and integrating the requirements of multiple sub-maps, the core map objectives are extracted, and the types and areas of real-world scenes that the drones need to collect are clarified. Cross-combination includes: function and terrain cross-combination: such as the "main city" needing to be located in "plains"; style and plot cross-combination: such as the "abandoned factory" needing to reflect a "post-apocalyptic style"; gameplay and visual cross-combination: such as the "sniper point" needing to be located in "high ground" with "open view"; tools such as requirement matrices, cross-analysis tables, and decision trees can be used to help determine the optimal map objectives.

[0041] Based on the map target, drones are deployed to conduct automatic or semi-automatic inspections of corresponding real-world locations, collecting image data of the actual scene. Suitable real-world locations, such as cities, mountains, and industrial areas, are selected according to the map target. The drone's flight path is planned to ensure coverage of all target areas. Using onboard high-definition cameras, LiDAR, and other equipment, the drones collect image data of terrain, buildings, and vegetation. The collected images are initially labeled, such as terrain type, architectural style, and vegetation density. Autonomous flight, GPS navigation, and obstacle avoidance algorithms are employed to ensure the safety and data quality of the inspection. After the drone inspection is completed, the collected image data is organized, classified, and stored, providing basic data for subsequent map generation. Image classification: Classified by terrain, building, and vegetation types; Quality screening: Blurry, duplicate, and non-compliant images are removed; Data annotation: Detailed annotations are applied to the images, such as coordinates, altitude, and angle; Data storage: Image data is stored in a database for easy retrieval later.

[0042] Furthermore, based on the recognition of multiple real-world scene image materials, the priority of multiple real-world scene image materials is determined, and the multiple real-world scene image materials are filtered in terms of image quality and priority to output multiple candidate real-world scene image materials. According to the matching of multiple candidate real-world scene image materials and multiple sub-map requirements, a corresponding primary map combination is constructed. At this time, the primary map combination records the content that the multiple candidate real-world scene image materials match the game's map requirements, which is compatible with the overall consideration of the recognition of multiple real-world scene image materials and ensures the accuracy of the priority of multiple real-world scene image materials.

[0043] At this point, the purpose of image recognition is to extract key features from each image and evaluate its matching degree with the game map requirements based on these features, thereby determining its priority. The recognition content includes: terrain features: such as mountains, plains, water areas, building complexes, etc.; visual style: such as hue, lighting, architectural style; scene elements: such as vegetation, roads, landmarks, crowds, etc.; image quality: such as resolution, sharpness, noise level, exposure, etc.; functional matching degree: such as whether it is suitable for use as a task area, resource point, spawn point, etc. The priority determination method is to use a weighted scoring model, for example: terrain matching degree: 40%; visual style matching degree: 30%; image quality: 20%; scene element richness: 10%. Each image is scored according to the above dimensions, and the final weighted score is used to determine the priority.

[0044] After prioritizing, the system will perform a secondary screening based on image quality, eliminating low-quality or poorly matched images and retaining high-quality candidate materials. Screening criteria: Image quality threshold: such as resolution not lower than 4K, clarity score >80, and no obvious noise; Priority threshold: such as comprehensive score >70 points; Diversity requirements: avoid too many homogeneous images and ensure coverage of different terrains and styles; Output results: The screened images serve as candidate materials and proceed to the next matching stage.

[0045] The process involves matching candidate images with submap requirements to construct a primary map composite. Matching methods include: each submap requirement corresponds to one or more candidate images; matching criteria include terrain, style, and function; a candidate image list is generated for each submap region; composite construction involves combining the candidate image lists of all submaps into a primary composite of the overall map; the primary composite records the candidate images corresponding to each submap region and their matching degree; furthermore, the primary map composite is not only a collection of images but also records detailed matching information for subsequent evaluation and optimization; the recorded content includes: the name and requirements of each submap region; the corresponding candidate image list; the matching score and matching dimension of each image; and the overall composite score.

[0046] refer to Figure 3 In step S12, multiple primary maps are determined based on the combination of primary maps, and the three best primary maps are determined according to the matching coefficients of the multiple primary maps and each actual scene image material relative to the map requirements.

[0047] In the specific implementation of this invention, the specific steps are as follows:

[0048] S121: Based on the recognition of the primary map combination, determine the image quality parameters of multiple candidate real-world scene image materials, determine the corresponding image selection method based on the image quality parameters of the multiple candidate real-world scene image materials, and determine multiple primary maps based on the multiple candidate real-world scene image materials.

[0049] S122: Collect image materials from various real-world scenes and the map requirements of the game, and determine the corresponding matching coefficient based on the matching between the image materials from various real-world scenes and the map requirements of the game;

[0050] S123: Determine multiple image combinations based on multiple primary maps and corresponding matching coefficients; determine the best image combination based on the selection of multiple image combinations, and output the three best primary maps.

[0051] In the embodiments of this application, image quality parameters of multiple candidate real-world scene image materials are determined based on the identification of the primary map combination, and a corresponding image selection method is determined based on the image quality parameters of the multiple candidate real-world scene image materials. This method determines multiple primary maps based on the multiple candidate real-world scene image materials, taking into account the overall image quality parameters of the multiple candidate real-world scene image materials and ensuring the accuracy of the corresponding image selection method.

[0052] At this point, a technical quality assessment is conducted on the actual scene image materials in the initial map assemblage. First, it's necessary to identify and extract the key quality parameters for each image. These parameters typically include: resolution: the pixel size of the image, affecting the detail in the game; sharpness: the sharpness of the image, affecting the presentation of edges and details; color saturation: the vibrancy of colors, affecting the visual style of the game; contrast: the difference between light and dark, affecting the three-dimensionality and layering of the image; noise level: random brightness or color variations in the image, affecting purity; compositional balance: the rationality of the distribution of image elements, affecting visual aesthetics; lighting conditions: the direction, intensity, and quality of light, affecting atmosphere creation; dynamic range: the difference between the brightest and darkest parts of the image, affecting realism. These parameters can be automatically extracted using computer vision algorithms or combined with manual evaluation for a comprehensive judgment.

[0053] Based on the determined image quality parameters, a suitable image selection method needs to be designed. These methods typically include: weighted scoring: assigning weights to each quality parameter and calculating a comprehensive score for each image; threshold screening: setting a minimum threshold for each parameter and eliminating images that do not meet the criteria; cluster analysis: grouping images according to quality parameters and selecting representative images from each group; multi-objective optimization: balancing multiple quality parameters to find the optimal solution; and machine learning models: using a trained model to predict the suitability of an image. The choice of method depends on the specific needs of the game, the quantity and quality distribution of the image assets, and generally, using a combination of methods yields better results.

[0054] The process involves transforming high-quality real-world scene images into preliminary maps usable for the game; adjusting sizes, correcting colors, and removing unwanted elements; ensuring all images maintain a consistent visual style; identifying and marking key terrain features in the images; converting real-world scenes into images suitable for the game environment; breaking down images into different layers (such as terrain, vegetation, and buildings); and ensuring natural transitions between adjacent images. Through these processes, the original real-world scene images are transformed into preliminary maps that can be directly used for game map generation. These preliminary maps are not only of high technical quality but also have a unified style and clear terrain features, laying a solid foundation for subsequent map generation and game development.

[0055] Furthermore, image materials from various real-world scenarios and the map requirements of the game are collected. Based on the matching of image materials from various real-world scenarios and the map requirements of the game, the corresponding matching coefficients are determined. This approach takes into account the overall matching of image materials from various real-world scenarios and the map requirements of the game, ensuring the accuracy of the corresponding matching coefficients.

[0056] At this point, image materials from various actual scenes and game map requirements are collected. Actual scene image materials: images obtained through on-site photography or digital modeling in the early stages; game map requirements: design document content including terrain, style, functional areas, gameplay mechanics, etc.; image materials: image metadata (such as resolution, shooting location, style tags, etc.) are read in batches from the material library; map requirements: structured requirements are extracted from the game design document (GDD), such as "the starting village should be in a green hilly style" and "the main city should be in a modern urban style".

[0057] The matching coefficient is determined based on the matching of image materials from various actual scenes with the requirements of the game map. The calculation of the matching coefficient is based on the matching degree evaluation of multiple dimensions, usually using a weighted scoring method. The matching dimensions include: terrain matching degree: whether the terrain in the image matches the terrain type required by the map; style matching degree: whether the overall visual style of the image matches the game's art style; element matching degree: whether the image contains the key elements required by the map (such as buildings, vegetation, water bodies, etc.); and function matching degree: whether the image is suitable as the background of a specific functional area (such as a combat area or a rest area). The formula is: Matching coefficient = (terrain matching degree × 0.3) + (style matching degree × 0.3) + (element matching degree × 0.2) + (function matching degree × 0.2).

[0058] The system organizes the matching coefficient calculation results for each image into structured data, usually stored in the form of tables or databases, for easy subsequent filtering and retrieval; the output includes: image ID; matching coefficient; scoring details for each dimension; applicable map area.

[0059] Specifically, assuming the game project is "Fantasy World", its map requirements include: Beginner Village: green hills, low buildings, and abundant vegetation; Main City: stone buildings, dense layout, and a central square; Desert Area: yellow sand terrain, sparse vegetation, and ruins; The image library contains 200 images of different scenes.

[0060] Suppose we have an image A with the following characteristics: Terrain: Green hills (perfectly matches the requirements of the starting village) → Score 100; Style: Cartoon style (consistent with the game style) → Score 90; Elements: Low buildings and abundant vegetation (meets the requirements) → Score 95; Function: Suitable as a background for the starting village → Score 90.

[0061] Matching coefficient = (100 × 0.3) + (90 × 0.3) + (95 × 0.2) + (90 × 0.2) = 30 + 27 + 19 + 18 = 94; Therefore, the matching coefficient of image A is 94; A matching coefficient illustration table is shown in Table 1:

[0062] Table 1: Matching Coefficient Diagram

[0063] Image ID Matching coefficient Terrain rating Style rating Element rating Functionality rating Applicable Areas A001 94 100 90 95 90 Beginner Village A002 85 80 90 85 85 Main City A003 78 70 85 80 75 desert areas

[0064] By using multi-dimensional weighted scoring, the degree of matching between each image and the map requirements is objectively evaluated; the matching coefficient provides a quantitative basis for subsequent image selection and map combination; it avoids the subjectivity of manual selection and improves the efficiency of material utilization; and it provides a replicable and scalable evaluation method for large-scale map generation.

[0065] Therefore, multiple image combinations are determined based on multiple primary maps and their corresponding matching coefficients; the optimal image combination is determined based on the selection of multiple image combinations to output the three best primary maps. This approach takes into account the overall consideration of multiple primary maps and their corresponding matching coefficients, ensuring the accuracy of multiple image combinations.

[0066] At this point, multiple image combinations are determined based on multiple primary maps and their corresponding matching coefficients. Input: A set of primary maps, each image having its matching coefficient calculated via S122. Objective: Generate multiple image combinations, each consisting of several images, representing a map stitching scheme. Simultaneously, images with high scores are prioritized based on matching coefficients. The images are combined according to map structure (e.g., "beginner village - main city - desert"). Within each combination, the images must satisfy basic coherence in terms of style, terrain, and transitions.

[0067] The optimal image combination is determined through a selection process based on multiple image combinations. Selection criteria include: average matching coefficient of images within the combination; stylistic consistency between images (e.g., a natural transition between the starting village and the main city); map structural rationality (e.g., reasonable distribution of functional areas); and special requirements (e.g., whether it includes key landmarks or unique terrain). Each combination is scored, for example: average matching coefficient × 0.5; stylistic consistency score × 0.3; structural rationality score × 0.2. The top combinations are ranked by total score, and the best combinations are selected. Representative images are extracted from these optimal combinations to form the three final primary maps. Each image represents a typical style or region, ensuring the diversity and quality of subsequent map generation.

[0068] Specifically, the game project "Fantasy World" required the generation of a basic map containing a starting village, a main city, and a desert. Nine actual scene images were collected, and the matching coefficient of each image was calculated using S122. Five different image combinations were constructed, each containing one image each of the starting village, the main city, and the desert. Each combination was scored across multiple dimensions, and three best combinations were selected. Three optimal images were extracted from the best combinations: starting village: C001; main city: C002; desert: C003. The three output images not only had high matching coefficients but also had a consistent style and reasonable structure, providing a high-quality material foundation for subsequent map generation.

[0069] The three output images not only have a high matching coefficient, but also have a consistent style and reasonable structure, providing a high-quality material foundation for subsequent map generation. Through multi-dimensional scoring, the advantages and disadvantages of each combination were objectively evaluated. Finally, the three best primary maps were selected to ensure the diversity and quality of map generation.

[0070] refer to Figure 4 In step S13, the corresponding map bias elements are determined based on the detection of the three better primary maps. The best map is determined based on multiple map bias elements, the character form of the game character and the character training requirements. The best map is used as a stage image of the game map.

[0071] In the specific implementation of this invention, the specific steps are as follows:

[0072] S131: Perform image detection on the three best primary maps and output the corresponding map bias regions. Determine multiple map features based on the region recognition of each map bias region. Determine the corresponding map bias elements based on the multiple map features and the image content presented by the corresponding primary maps.

[0073] S132: Collect basic information of game characters, and determine the character form and training requirements of game characters based on the recognition of basic information of game characters. Determine the first image control content based on multiple map bias elements and the character form of game characters, and determine the second image control content based on multiple map bias elements and character training requirements.

[0074] S133: The optimal map is determined based on the synthesis of the first image control content, the second image control content, and the three better primary maps. The optimal map is presented in the corresponding stage as the corresponding stage image.

[0075] In the embodiments of this application, image detection is performed on three preferred primary maps, and corresponding map bias regions are output. Multiple map features are determined based on the region recognition of each map bias region. The corresponding map bias elements are determined based on the multiple map features and the image content presented by the corresponding primary maps. This approach takes into account the overall consideration of multiple map features and the image content presented by the corresponding primary maps, ensuring the accuracy of the corresponding map bias elements.

[0076] At this point, image detection is performed on the three best primary maps, and the corresponding map bias regions are output. Input: three selected primary maps (such as the beginner village, the main city, and the desert); Objective: to identify visually or functionally "prominent" or "concentrated" regions in each image; using image recognition algorithms (such as edge detection, color clustering, semantic segmentation, etc.); to detect regions with significant visual differences or functional significance; and output these regions as "map bias regions".

[0077] Based on the region identification of each map's emphasis area, multiple map features are determined. Input: List of map emphasis areas; Objective: Extract "map features" from each emphasis area; Perform feature extraction on each emphasis area, such as terrain, buildings, vegetation, roads, etc.; Use image classification or feature description algorithms to identify the representative features of the area; Output: List of "map features".

[0078] Based on multiple map features and the image content presented by the corresponding primary map, determine the corresponding map bias elements. Input: map features + original image content; Objective: combine features with image content to determine "map bias elements"; perform semantic interpretation on each map feature, combined with game design requirements; determine the "importance" or "functional significance" of the feature in the map; output is "map bias elements".

[0079] Specifically, the game project "Fantasy World" has completed the initial map screening. Now, image detection and feature analysis need to be performed on three maps (starter village, main city, and desert). The three images C001, C002, and C003 are detected to identify their respective dominant areas. Feature analysis is performed on each dominant area to extract features such as buildings, terrain, and vegetation. Combining these features with the image content, the functional significance of each dominant area in the game is determined. Results: The "Central Hut Area" in the starter village (C001) has been identified as the player spawn point; the "Central Square" in the main city (C002) has been identified as the commercial and quest center; and the "Southeast Oasis" in the desert (C003) has been identified as a supply and hidden quest location.

[0080] Furthermore, basic information about the game characters is collected, and the character form and training requirements are determined based on the identification of the basic information. The first image control content is determined based on multiple map bias elements and the character form of the game characters. The second image control content is determined based on multiple map bias elements and character training requirements. This overall consideration of multiple map bias elements and character training requirements ensures the accuracy of the second image control content.

[0081] At this point, the input is: basic data of the game character, including character level, class, skills, equipment, attributes, etc.; the goal is: to obtain basic information about the character to provide data support for subsequent analysis; to extract character data from the game database; to standardize the data to ensure consistent format; and to output structured basic character information.

[0082] Specifically, let's assume we have a character basic information table: Character ID: P001; Character Name: Shadow Blade; Class: Assassin; Level: 25; Main Attributes: Agility, Critical Hit; Equipment: Dual-wielding Daggers, Light Armor; Skills: Stealth, Backstab, Poison Blade; Game Progress: Completed the newbie village quests, just entered the main city.

[0083] Input: Structured basic character information; Objective: Identify the character's morphological characteristics and training needs; Analyze character data using a rule engine or machine learning model; Determine the character's form (e.g., melee, ranged, tank) based on profession, attributes, skills, etc.; Determine the character's training requirements (e.g., skills to be improved, suitable challenges, etc.) based on level, progress, equipment, etc.; Output: Character form and character training requirements.

[0084] Specifically, based on the basic information of P001: Character form identification: Class is Assassin, main attributes are Agility and Critical Hit, equipped with light armor and dual daggers; skills are biased towards single-target burst and stealth; Conclusion: Character form is "highly mobile melee assassin"; Character training requirements identification: Level 25, just entered the main city, needs to adapt to a more complex combat environment; Equipment is basic light armor and daggers, needs to upgrade equipment; Skills are mainly single-target attacks, needs to improve group combat ability; Conclusion: Character training requirements are "upgrade equipment level, learn group combat skills, adapt to complex terrain combat".

[0085] The first image adjustment content is determined based on multiple map bias elements and the character form of the game character. Input: map bias elements (from S131) ​​and character form; Objective: adjust the visual appearance of the map according to the character form; analyze the matching degree between the character form and the map bias elements; determine the image elements that need to be adjusted (such as lighting, color, texture, etc.); generate the first image adjustment content, mainly focusing on the optimization of visual experience.

[0086] Specifically, assuming the map's key elements include: the main city's central square (commercial area, bright and open); and the main city's dark alleys (hidden areas, dark and narrow); combined with the character form of a "highly mobile melee assassin": the first image adjustment content is as follows: Central Square: Enhance light and shadow contrast, highlight usable shadow areas; increase visual cues for climbable buildings; Dark Alleys: Reduce overall brightness, enhance the visual effect of hidden areas; increase the detail of objects that can be hidden; Overall: Improve edge clarity, making it easier for assassins to judge movement routes; enhance the visual feedback of dynamic elements.

[0087] The second image control content is determined based on multiple map biases and character training requirements. Input: map biases and character training requirements; Objective: adjust the map's functional performance according to character training needs; analyze the correlation between character training requirements and map biases; determine the functional elements that need to be adjusted (such as mission trigger points, training areas, enemy distribution, etc.); generate the second image control content, mainly focusing on the functional optimization of the game experience.

[0088] Specifically, in line with the training requirements of "upgrading equipment levels, learning group combat skills, and adapting to combat in complex terrain": Second image adjustment content: Central Square: Increase the visual prominence of equipment vendors; add signs for group combat training grounds; Dark Alley: Set up training points for complex terrain; add prompts for high-level equipment that can be picked up; Overall: Adjust enemy distribution, add enemy combinations suitable for group combat; set up guidance signs for terrain utilization training.

[0089] Therefore, the optimal map is determined based on the synthesis of the first image control content, the second image control content, and the three best primary maps. The optimal map is presented in the corresponding stage as the corresponding stage image. This takes into account the overall consideration of the synthesis of the first image control content, the second image control content, and the three best primary maps, ensuring the accuracy of the optimal map. At the same time, the introduction of three primary maps takes into account the overall consideration of multiple map bias elements, the character form of the game character, and the character training requirements, improving the accuracy of the optimal map and completing further image control of the three primary maps.

[0090] At this point, the optimal map is determined based on the synthesis of the first image control content, the second image control content, and three better primary maps. The first image control content and the second image control content are integrated. Input: first image control content (visual level) and second image control content (functional level); Objective: to integrate the two types of control content into a unified control scheme; to establish a control content mapping table and clarify the specific parameters of each control; to ensure that visual control and functional control do not conflict; and to generate a comprehensive control instruction set.

[0091] The process involves combining the control content with three optimal primary maps. Inputs include: a comprehensive control command set and three primary maps. The objective is to apply the control content to the primary maps. The primary maps are then broken down into multiple layers. The parameters of each layer are adjusted according to the control commands. The adjusted layers are then recombined into a new image. Simultaneously, inputs include: multiple composite maps. The objective is to select the best map from the composite results. An evaluation index system is established (visual quality, functional matching, performance, etc.). Each composite image is scored. The image with the highest overall score is selected as the best map. The best map is presented in the corresponding stage as the stage image.

[0092] refer to Figure 5 In step S14, when the game character needs to traverse the overall map, the stage positions of multiple best maps are marked, and the map synthesis mode of multiple best maps is determined based on the multiple best maps, the corresponding stage positions, and the game character's play objectives at each stage.

[0093] In the specific implementation of this invention, the specific steps are as follows:

[0094] S141: Match the corresponding best map in different stages of the game map, monitor the position of the game character on the game map in real time, and trigger the overall construction of the game map based on the traversal of the game map by the game character. At this time, determine the stage position of multiple best maps based on the detection of the best map.

[0095] S142: Collect image content from multiple optimal maps, determine the first synthesis coefficient based on the image content of multiple optimal maps and the corresponding stage position, and determine the second synthesis coefficient based on the stage position of multiple optimal maps and the game character's play objectives at each stage.

[0096] S143: Determine the map synthesis mode of multiple optimal maps based on the first synthesis coefficient, the second synthesis coefficient, and the mapping relationship of map synthesis modes.

[0097] In the embodiments of this application, the corresponding best map is matched in different stages of the game map, the position of the game character on the game map is monitored in real time, and the overall construction of the game map is triggered according to the traversal of the game map by the game character. At this time, the stage position of multiple best maps is determined based on the detection of the best map, which is compatible with the overall consideration of the detection of the best map and ensures the accuracy of the stage position of multiple best maps.

[0098] At this point, the system will match the "best maps" generated in the early stages to different game stages based on the phased goals of the game design (such as chapters, quest lines, plot progression, etc.). Each stage corresponds to one or more maps, and these images represent the main scenes or environments of that stage. The matching criteria are as follows: plot development: for example, "newbie village" corresponds to "forest map", "mid-to-late game" corresponds to "desert map"; mission type: for example, "infiltration mission" is suitable for "urban street battle map", "wilderness survival" is suitable for "forest / wilderness map"; gameplay mechanics: for example, "flying mission" is suitable for "canyon map", "underwater exploration" is suitable for "lake / seabed map".

[0099] The system obtains the character's world coordinates in real time through the game engine's built-in positioning system (such as Unity's Transform and Unreal's ActorLocation) and maps them to the map grid or tile system; it uses grid partitioning or quadtree structures to efficiently manage the character's position; it sets "hotspot zones" to trigger specific events or map loading when the character enters these zones; and it records data such as the character's movement path, dwell time, and exploration progress to dynamically adjust the map generation strategy.

[0100] When a character explores a certain area or completes specific conditions, the system will trigger the overall construction of the map. This process includes dynamically loading new areas, updating the content of existing areas, and adjusting the complexity or difficulty of the map. Examples of trigger conditions: the character reaches the map boundary or the entrance to a new area; completes a main quest or side quest; the exploration progress reaches a preset threshold (such as "60% of the current area has been explored"); the character's level or equipment reaches a specific requirement; at the same time, new terrain, buildings, NPCs, enemies, etc. are loaded; environmental parameters such as lighting, weather, and time are adjusted; and new quest points, resource points, or event points are generated.

[0101] Image processing techniques (such as edge detection, feature point matching, and semantic segmentation) are used to analyze the content of the best maps and determine their relative positions and orientations within the overall map. Simultaneously, map boundaries are identified to determine the stitching locations. Image feature points are extracted using algorithms such as SIFT, SURF, or ORB for map alignment and fusion. The maps are divided into different semantic regions (such as "water areas," "roads," and "buildings") for logical map partitioning. Output results include: the coordinates of each best map within the overall map (such as (x,y) coordinates or grid index); the relative orientations between maps (such as "the forest map is north of the desert map"); and the map stitching method (such as "seamless stitching" or "connected via portal").

[0102] The system enables the transformation from static maps to dynamic game worlds; it not only matches the best map to different game stages, but also ensures smooth map loading and seamless transitions through real-time monitoring and dynamic triggering mechanisms. This approach is particularly suitable for large open-world games and can effectively enhance players' immersion and gaming experience.

[0103] Specifically, assuming the game "Wilderness Adventure" is divided into three stages: beginner stage, growth stage, and ultimate stage, the corresponding best maps are the forest map, desert map, and ancient ruins map, respectively; the system will match these three maps to the corresponding game stages.

[0104] After the game starts, the character's initial position is at coordinates (0,0) on the forest map. The system updates the character's position every second and records its movement trajectory. When the character moves to the boundary of the forest map (e.g., (1000,500)), the system detects that the character is about to enter a new area. When the character reaches the exit of the forest map (coordinates (1000,500)), the system triggers map construction: loading the terrain, enemies, and resource points of the desert map; adjusting environmental parameters: changing from "cloudy" in the forest to "sunny" in the desert; generating a new task point: finding hidden treasure chests in the desert.

[0105] By edge detection, the system identifies the eastern boundary of the forest map and the western boundary of the desert map; using feature point matching, the system aligns the boundaries of the two maps to ensure seamless stitching; finally, the system determines the location of the forest map to be (0,0) to (1000,1000); the location of the desert map to be (1000,0) to (2000,1000); and the stitching method for the two maps to be seamless (direct connection).

[0106] The system enables the transformation from static maps to dynamic game worlds. Taking Wilderness Adventure as an example, we not only match the best map to different game stages, but also ensure smooth map loading and seamless transitions through real-time monitoring and dynamic triggering mechanisms. This method is particularly suitable for large open-world games and can effectively enhance players' immersion and gaming experience.

[0107] Furthermore, image content from multiple optimal maps is collected. A first composite coefficient is determined based on the image content of multiple optimal maps and their corresponding stage positions. A second composite coefficient is determined based on the stage positions of multiple optimal maps and the game character's play objectives at each stage. This approach takes into account the overall consideration of the stage positions of multiple optimal maps and the game character's play objectives at each stage, ensuring the accuracy of the second composite coefficient.

[0108] At this point, visual and structural information of each optimal map is collected for subsequent calculation of synthesis coefficients. Image content includes, but is not limited to: visual features such as color distribution, texture type, light intensity, building density, etc.; terrain structure such as elevation map, water distribution, road network, vegetation coverage, etc.; scene elements such as buildings, NPCs, props, enemy distribution, etc.; style consistency such as art style, color saturation, model detail, etc. These contents can be collected through image recognition algorithms (such as CNN, OpenCV feature extraction) or manual annotation (such as designers marking key areas).

[0109] The first composite coefficient is determined based on the image content and corresponding stage positions of multiple best maps. The first composite coefficient is used to measure the visual and structural compatibility of different maps and is an important basis for map stitching and transition effects. The calculation method is as follows: Image content similarity: compare the color distribution, texture style, terrain structure, etc. of two maps and calculate the similarity score; Stage position relationship: consider the spatial relationship of the maps in the game world (such as adjacent, overlapping, interval, etc.) and adjust the weights.

[0110] First formula for the coefficient of composition:

[0111] Among them, α, β, and γ are weighting coefficients that can be adjusted according to game requirements; the first synthesis coefficient is a value between 0 and 1, and the higher the value, the more suitable the two maps are for seamless stitching.

[0112] The second synthesis coefficient is determined based on the stage positions of multiple optimal maps and the game character's play objectives at each stage. The second synthesis coefficient is used to measure whether the map splicing conforms to the gameplay and plot logic. The calculation basis includes: stage position relationship: such as whether the maps are adjacent, whether they belong to the same task line, whether there is plot connection, etc.; play objective matching degree: such as whether the character's goal in the current stage (exploration, combat, puzzle solving) is consistent with the goal of the next stage; difficulty curve adaptation: such as whether the enemy strength and resource distribution of the map conform to the character's growth curve.

[0113] Second formula for the coefficient of composition:

[0114]

[0115] Among them, δ, ϵ, and ζ are weighting coefficients; the second composite coefficient is also a value between 0 and 1, and the higher the value, the greater the improvement of the game experience by map stitching.

[0116] Specifically, let's say we're developing an open-world RPG called "Wilderness Adventure," which includes three main phases: Starter Village (Forest Map): Characters learn basic controls, with the objective of "collecting 10 herbs"; Transfer Station (Mountain Map): Characters learn combat skills, with the objective of "defeating 5 wolves"; Final Boss Area (Volcano Map): Characters challenge the final boss, with the objective of "defeating the volcano dragon."

[0117] Forest Map: Visual Features: Predominantly green, soft lighting, dense vegetation; Terrain Structure: Mainly flat terrain with a few small hills; Scene Elements: Dense distribution of herbs, relatively weak enemies (such as rabbits and slimes); Mountain Map: Visual Features: Predominantly brown, strong lighting, distinct rock textures; Terrain Structure: Significantly undulating, with cliffs and caves; Scene Elements: Numerous wild wolves, a few treasure chests; Volcano Map: Visual Features: Predominantly red and black, intense lighting, flowing lava; Terrain Structure: Extremely undulating, with lava pools and craters; Scene Elements: The volcano dragon is located in the center of the map, surrounded by elite monsters.

[0118] Calculate the first composite coefficient: forest map → mountain map; visual similarity: 0.6 (natural transition from green to brown); structural similarity: 0.7 (reasonable transition from flat to undulating); location weight: 0.9 (adjacent maps);

[0119] ;

[0120] Mountain map → Volcano map: Visual similarity: 0.4 (strong contrast between brown and red); Structural similarity: 0.8 (logically reasonable transition between undulation and extreme undulation); Location weight: 0.8 (adjacent maps);

[0121] ;

[0122] Calculate the second composite coefficient:

[0123] Forest Map → Mountain Map: Objective Consistency: 0.8 (Gathering → Combat, different task types but logically coherent); Difficulty Suitability: 0.9 (Weak enemies → Medium enemies, reasonable difficulty curve); Story Continuity: 0.7 (Beginner → Advanced, natural story progression);

[0124] ;

[0125] Mountain Map → Volcano Map: Objective Consistency: 0.9 (Combat → Boss Battle, highly consistent mission type); Difficulty Suitability: 0.8 (Medium Enemies → Elite Enemies, reasonable difficulty increase); Story Continuity: 0.9 (Advanced → Final Challenge, story climax);

[0126] ;

[0127] Forest → Mountain: Good visual and structural compatibility (first composite coefficient = 0.69), excellent gameplay and story integration (second composite coefficient = 0.82); Mountain → Volcano: Strong visual contrast (first composite coefficient = 0.56), but gameplay and story are highly consistent (second composite coefficient = 0.86); Application suggestions: For transitions with a low first composite coefficient (such as mountain → volcano), a transition area (such as a lava zone) can be added to alleviate visual abruptness; for transitions with a high second composite coefficient, they can be prioritized for main quest progression to ensure a smooth player experience. This method not only ensures the visual rationality of map stitching but also takes into account the logic of gameplay, making it an important technical means for constructing large open-world maps.

[0128] Therefore, the map synthesis mode of multiple optimal maps is determined based on the mapping relationship between the first synthesis coefficient, the second synthesis coefficient, and the map synthesis mode. This takes into account the overall consideration of the mapping relationship between the first synthesis coefficient, the second synthesis coefficient, and the map synthesis mode, and ensures the accuracy of the map synthesis mode of multiple optimal maps.

[0129] At this point, the optimal map synthesis mode for multiple maps is determined based on the first synthesis coefficient, the second synthesis coefficient, and the mapping relationship of map synthesis modes. The first synthesis coefficient represents the visual and structural compatibility of the map (range: 0~1, the higher the coefficient, the more compatible the map). The second synthesis coefficient represents the continuity of the map in terms of gameplay and storyline (range: 0~1, the higher the coefficient, the more coherent the map). Map synthesis mode mapping relationship: a predefined rule table used to select the appropriate map synthesis mode based on the values ​​of the first synthesis coefficient and the second synthesis coefficient.

[0130] Map compositing patterns refer to the methods of stitching or merging two or more maps. A map compositing pattern matching table is collected, as shown in Table 2:

[0131] Table 2: Schematic Diagram of Map Synthesis Mode

[0132] Pattern Number Mode Name Applicable Scenarios Features M1 Seamless splicing Both the first and second synthesis coefficients are relatively high. The visuals and gameplay transition seamlessly, and players barely notice the change. M2 Gradual transition The first synthesis coefficient is lower, and the second synthesis coefficient is higher. Visually, the transitions are achieved through gradient layers (such as fog and light and shadow), creating a seamless gameplay experience. M3 Portal Both the first and second synthesis coefficients are low. Forced switching via specific trigger points (such as portals) is suitable for plot twists. M4 Dynamic loading The first synthesis coefficient is higher, and the second synthesis coefficient is lower. The visuals are natural, but the gameplay is clearly divided (such as different mission areas). M5 Plot switch The first synthesis coefficient is lower, and the second synthesis coefficient is higher. Plot

[0133] The data acquisition mapping rule illustration table is shown in Table 3:

[0134] Table 3: Mapping Rules Illustration

[0135] First range of synthesis coefficients Second synthesis coefficient range Recommended mode >0.7 >0.7 M1 (Seamless splicing) <0.6 >0.7 M2 (gradual transition) <0.6 <0.6 M3 (Portal Type) >0.7 <0.6 M4 (Dynamically Loaded) 0.6~0.7 >0.7 M5 (Story Switch)

[0136] Based on the values ​​of the first and second composite coefficients, refer to the table to select the most suitable composite mode; for example: if the first composite coefficient = 0.8 and the second composite coefficient = 0.9 → select M1 (seamless splicing); if the first composite coefficient = 0.5 and the second composite coefficient = 0.8 → select M2 (gradual transition); if the first composite coefficient = 0.4 and the second composite coefficient = 0.5 → select M3 (portal style).

[0137] Specifically, the transition from forest to mountainous terrain: First synthesis coefficient = 0.69, second synthesis coefficient = 0.82; if the first synthesis coefficient is between 0.6 and 0.7, and the second synthesis coefficient is > 0.7, then the recommended mode is M5 (story transition); Implementation: A story event (such as a flash flood blocking the road) is set between the forest and the mountains. Players need to complete specific tasks (such as building a bridge) to enter the mountains; Visually, the transition is achieved through a flash flood scene, which, although less natural, creates a tighter connection to the storyline.

[0138] Mountain to Volcano Transition: First synthesis coefficient = 0.56, second synthesis coefficient = 0.86; First synthesis coefficient < 0.6, second synthesis coefficient > 0.7 → Recommended mode M2 ​​(gradual transition); Implementation: Design a lava valley transition zone between the mountains and the volcano, visually achieving the transition through gradually increasing lava and smoke effects; Gameplay remains consistent, players can freely explore the transition area, but the difficulty gradually increases.

[0139] The final compositing mode diagram is shown in Table 4.

[0140] Table 4. Schematic diagram of the final synthesis mode

[0141] Map transition First Combination Coefficient Second Combination Coefficient Synthesis Mode Implementation Forest → Mountain 0.69 0.82 M5 (Story Switch) Flash flood event triggers mission Mountains → Volcanoes 0.56 0.86 M2 (gradual transition) Lava Valley Gradient Zone

[0142] refer to Figure 6 In step S15, the corresponding overall map is determined based on the map synthesis mode, the stage position and position order of multiple best maps, multiple map abnormal areas are determined based on the detection of the overall map, and the corresponding optimization area is determined based on the multiple map abnormal areas and the virtual simulation of the game character, so as to automatically update the overall map.

[0143] In the specific implementation of this invention, the specific steps are as follows:

[0144] S151: Collect the map compositing mode, determine the map compositing information based on the parsing of the map compositing mode, determine multiple sub-map compositing projects based on the identification of the map compositing information, trigger the compositing of multiple best maps in different dimensions based on the multiple sub-map compositing projects, at this time, collect the stage position and position order of multiple best maps, and determine the corresponding overall map based on the compositing of multiple best maps in different dimensions based on the stage position and position order of multiple best maps.

[0145] S152: Perform anomaly detection on the overall map, determine multiple abnormal map features based on the anomaly detection of the overall map, and determine multiple map anomaly areas based on the shape, location and regional distribution of the multiple abnormal map features and the overall map.

[0146] S153: Collect multiple abnormal map areas and game characters, and conduct virtual simulations of the game characters over multiple abnormal map areas to present the virtual effects of the game characters traversing the multiple abnormal map areas. Based on the virtual effects, deduce the corresponding optimization areas, and trigger the overall map to update automatically based on the replacement of abnormal map areas by the optimization areas.

[0147] In the embodiments of this application, the map synthesis mode is collected, map synthesis information is determined based on the parsing of the map synthesis mode, multiple sub-map synthesis projects are determined based on the identification of the map synthesis information, and the synthesis of multiple optimal maps in different dimensions is triggered based on the multiple sub-map synthesis projects. At this time, the stage position and position order of the multiple optimal maps are collected, and the corresponding overall map is determined based on the synthesis of the stage position and position order of the multiple optimal maps in different dimensions. This approach incorporates the overall consideration of map synthesis information identification and ensures the accuracy of the multiple sub-map synthesis projects.

[0148] At this point, the map blending mode selected in S143 (such as "gradient transition") is converted into specific blending information, including: blending method: whether it is "horizontal blending", "vertical blending" or "center radial"; transition mechanism: whether it is "color gradient", "terrain smoothing", "obstruction transition", etc.; triggering conditions: such as "player reaches a certain coordinate" or "completes a certain task"; transition range: such as "10 pixels" or "50 meters distance", etc. This information will guide the subsequent blending behavior.

[0149] The overall map composition task is broken down into multiple sub-map composition projects. Each project is responsible for the composition work of one dimension, such as: terrain dimension: responsible for height map, slope, and water distribution; texture dimension: responsible for surface material and vegetation distribution; lighting dimension: responsible for shadows, global illumination, and weather effects; gameplay dimension: responsible for NPC distribution, quest points, and path planning. Each sub-project is executed independently, but they need to work together to form a unified map.

[0150] In each dimension, specific compositing operations are performed according to the requirements of the sub-map compositing project; for example: terrain dimension: smooth the edges of forests and mountains to form a natural transition; texture dimension: gradually replace grass textures with rock textures in the transition area from forest to mountains; lighting dimension: adjust the lighting intensity in the transition area to simulate changes in natural light; gameplay dimension: add rest points and supply stations in the transition area to enhance the player experience.

[0151] Record the positional relationship and order of each best map within the overall map. For example, stage position: such as "Forest (starting point)", "Mountain (mid-game)", "Volcano (end point)"; positional order: such as "from south to north", "from low to high", "from beginner area to advanced area". This information is used to ensure the spatial and logical coherence of the map. By combining all the above information, the composite results of each dimension are integrated into a complete overall map. This map is not only visually coherent, but also logically reasonable in terms of gameplay and plot.

[0152] Furthermore, anomaly detection is performed on the overall map, and multiple abnormal map features are identified based on the anomaly detection of the overall map. Multiple abnormal map regions are identified based on the shape, location, and regional distribution of the multiple abnormal map features and the overall map. This comprehensive consideration of the shape, location, and regional distribution of the multiple abnormal map features ensures the accuracy of the multiple abnormal map regions.

[0153] At this point, a comprehensive scan and analysis of the overall map is performed to identify anomalies. Anomaly detection typically includes the following aspects: terrain anomalies: such as cliff breaks, abrupt slope changes, and asymmetrical water features; texture anomalies: such as texture misalignment, color abrupt changes, and material mismatches; lighting anomalies: such as uneven lighting, misaligned shadows, and abrupt changes in lighting intensity; gameplay anomalies: such as interrupted paths, abnormal obstacles, and missing task points. Detection methods typically include: automatic algorithm detection: such as edge detection, texture analysis, and lighting calculation; manual review: manually checked by artists or designers; and player feedback: player reports collected through test versions.

[0154] After an anomalies are detected, they need to be transformed into specific anomaly map features, that is, the anomalies are classified and described. Anomaly features typically include: morphological features: such as broken lines, protrusions, depressions, etc.; location features: such as coordinates (100,200), regions A and B, etc.; distribution features: such as concentrated distribution, dispersed distribution, linear distribution, etc.; severity: such as minor, moderate, severe, etc. These features will help the development team quickly understand the nature and scope of impact of the anomalies.

[0155] The scattered anomaly features are integrated into specific map anomaly areas, that is, the specific areas in the map that need to be repaired are determined; the map is divided into multiple anomaly areas according to the distribution of anomalies; the boundary range of each anomaly area is defined; the priority of repair is determined according to the severity and impact of the anomalies; repair recommendations: a preliminary repair plan is proposed for each anomaly area.

[0156] Therefore, multiple map anomaly areas and game characters are collected, and the game characters are used to virtually simulate the map anomaly areas to present the virtual effects of the map anomaly areas during the game characters' traversal. Based on these virtual effects, corresponding optimization areas are deduced. The replacement of map anomaly areas by optimization areas triggers the autonomous update of the overall map. This method introduces a map merging mode and further merges multiple best maps to improve the accuracy of the overall map.

[0157] At this point, collect the abnormal map areas that need optimization and the game characters used for testing; specifically, this includes: abnormal area collection: select the area to be used for virtual simulation from the multiple abnormal map areas identified in S152; game character collection: select representative game characters, including characters of different types (such as warriors, mages, assassins, etc.), different abilities (such as jump height, movement speed, climbing ability, etc.), and different equipment (such as climbing tools, flying items, etc.); the collection methods usually include: reading abnormal area information from the database; selecting test characters from the character library; setting test parameters (such as character initial position, target position, test conditions, etc.).

[0158] The virtual simulation involves simulating a game character's traversal of anomaly zones, observing and recording the impact of anomalies on the character's behavior. Virtual simulations typically include the following aspects: path planning: how the character plans the path from the starting point to the destination; movement behavior: the character's movement methods within the anomaly zone (e.g., walking, running, jumping, climbing); interaction behavior: the character's interaction with the anomaly zone (e.g., triggering mechanisms, using items, interacting with the environment); and feedback recording: recording the character's behavioral data within the anomaly zone (e.g., movement time, number of failures, lag). Simulation methods typically include: automated testing: using AI to control the character for automated simulations; manual testing: testers manually controlling the character for simulations; and hybrid testing: combining the advantages of automated and manual testing.

[0159] Presenting the results of virtual simulations in a visual manner facilitates the analysis of the impact of anomalies on the game experience. Virtual effects typically include the following aspects: visual effects: such as characters getting stuck in cracks, falling into the void, or passing through walls; gameplay effects: such as tasks being impossible to complete, paths being blocked, or resources being unavailable; performance effects: such as frame rate drops, loading delays, and memory overflows; and experience effects: such as player frustration, disorientation, and decreased immersion. Presentation methods typically include: heatmaps: showing the movement density and dwell time of characters in abnormal areas; pathmaps: showing the movement trajectory and failure points of characters; data charts: displaying performance indicators and experience scores; and video replays: recording the behavior of characters in abnormal areas.

[0160] Based on the results of the virtual simulation, analyze the impact of the anomaly on the game experience and deduce the areas that need optimization. The determination of optimization areas is usually based on the following factors: Impact level: the extent of the anomaly's impact on the game experience (e.g., whether it causes mission failure, whether it affects core gameplay, etc.); Occurrence frequency: the frequency with which the anomaly occurs during the game (e.g., whether it is triggered every time you pass by, or whether it occurs occasionally); Repair difficulty: the workload and complexity required to repair the anomaly (e.g., simple terrain adjustment or redesign of the area); Priority: the optimization priority determined by a combination of impact level, occurrence frequency, and repair difficulty. The impact level of the anomaly is determined by statistically analyzing the simulation data. The priority of the anomaly is jointly evaluated by team members such as designers, artists, and programmers. Player evaluations and suggestions on the anomaly are collected through test versions.

[0161] Based on the identified optimization areas, abnormal areas in the original map are replaced or modified, triggering an automatic update of the overall map. Autonomous updates typically include the following steps: Optimization scheme formulation: Based on the characteristics of the optimization area, a specific optimization scheme is formulated (e.g., adjusting terrain height, modifying texture maps, redesigning paths, etc.); Optimization resource generation: Optimized map resources are generated (e.g., new terrain models, texture maps, colliders, etc.); Abnormal area replacement: Abnormal areas in the original map are replaced with optimized resources; Map integration update: The replaced areas are integrated with the overall map to ensure seamless integration; Quality verification: The updated map undergoes quality checks to ensure that the optimization is effective and has not introduced new problems. Update methods typically include: Automated update: Resource replacement and integration are completed automatically using tools; Semi-automatic update: Some tasks are completed by tools, while others require manual intervention; Manual update: Optimization and update are completed entirely manually.

[0162] Please see Figure 7 , Figure 7 This is a schematic diagram of the structural composition of the game map generation system in an embodiment of the present invention; the game map generation system includes:

[0163] The primary map combination module 21 is used to determine the primary map combination based on the game's map requirements and multiple real-world scene image materials;

[0164] The primary map module 22 is used to determine multiple primary maps based on the combination of primary maps, and to determine the three best primary maps based on the matching coefficients of multiple primary maps and each actual scene image material relative to the map requirements.

[0165] The optimal map module 23 is used to determine the corresponding map bias elements based on the detection of the three better primary maps. The optimal map is determined based on multiple map bias elements, the character form of the game character and the character training requirements. The optimal map serves as a stage image of the game map.

[0166] The map merging mode module 24 is used to mark the stage positions of multiple best maps when the game character needs to traverse the overall map, and to determine the map merging mode of multiple best maps based on the multiple best maps, the corresponding stage positions and the game character's play objectives at each stage.

[0167] The map update module 25 is used to determine the corresponding overall map based on the map synthesis mode, the stage position and position order of multiple best maps, identify multiple abnormal map areas based on the detection of the overall map, and determine the corresponding optimization areas based on the multiple abnormal map areas and the virtual simulation of the game character, so as to automatically update the overall map.

[0168] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A method for generating a game map, characterized in that, include: The initial map combination was determined based on the game's map requirements and multiple real-world scene image materials; Multiple primary maps are determined based on the combination of primary maps. The three best primary maps are then determined based on the matching coefficients of the multiple primary maps and the image materials of each actual scene relative to the map requirements. The matching coefficients are calculated based on the matching degree evaluation of multiple dimensions, including terrain matching degree, style matching degree, element matching degree, and function matching degree. Based on the detection of the three best primary maps, the corresponding map weight elements are determined. Based on multiple map weight elements, the character form of the game character and the character training requirements, the best map is determined. The best map serves as the stage image of the game map. This includes: performing image detection on the three best primary maps and outputting the corresponding map weight areas; determining multiple map features based on the region recognition of each map weight area; and determining the corresponding map weight elements based on the multiple map features and the image content presented by the corresponding primary maps. When a game character needs to traverse the overall map, mark the stage positions of multiple best maps, and determine the map synthesis mode of multiple best maps based on multiple best maps, their corresponding stage positions, and the game character's play objectives at each stage. Based on the map synthesis mode, the stage positions and position order of multiple best maps, the corresponding overall map is determined. Based on the detection of the overall map, multiple abnormal map areas are identified. Based on the multiple abnormal map areas and the virtual simulation of the game characters, the corresponding optimization areas are determined to automatically update the overall map.

2. The method for generating a game map according to claim 1, characterized in that, The process of determining the initial map combination based on the game's map requirements and multiple real-world scene image assets includes: The system collects the game's map requirements, determines multiple sub-map requirements based on the initial analysis of the game's map requirements, determines the corresponding map target based on the cross-composite of multiple sub-map requirements, and triggers the drone to conduct target inspection in the target area based on the map target in order to collect multiple real-world scene image materials. Based on the recognition of multiple real-world scene image materials, the priority of multiple real-world scene image materials is determined, and the multiple real-world scene image materials are filtered in terms of image quality and priority to output multiple candidate real-world scene image materials. According to the matching of multiple candidate real-world scene image materials and multiple sub-map requirements, a corresponding primary map combination is constructed. At this time, the primary map combination records the content that the multiple candidate real-world scene image materials match the game's map requirements.

3. The method for generating a game map according to claim 2, characterized in that, The process involves determining multiple primary maps based on a combination of primary maps, and then selecting three optimal primary maps based on the matching coefficients between these primary maps and the actual scene image materials relative to the map requirements. These three primary maps include: Based on the identification of the primary map combination, the image quality parameters of multiple candidate real-world scene image materials are determined, and the corresponding image selection method is determined based on the image quality parameters of the multiple candidate real-world scene image materials, so as to determine multiple primary maps based on the multiple candidate real-world scene image materials. Collect image materials from various real-world scenarios and game map requirements, and determine the corresponding matching coefficients based on the matching between the image materials from various real-world scenarios and the game map requirements; Multiple image combinations are determined based on several primary maps and their corresponding matching coefficients; the optimal image combination is then selected based on the selection of these multiple image combinations, resulting in the output of three optimal primary maps.

4. The method for generating a game map according to claim 1, characterized in that, The process of determining corresponding map bias elements based on the detection of three better primary maps, determining the optimal map based on multiple map bias elements, the character form of the game character, and the character training requirements, with the optimal map serving as a stage image of the game map, also includes: Collect basic information about the game characters, and determine the character form and training requirements based on the identification of the basic information of the game characters. Determine the first image control content based on multiple map bias elements and the character form of the game characters, and determine the second image control content based on multiple map bias elements and character training requirements. The optimal map is determined by synthesizing the first image control content, the second image control content, and three better primary maps. The optimal map is presented in the corresponding stage as the corresponding stage image.

5. The method for generating a game map according to claim 1, characterized in that, When a game character needs to traverse the overall map, the stage positions of multiple optimal maps are marked. A map merging mode is determined based on these optimal maps, their corresponding stage positions, and the game character's objectives at each stage. This includes: The game map matches the corresponding best map at different stages, monitors the position of the game character on the game map in real time, and triggers the overall construction of the game map based on the traversal of the game map by the game character. At this time, the stage position of multiple best maps is determined based on the detection of the best map.

6. The method for generating a game map according to claim 5, characterized in that, The method of marking the stage positions of multiple optimal maps when the game character needs to traverse the overall map, and determining the map synthesis mode of multiple optimal maps based on multiple optimal maps, corresponding stage positions, and the game character's play objectives at each stage, also includes: Collect image content from multiple optimal maps, determine the first synthesis coefficient based on the image content of multiple optimal maps and the corresponding stage position, and determine the second synthesis coefficient based on the stage position of multiple optimal maps and the game character's play objectives at each stage. The map synthesis mode of multiple optimal maps is determined based on the first synthesis coefficient, the second synthesis coefficient, and the mapping relationship of map synthesis mode.

7. The method for generating a game map according to claim 1, characterized in that, The process involves determining a corresponding overall map based on the map synthesis mode, the stage positions and position order of multiple optimal maps, identifying multiple map anomaly areas based on the detection of this overall map, and determining corresponding optimization areas based on the multiple map anomaly areas and the virtual simulation of the game character, in order to autonomously update the overall map, including: The map compositing pattern is collected, and the map compositing information is determined based on the parsing of the map compositing pattern. Multiple sub-map compositing projects are determined based on the identification of the map compositing information. Based on the multiple sub-map compositing projects, the compositing of multiple optimal maps in different dimensions is triggered. At this time, the stage positions and position orders of the multiple optimal maps are collected, and the corresponding overall map is determined based on the compositing of the stage positions and position orders of the multiple optimal maps in different dimensions.

8. The method for generating a game map according to claim 7, characterized in that, The process of determining the corresponding overall map based on the map synthesis mode, the stage positions and position order of multiple optimal maps, identifying multiple map anomaly areas based on the detection of the overall map, and determining corresponding optimization areas based on the multiple map anomaly areas and the virtual simulation of the game character, in order to autonomously update the overall map, also includes: Anomaly detection is performed on the overall map. Based on the anomaly detection, multiple abnormal map features are identified. Based on the shape, location, and regional distribution of the multiple abnormal map features, multiple abnormal map regions are identified. Collect multiple abnormal map areas and game characters, and conduct virtual drills of the game characters over these abnormal map areas to present the virtual effects of the game characters traversing these areas. Based on these virtual effects, deduce the corresponding optimization areas, and trigger the overall map to update automatically based on the replacement of abnormal map areas by the optimization areas.

9. A map generation system for a game, characterized in that, The game map generation system is applied to the game map generation method as described in any one of claims 1-8, and the game map generation system includes: The basic map combination module is used to determine the basic map combination based on the game's map requirements and multiple real-world scene image materials; The primary map module is used to determine multiple primary maps based on a combination of primary maps. It then selects the three best primary maps based on the matching coefficients of the multiple primary maps and the image materials of each actual scene relative to the map requirements. The matching coefficients are calculated based on a multi-dimensional matching degree evaluation, including terrain matching degree, style matching degree, element matching degree, and function matching degree. The optimal map module is used to determine the corresponding map weighting elements based on the detection of the three best primary maps. The optimal map is determined based on multiple map weighting elements, the character form of the game character, and the character training requirements. The optimal map serves as a stage image of the game map, including: image detection of the three best primary maps and outputting the corresponding map weighting regions; determining multiple map features based on the region recognition of each map weighting region; and determining the corresponding map weighting elements based on the multiple map features and the image content presented by the corresponding primary maps. The map merging mode module is used to mark the stage positions of multiple best maps when the game character needs to traverse the overall map. Based on the multiple best maps, the corresponding stage positions, and the game character's play objectives at each stage, the map merging mode of multiple best maps is determined. The map update module is used to determine the corresponding overall map based on the map synthesis mode, the stage position and position order of multiple best maps, identify multiple abnormal map areas based on the detection of the overall map, and determine the corresponding optimization areas based on the multiple abnormal map areas and the virtual simulation of the game characters, so as to automatically update the overall map.

Images