AI-based virtual clothing interaction method, system and storage medium

Abstract

The present application relates to an AI-based virtual clothing interaction method, system and storage medium, the method comprising the following steps: delineating a number of warps and a number of wefts based on a virtual clothing model, and delineating the fabric of the virtual clothing model into a number of cloth pieces based on the warps and wefts; setting the durability attributes of the cloth pieces and setting the initial durability and minimum durability; determining the attack mode and its corresponding durability damage value; recording the attack received by each piece of cloth, and calculating the cumulative value of the durability damage value received by each piece of cloth; inputting the attack sequence received by each piece of cloth, the cumulative value of the durability damage of various attack types and the initial durability of the corresponding piece of cloth into a pre-trained model, and obtaining the damage effect; selecting the corresponding damage effect map, and replacing it with the corresponding piece of cloth of the virtual clothing. The present application has the advantage of realizing the dynamic damage effect of clothing without relying on high computing power.

Description

Technical Field

[0001] The present application relates to the field of computer graphics, and in particular to an AI-based virtual clothing interaction method, system and storage medium. Background Art

[0002] At present, the development of clothing effects in the gaming industry is becoming more and more mature, especially in the interaction and dynamic performance of virtual clothing. In the game, clothing interaction between different players has become an important factor in improving the gaming experience. For example, in many combat games, the change in the appearance of clothing after being hit is a key visual effect. In order to achieve this effect, cloth simulation engines such as Unreal Engine 5 are usually used.

[0003] Fabric simulation engines can simulate the real reactions of clothing to various blows, including tearing, scorching, and explosion damage. This simulation technology is suitable for large-scale games because it can provide highly realistic visual effects and produce real-time changes based on the effects of the blows. However, for small-scale games, the application of fabric simulation engines faces many challenges. First of all, this type of engine has very high requirements for computing resources. It is very computationally intensive to implement the calculation of damage effects and simulations, which may cause stuttering when the game is running, and the requirements for hardware equipment are also very high. Secondly, due to the complexity of fabric simulation engines, small-scale games face high technical barriers when developing and applying these engines.

[0004] In addition, the cloth simulation engine also has many inconveniences during the game update process. If the game needs to update the cloth material or introduce a new weapon system, such as new guns and ammunition, the cloth simulation effect needs to be retrained and debugged. This process is cumbersome and time-consuming, and is not suitable for the needs of frequently updated games. Therefore, for small-scale games, it is particularly important to find a solution that can achieve dynamic clothing damage effects without relying on high computing power. Summary of the invention

[0005] In order to achieve dynamic clothing damage effects without relying on high computing power, the present application provides an AI-based virtual clothing interaction method, system and storage medium.

[0006] In the first aspect, the present application provides an AI-based virtual clothing interaction method, which adopts the following technical solutions:

[0007] A virtual clothing interaction method based on AI, comprising the following steps:

[0008] S1. Defining a plurality of warps and a plurality of wefts based on the virtual clothing model, and demarcating the fabric of the virtual clothing model into a plurality of cloth blocks based on the warps and wefts, wherein the cloth blocks are divided into sensitive cloth blocks corresponding to sensitive areas of the torso and non-sensitive cloth blocks corresponding to non-sensitive areas of the torso;

[0009] S2. Set the durability attribute of the cloth and set the initial durability and the minimum durability. The minimum durability of the sensitive cloth is greater than zero, and the minimum durability of the non-sensitive cloth is zero. When the durability of the cloth is lower than the initial durability, there is a corresponding damaged effect. When the durability of the cloth is zero, there is a corresponding damaged effect.

[0010] S3. Determine the attack mode and its corresponding durability damage value, wherein the attack mode includes bullet damage, burning damage, and explosion damage, and the durability damage value reflects the degree of damage to the fabric. Sensitive fabrics and non-sensitive fabrics will show different damage effects after being hit differently;

[0011] S4. record the blows received by each piece of cloth, and calculate the cumulative value of the durability damage value received by each piece of cloth;

[0012] S5. Input the order of attacks on each cloth piece, the cumulative value of durability damage of various attack types and the initial durability of the corresponding cloth piece into the pre-trained model, and obtain the damage effect; wherein the pre-trained model is trained based on the damage effect of the actual cloth piece under different attack orders and degrees;

[0013] S6. Select the corresponding damaged effect map and replace it on the corresponding piece of cloth of the virtual clothing.

[0014] Optionally, the damaged effect map is designed with a plurality of universal connection anchor points corresponding to the warp and weft lines, and the universal connection anchor points are edge areas for providing connections for adjacent damaged effect maps and / or normal cloth maps.

[0015] Optionally, the following steps are further included between S5 and S6:

[0016] A secondary adjustment is made to the combined damage effect of sensitive and non-sensitive areas.

[0017] Optionally, the secondary adjustment of the combined damage effect of the sensitive area and the non-sensitive area includes the following steps:

[0018] Traverse the cloth blocks and obtain the durability of the surrounding cloth blocks of each cloth block;

[0019] If the durability of the surrounding cloth pieces is lower than the preset adjustment threshold, the cloth piece will be directly regarded as a damaged cloth piece; wherein, the damaged cloth piece corresponds to the damage effect;

[0020] If there are sensitive cloth blocks around the damaged cloth block, the joint damaged effect map is called and the joint damaged effect map is spliced ​​onto the sensitive cloth block map based on the position of the sensitive cloth block;

[0021] Based on the current durability of the sensitive cloth block, the sensitive cloth block damage effect map is called to replace the map corresponding to the damage effect calculated by the pre-trained model, wherein the damage effect of the sensitive cloth block is at most a local damage effect.

[0022] Optionally, the training steps of the pre-trained model include:

[0023] S51. Collecting raw data, wherein the data set includes a plurality of different types and different orders of attacks on the actual cloth pieces, and the damage effects caused by the attacks of the various types in the order;

[0024] S52. Calculate the durability damage value suffered by each actual cloth block, and add up the durability damage values ​​of different attack types to obtain the total durability damage value suffered by each cloth block;

[0025] S53. organize the initial durability, attack type and order, and durability damage accumulation value of each real cloth piece into a structured input data format, and use the resulting damage effect as output to form a data set;

[0026] S54. A deep learning neural network model is selected as a machine learning model, and a model structure is constructed, wherein the model structure includes an input layer, a hidden layer, and an output layer corresponding to the input data format, the output layer uses a Softmax activation function to convert the output result into a probability distribution, the optimizer uses adam, categorical_crossentropy is selected as the loss function, and accuracy is used as the evaluation indicator;

[0027] S55. Divide the data set into a training set, a validation set, and a test set;

[0028] S56. Use the training set data to batch train the model and optimize the network weights by gradient descent method;

[0029] S57. Improve model performance by adjusting hyperparameters, where hyperparameters include learning rate, batch size, and regularization term;

[0030] S58. Monitor model performance on the validation set and stop training when performance stops improving to prevent overfitting.

[0031] S59. Use validation set data to evaluate model performance and adjust network structure or parameters.

[0032] Optionally, S3 includes the following steps:

[0033] S31. Select a reference cloth piece;

[0034] S32. Collect the number of shots when the reference cloth is broken after the simulated bullets of different selected calibers are randomly shot at the reference cloth for several times, and calculate the average number of shots based on the collected data, and calculate the durability damage value corresponding to the caliber = initial durability / average number of shots;

[0035] S33. Collect the time required for the flame of the selected temperature to cause the reference cloth to burn at different preset distances, and calculate the average time required for burning based on the collected data, and calculate the durability damage value corresponding to the distance = initial durability / average time required for burning;

[0036] S34. Collect the times required for the selected model of explosive to explode the reference cloth at different preset distances, and calculate the average number of explosions based on the collected data, and calculate the durability damage value corresponding to the distance = initial durability / average number of explosions.

[0037] In the second aspect, the present application provides an AI-based virtual clothing interactive system, which adopts the following technical solutions:

[0038] An AI-based virtual clothing interactive system, comprising:

[0039] A cloth block division module is used to define a plurality of warps and a plurality of wefts based on the virtual clothing model, and divide the cloth of the virtual clothing model into a plurality of cloth blocks based on the warps and wefts, wherein the cloth blocks are divided into sensitive cloth blocks corresponding to sensitive areas of the torso and non-sensitive cloth blocks corresponding to non-sensitive areas of the torso;

[0040] Durability setting module, used to set the durability attribute of the cloth and set the initial durability and minimum durability. The minimum durability of sensitive cloth is greater than zero, and the minimum durability of non-sensitive cloth is zero. When the durability of the cloth is lower than the initial durability, there is a corresponding damage effect, and when the durability of the cloth is zero, there is a corresponding damage effect;

[0041] The attack damage setting module is used to determine the attack mode and its corresponding durability damage value, wherein the attack mode includes bullet damage, burning damage, and explosion damage. The durability damage value reflects the degree of damage to the fabric. Sensitive fabrics and non-sensitive fabrics will show different damage effects after being hit differently.

[0042] A strike recording and calculation module is used to record the strikes received by each piece of cloth and calculate the cumulative value of the durability damage received by each piece of cloth;

[0043] A damage effect determination module is used to input the order of attacks on each piece of cloth, the cumulative value of durability damage of various attack types and the initial durability of the corresponding piece of cloth into the pre-trained model, and obtain the damage effect; wherein the pre-trained model is trained based on the damage effects of actual pieces of cloth under different attack orders and degrees;

[0044] The texture usage module is used to select the corresponding damaged effect texture and replace it on the corresponding cloth piece of the virtual clothing.

[0045] In a third aspect, the present application provides a computer device, which adopts the following technical solution:

[0046] A computer device comprising:

[0047] one or more processors;

[0048] Memory;

[0049] One or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs are configured to: execute the virtual clothing interaction method described above.

[0050] In a fourth aspect, the present application provides a computer-readable storage medium, which adopts the following technical solution:

[0051] A computer-readable storage medium stores a computer program that can be loaded by a processor and execute the above method.

[0052] The storage medium stores at least one instruction, at least one program, code set or instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by the processor to implement: the virtual clothing interaction method as described above.

[0053] In summary, the present application includes at least one of the following beneficial technical effects:

[0054] 1. Through the combination of experimental data and pre-trained models, the generated damage effect is more realistic, which can accurately reflect the damage of different materials and impact methods, and significantly improve the visual realism of virtual clothing after impact.

[0055] 2. Compared with traditional cloth simulation engines, this solution greatly reduces the demand for computing resources and improves the performance and smoothness of small-scale games through pre-trained models and reasonable damage effect generation strategies.

[0056] 3. Specific secondary adjustments are made to sensitive areas and non-sensitive areas to make the damage effect of sensitive areas more reasonable, avoiding the problem of complete damage of sensitive areas being subject to network supervision under realistic conditions, and ensuring the consistency and realism of visual effects.

[0057] 4. The use of preset textures and joint damage effect stitching strategies simplifies the generation process of damage effects, improves development efficiency and controllability of effects, and makes the design and application of damage effects easier. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 A schematic diagram of an application environment of an AI-based virtual clothing interaction method in one embodiment of the present invention is shown.

[0059] Figure 2 A flowchart of an AI-based virtual clothing interaction method according to an embodiment of the present invention is shown.

[0060] Figure 3 A flowchart of sub-step S3 in one embodiment of the present invention is shown.

[0061] Figure 4 A flowchart of sub-step S5 in one embodiment of the present invention is shown.

[0062] Figure 5 A schematic diagram of an AI-based virtual clothing interaction device according to an embodiment of the present invention is shown.

[0063] Figure 6 A schematic diagram of a computer device according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0064] The present application is further described in detail below in conjunction with the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0065] In the following description, for the purpose of explanation, many specific details are set forth in order to provide a thorough understanding of the inventive concept. As part of this specification, some of the drawings in the drawings of the present disclosure represent structures and devices in the form of block diagrams to avoid making the disclosed principles complicated and difficult to understand. For the sake of clarity, not all features of the actual implementation are necessary to be described. In addition, the language used in this disclosure has been selected primarily for readability and instructional purposes, and may not be selected to delineate or limit the subject matter of the present invention, thereby resorting to the necessary claims to determine such inventive subject matter. References to "a specific implementation" or "specific implementations" in this disclosure mean that the specific features, structures or characteristics described in conjunction with the specific implementation are included in at least one specific implementation, and multiple references to "a specific implementation" or "specific implementations" should not be understood as necessarily referring to the same specific implementation.

[0066] Unless expressly limited, the terms "a", "an", and "the" are not intended to refer to a singular entity, but rather to include general categories of which specific examples may be used for illustration. Thus, the use of the terms "a" or "an" may mean any number of at least one, including "one", "one or more", "at least one", and "one or more than one". The term "or" means any of the alternatives and any combination of the alternatives, including all alternatives, unless the alternatives are expressly indicated to be mutually exclusive. The phrase "at least one of" when combined with a list of items refers to a single item in the list or any combination of the items in the list. The phrase does not require all of the listed items unless expressly limited to that.

[0067] The embodiment of the present application discloses an AI-based virtual clothing interaction method. The AI-based virtual clothing interaction method provided by the embodiment of the present invention can be applied in the following aspects: Figure 1 In the application environment, the AI-based virtual clothing interaction method is applied in the AI-based virtual clothing interaction system, and the AI-based virtual clothing interaction system includes a client and a server, wherein the client communicates with the server through a network. The client, also known as the user end, refers to a program corresponding to the server that provides local services to the client. Furthermore, the client is a computer-side program, an APP program for a smart device, or a third-party applet embedded in other APPs. The client can be installed on, but not limited to, various computer devices such as personal computers, laptops, smart phones, tablet computers, and portable wearable devices. The server can be implemented with an independent server or a server cluster consisting of multiple servers.

[0068] Reference Figure 2 The AI-based virtual clothing interaction method includes the following steps S1-S6.

[0069] S1. Based on the virtual clothing model, a number of warps and a number of wefts are defined, and based on the warps and wefts, the fabric of the virtual clothing model is defined into a number of cloth blocks, wherein the cloth blocks are divided into sensitive cloth blocks corresponding to sensitive areas of the torso and non-sensitive cloth blocks corresponding to non-sensitive areas of the torso.

[0070] Specifically, S1 includes the following steps S11-S15.

[0071] S11. Import the model of the virtual clothing and ensure that the model includes sufficient details and resolution for subsequent cloth division.

[0072] S12. Delineate the warp and weft on the virtual clothing model. The delineation of the warp and weft may be based on the structure and design of the clothing, for example, according to the main direction and texture of the clothing. Delineate the warp and weft on the virtual clothing model. The delineation of the warp and weft may be based on the structure and design of the clothing, for example, according to the main direction and texture of the clothing. In some embodiments, the warp and weft are drawn using graphics processing software (such as Blender or Maya) or a programming language (such as Python with a graphics library).

[0073] The warp and weft threads are the connection points between adjacent pieces of cloth. In order to make the sizes of the divided pieces of cloth as close as possible so that the degree of durability reduction when they are hit is closer, the spacing between the warp threads and the spacing between the weft threads are as equidistant as possible.

[0074] S13. Based on the designated warp and weft lines, the virtual clothing model is divided into a number of cloth pieces. The size and shape of the cloth pieces are adjusted according to the design requirements to ensure that the size of each cloth piece after division is reasonable and uniform.

[0075] In some embodiments, the clothing model may be meshed using an automatic segmentation tool or a manual segmentation tool to generate cloth pieces.

[0076] S14. Classify the divided cloth pieces into sensitive cloth pieces and non-sensitive cloth pieces. The sensitive cloth pieces correspond to sensitive areas of the human body, such as the chest, lower abdomen, crotch, etc.

[0077] The purpose of the classification into sensitive and non-sensitive fabrics is to accommodate current audit requirements, where sensitive areas of the torso must not be exposed.

[0078] In some embodiments, software may be used to manually mark or automatically identify sensitive areas, and the cloth pieces corresponding to these areas may be classified as sensitive cloth pieces, and the cloth pieces in other areas may be classified as non-sensitive cloth pieces.

[0079] S15. Mark each cloth piece, record its location, type (sensitive cloth piece or non-sensitive cloth piece) and other attributes (such as initial durability). Store the division and classification information of the cloth pieces in the model file, or save it in the form of an independent data file for use in subsequent steps.

[0080] Specifically, in one embodiment, a 3D model of a virtual T-shirt is imported using 3D modeling software (such as Blender). According to the structure of the T-shirt, warps and wefts are delineated on the model. The warps are divided along the longitudinal direction of the T-shirt, and the wefts are divided along the transverse direction. Using the "Mark Seam" function in Blender, the warps and wefts are delineated along the main direction of the T-shirt to ensure uniform distribution. Based on the delineated warps and wefts, the T-shirt model is divided into several pieces of cloth. For example, the front part is divided into multiple pieces of cloth, including the chest area and the abdomen area. Using the "UV Mapping" tool in Blender, the surface of the T-shirt is divided into several UV islands, each UV island corresponding to a piece of cloth.

[0081] Mark the chest area as sensitive because this area corresponds to the sensitive part of the human body. Mark the sleeves, abdomen, back and other areas as non-sensitive. Mark each piece of cloth and record its location, type (sensitive or non-sensitive), initial durability and other information. Store this information in the model file or export it as a separate data file (such as JSON or XML format) for use in subsequent steps.

[0082] S2. Set the durability attribute of the cloth and set the initial durability and minimum durability, where the minimum durability of the sensitive cloth is greater than zero, and the minimum durability of the non-sensitive cloth is zero. When the durability of the cloth is lower than the initial durability, there is a corresponding damaged effect, and when the durability of the cloth is zero, there is a corresponding damaged effect.

[0083] In different embodiments, the initial durability of different cloth pieces can be set to be the same or different. This is because in reality, due to the structural characteristics of cloth pieces at different locations, the corresponding durability damage values ​​when they are hit by the same blow are different, that is, some cloth pieces are more easily damaged, and some cloth pieces are more durable. Therefore, the initial durability can be set to be different, so that the same blow will cause the same durability damage value to cloth pieces in different areas.

[0084] In addition, the minimum durability of sensitive and non-sensitive cloth blocks is set. For example, the minimum durability of sensitive cloth blocks is set to 10, and the minimum durability of non-sensitive cloth blocks is set to 0. When the durability of non-sensitive cloth blocks reaches 0, they are considered damaged. Damaged cloth blocks can correspond to special maps, or directly correspond to no maps.

[0085] Of course, the initial durability can be considered the same, so even if it is not damaged after a series of blows in reality, the virtual cloth will be directly considered damaged if its durability is lower than zero after such a blow. Although it is different from reality, it is convenient for computers to perform simple calculations.

[0086] For example, using 3D modeling software (such as Blender) or game engines (such as Unity), initialize the durability value for each piece of cloth. In Blender, you can set the durability value by adding custom attributes. In the properties panel of the cloth, add the initial_durability and minimum_durability fields and set the corresponding values. Add a durability field to each piece of cloth in the data structure. For example, use a Python script to set the durability attribute for each piece of cloth in the Blender model:

[0087] importbpy

[0088] #Assume that the cloth objects of the T-shirt model are all in a collection

[0089] tshirt_blocks=bpy.data.collections['TShirtBlocks']

[0090] forblock in tshirt_blocks.objects:

[0091] ifblock.name in sensitive_areas:#If it is a sensitive cloth block

[0092] block['initial_durability']=100

[0093] block['minimum_durability']=30

[0094] else:#Non-sensitive cloth

[0095] block['initial_durability']=70

[0096] block['minimum_durability']=0

[0097] S3. Determine the striking mode and its corresponding durability damage value, wherein the striking mode includes bullet damage, burning damage, and explosion damage. The durability damage value reflects the degree of damage to the fabric. Sensitive fabrics and non-sensitive fabrics will show different damage effects after being struck differently.

[0098] It should be noted that when implementing AI-based virtual clothing interaction methods, different attacks and attack sequences correspond to specific damage effect maps. The damage effect map is designed by programmers for each designated piece of cloth based on the data collected in reality and the corresponding damage conditions (these data come from the data set of the pre-trained model).

[0099] The design of the damage effect map is based on real-life experimental data, which includes the actual damage of fabrics under different types of attacks (such as bullets, burning, explosions) and different attack sequences. For example, according to different attack types and attack sequences, the damage effect is divided into multiple types, such as: slight tear, moderate tear, severe tear, scorch, explosion damage, etc. Each damage effect corresponds to a different map, showing different degrees and types of damage. It should be noted that only limited examples are given here, and should not be limited or misled by the examples. In fact, there can be dozens or hundreds of maps, and the damage effects such as tearing and scorching can coexist and correspond to certain maps.

[0100] Programmers use image editing software (such as Photoshop, GIMP) to design damage effect maps. The map design should be as realistic as possible, reflecting the appearance of real cloth after suffering a corresponding blow. The material, texture and color of the cloth need to be considered during the design to ensure that the map looks natural and harmonious when applied to the virtual clothing.

[0101] In addition, the damage effect map is designed with several universal connection anchor points corresponding to the warp and weft lines. The universal connection anchor points are used to provide edge areas for connecting adjacent damage effect maps and / or normal cloth maps. Universal connection anchor points are designed at the warp and weft lines of the damage effect map to ensure that adjacent damage effect maps and / or normal cloth maps can be smoothly connected. In other words, the design of the anchor points ensures that in the game, the damage effects between cloth pieces can be seamlessly transitioned to avoid visually abrupt edges.

[0102] In the game engine, anchor points are used to dynamically adjust the connection between adjacent pieces of cloth. For example, when two pieces of cloth are hit, anchor points help calculate and render the damage transition effect between them. The use of anchor points ensures the continuity and consistency of the damage effect on the virtual clothing, improving the visual realism of the game.

[0103] For example, we can obtain data about the damage of cotton cloth under bullet impact from the dataset of the pre-trained model, including different cases of slight tear, moderate tear and severe tear.

[0104] Use Photoshop to design three types of damage effect maps, corresponding to slight tearing, moderate tearing and severe tearing respectively.

[0105] Slightly torn map: Draw some tiny cracks and holes on the surface of the cloth, with a slightly darker color.

[0106] Medium Tears Map: Paints larger rips and torn edges on the surface of the cloth, adding some detail of broken fibers.

[0107] Heavy Torn Map: Paints large areas of tearing and missing parts on the surface of the cloth, with darker colors and burn marks on the edges.

[0108] Add universal connection anchor points around the edges of each damage texture. These anchor points can be alignment marks or transparent areas to achieve seamless connection between adjacent textures in the game.

[0109] Import the designed damaged effect map into the game engine (such as Unity) and apply it to the corresponding cloth pieces of the virtual T-shirt.

[0110] Specifically, refer to Figure 3 , S3 includes the following steps S31-S34.

[0111] S31. Select a reference piece of cloth.

[0112] Select a standard piece of fabric as a reference piece, for example, a 10cm x 10cm piece of cotton fabric. Use this reference fabric for all subsequent striking tests to ensure consistency and comparability of the results.

[0113] S32. Collect the number of times the reference cloth is broken after randomly shooting at it several times with simulated bullets of different selected calibers, and calculate the average number of shooting times based on the collected data, and calculate the durability damage value corresponding to the caliber = initial durability / average number of shooting times.

[0114] Select simulated bullets of different calibers, for example, 9mm and 5.56mm bullets, and prepare multiple reference cloth pieces of the same size.

[0115] Use 9mm simulated bullets to randomly shoot multiple reference cloth pieces, and record the number of shots required for each piece of cloth to be completely broken.

[0116] Experiment 1: Shot 15 times, the cloth completely shattered.

[0117] Experiment 2: After 12 shots, the cloth completely shattered.

[0118] Experiment 3: Shot 14 times, the cloth completely shattered.

[0119] Calculate the average number of shots: average = (15 + 12 + 14) / 3 = 13 times.

[0120] Calculate the durability damage value = initial durability / average number of shots.

[0121] Assuming the initial durability is 100, the durability damage value = 100 / 13≈7.69.

[0122] S33. Collect the time required for the flame of the selected temperature to cause the reference cloth block to burn at different preset distances, and calculate the average time required for combustion based on the collected data, and calculate the durability damage value corresponding to the distance = initial durability / average time required for combustion.

[0123] Select flames of different temperatures, for example, 500° C. and 1000° C. Prepare multiple reference cloth pieces of the same size.

[0124] Use a 500°C flame to burn the reference cloth at different preset distances (such as 10cm, 20cm, 30cm), and record the time required for each piece of cloth to burn until it is completely destroyed.

[0125] Experiment 1: The burning time is 30 seconds.

[0126] Experiment 2: Burning time is 28 seconds.

[0127] Experiment 3: Burning time is 32 seconds.

[0128] Calculate the average value of the burning time: average value = (30 + 28 + 32) / 3 = 30 seconds.

[0129] Calculating durability damage

[0130] Durability damage value = initial durability / average time required for burning.

[0131] Assuming the initial durability is 100, the durability damage value = 100 / 30≈3.33.

[0132] S34. Collect the times required for the selected model of explosive to explode the reference cloth at different preset distances, and calculate the average number of explosions based on the collected data, and calculate the durability damage value corresponding to the distance = initial durability / average number of explosions.

[0133] Select different types of explosives and prepare several reference cloth pieces of the same size.

[0134] Use explosive A to perform explosion tests on reference cloth at different preset distances (such as 1m, 2m, and 3m), and record the number of explosions required for each piece of cloth before it is completely exploded.

[0135] Experiment 1: The number of explosions is 3.

[0136] Experiment 2: The number of explosions is 4.

[0137] Experiment 3: The number of explosions is 3.

[0138] Calculate the average number of explosions: average = (3 + 4 + 3) / 3 = 3.33 times.

[0139] Calculating durability damage

[0140] Durability damage value = initial durability / average number of explosions.

[0141] Assuming the initial durability is 100, the durability damage value = 100 / 3.33≈30.03.

[0142] It is important to note that when calculating the durability damage value of explosion damage, it is necessary to define the standard for a piece of cloth to be considered "completely exploded". This standard can be based on the degree of tearing of the cloth and the area of ​​the tearing. The following is a specific description of the explosion standard:

[0143] 1. Number of cracked blocks

[0144] If the cloth is torn into several small pieces after the explosion, whether it is completely exploded can be defined according to the number of the torn small pieces.

[0145] Standard: If the cloth is broken into 3 or more pieces after the explosion, it is considered to be completely broken. And each piece occupies at least 10% of the original area of ​​the cloth.

[0146] 2. Crack area

[0147] The torn area refers to the ratio of the total area of ​​the torn part to the original area of ​​the cloth after the explosion.

[0148] Standard: If the total area of ​​the torn parts of the cloth after the explosion accounts for more than 50% of the original area of ​​the cloth, it is considered to be completely torn.

[0149] For example:

[0150] Experiment 1: The cloth is split into 4 pieces, each with an area of ​​20 square centimeters, 25 square centimeters, 15 square centimeters, and 40 square centimeters, with a total area of ​​100 square centimeters. The total area of ​​the split parts is 60 square centimeters, accounting for 60% of the original area.

[0151] Experiment 2: The cloth is split into three pieces, each with an area of ​​30 square centimeters, 40 square centimeters, and 30 square centimeters, with a total area of ​​100 square centimeters. The total area of ​​the split parts is 50 square centimeters, accounting for 50% of the original area.

[0152] Experiment 3: The cloth is split into two pieces, each with an area of ​​70 square centimeters and 30 square centimeters, with a total area of ​​100 square centimeters. The total area of ​​the split part is 30 square centimeters, accounting for 30% of the original area.

[0153] Conclusion: The results of Experiment 1 and Experiment 2 both meet the standard of complete explosion (number of cracked pieces ≥ 3 or cracked area ≥ 50%). The results of Experiment 3 do not meet the standard of complete explosion (number of cracked pieces < 3 and cracked area < 50%).

[0154] S4. Record the blows received by each piece of cloth, and calculate the cumulative value of the durability damage value received by each piece of cloth.

[0155] In this step, firstly, a data structure is created for each cloth piece to record the attack history and cumulative damage value of each cloth piece. Each time the cloth piece is attacked, the attack type and attack time are recorded. It should be noted that the attack type here is not limited to general attack types such as bullets and burning, but can also be specific values ​​such as specific bullet types, flame burning distances, flame burning temperatures, etc., in order to refine the attack type and correspond to the designed durability damage value.

[0156] Then, according to the durability damage value of each attack, the cumulative durability damage value of the cloth is calculated. In the subsequent steps, the current durability of the cloth and the order of attacks received can be judged to decide whether to update the damage effect.

[0157] S5. Input the order of attacks on each piece of cloth, the cumulative value of durability damage of various attack types and the initial durability of the corresponding piece of cloth into the pre-trained model, and obtain the damage effect; wherein, the pre-trained model is trained based on the damage effect produced by the actual piece of cloth under different attack orders and degrees.

[0158] Specifically, refer to Figure 4 , the training steps of the pre-trained model include S51-S59.

[0159] S51. Collecting original data, wherein the data set includes a plurality of different types and different orders of attacks suffered by real cloth pieces, and damage effects produced after being hit by the said types of attacks in the order.

[0160] In this step, different types of fabrics are selected, such as cotton, nylon, and denim. The initial size of each fabric sample is 10cm x 10cm.

[0161] Then determine the strike type. Here you can choose from three types: bullet strike, burning strike and explosion strike.

[0162] Then set the experimental parameters, and then conduct multiple experiments on each type of fabric, and record the damage under different attack types and sequences. For example, record the damage effect of the fabric after a single attack, multiple attacks of the same type, and different types of combined attacks. For each experiment, record the initial durability of the fabric, the type and sequence of each attack, the durability damage value of each attack, and the final damage effect.

[0163] For example,

[0164] Cotton cloth was selected as the experimental material, with an initial size of 10cm x 10cm and an initial durability of 100.

[0165] Select the following attack types: 9mm bullet attack, 500℃ flame burning, small grenade explosion.

[0166] Then conduct multiple experiments and record the damage under different types and sequences of blows.

[0167] Experiment 1: Single strike type

[0168] Initial durability: 100

[0169] Strike type: 9mm bullet

[0170] Number of hits: 15

[0171] Durability damage per hit: 7.69

[0172] Cumulative durability damage value: 15*7.69=115.35

[0173] Damage effect: Severe damage

[0174] Experiment 2: Combinations of different strike types

[0175] Initial durability: 100

[0176] Type and order of attacks: 9mm bullet (7.69), 500°C flame (3.33), small grenade (30.03)

[0177] Number of hits: 3 bullets, 2 flames, 1 grenade

[0178] Cumulative durability damage value: 3*7.69+2*3.33+1*30.03=23.07+6.66+30.03=59.76

[0179] Damage effect: Moderate damage effect type C

[0180] Experiment 3: Different striking orders

[0181] Initial durability: 100

[0182] Type and order of attacks: 500°C flame (3.33), 9mm bullet (7.69), small grenade (30.03)

[0183] Number of hits: 2 fire, 3 bullets, 1 grenade

[0184] Cumulative durability damage value: 2*3.33+3*7.69+1*30.03=6.66+23.07+30.03=59.76

[0185] Damage effect: Moderate damage effect M type

[0186] Through the above specific examples, it is demonstrated how to collect raw data, including the damage of fabrics under different attack types and sequences. It should be noted that the moderate damage effect corresponds to a cumulative durability damage value of 59.76, but due to the different attack sequences, the corresponding moderate damage effect types are also different, namely C type and M type, corresponding to different damage effect maps. In addition, here you can collect data for only one type of fabric, or you can collect data for multiple types of fabrics. It is ultimately reflected in the difference in the initial durability of the fabric.

[0187] S52. Calculate the durability damage value received by each actual cloth block, and add up the durability damage values ​​of different attack types to obtain the total durability damage value received by each cloth block.

[0188] In this step, the durability damage value of each cloth piece under different attack types is recorded. For each cloth piece, the durability damage values ​​of different attack types are accumulated to calculate the total durability damage value. The calculated total durability damage value and detailed information of each attack are organized into a structured data format.

[0189] S53. Organize the initial durability, attack type and order, and durability damage accumulation value of each real cloth piece into a structured input data format, and use the resulting damage effect as output to form a data set.

[0190] First, define a data structure for each piece of cloth that includes initial durability, attack type, attack order, cumulative durability damage value, and damage effect. Then, based on the collected data, organize the detailed information of each piece of cloth, including initial durability, attack type and order, cumulative durability damage value, and damage effect. It should be noted that each damage effect corresponds to a map. For the same cumulative durability loss value, there may be different damage effects and corresponding maps.

[0191] For example, for facial data, the data structure can be set as follows:

[0192] structured_data=[

[0193] {

[0194] 'material':'cotton',

[0195] 'initial_durability':100,

[0196] 'damage_sequence':[

[0197] {'type':'bullet','damage':7.69,'count':3},

[0198] {'type':'fire','damage':3.33,'count':2},

[0199] {'type':'explosion','damage':30.03,'count':1}

[0200] ],

[0201] 'cumulative_damage':59.76,

[0202] 'remaining_durability':40.24,

[0203] 'damage_effects':[

[0204] {'effect':'moderate','texture':'moderate_damage_texture_1.png'},

[0205] {'effect':'moderate','texture':'moderate_damage_texture_2.png'} ]

[0207] }.

[0208] S54. Select a deep learning neural network model as the machine learning model and construct a model structure, where the model structure includes an input layer, a hidden layer, and an output layer corresponding to the input data format. The output layer uses a Softmax activation function to convert the output result into a probability distribution. The optimizer uses adam, selects categorical_crossentropy as the loss function, and uses accuracy as the evaluation indicator.

[0209] For example, choose a deep learning framework such as TensorFlow or PyTorch. Then use the activation function, ReLU for the hidden layer and Softmax for the output layer.

[0210] S55. Divide the dataset into training set, validation set and test set.

[0211] S56. Use the training set data to batch train the model and optimize the network weights by gradient descent.

[0212] S57. Improve model performance by adjusting hyperparameters, where hyperparameters include learning rate, batch size, and regularization term.

[0213] S58. Monitor model performance on the validation set and stop training when performance no longer improves to prevent overfitting.

[0214] S59. Use validation set data to evaluate model performance and adjust network structure or parameters.

[0215] Optionally, the following steps are further included between S5 and S6:

[0216] A secondary adjustment is made to the combined damage effect of sensitive and non-sensitive areas.

[0217] Specifically, the step of performing secondary adjustment on the combined damage effect of the sensitive area and the non-sensitive area includes the following steps ad.

[0218] a. Traverse the cloth blocks and obtain the durability of the surrounding cloth blocks of each cloth block.

[0219] In this step, the system will traverse each piece of fabric in the virtual clothing and record the durability of the adjacent pieces of fabric around each piece of fabric. This is to evaluate the damage of each piece of fabric and the impact of its surrounding environment on its durability.

[0220] For example, suppose there is a virtual T-shirt made up of multiple pieces of cloth. The system will check each piece of cloth one by one, for example, check the piece of cloth located in the center of the chest, and record the durability values ​​of the four adjacent pieces above, below, left and right.

[0221] b. If the durability of the surrounding cloth pieces is lower than the preset adjustment threshold, the cloth piece will be directly regarded as a damaged cloth piece; wherein, the damaged cloth piece corresponds to the damaged effect.

[0222] If the durability of all adjacent pieces of a piece of cloth is lower than a preset threshold (for example, 20), the piece of cloth is marked as a "damaged piece of cloth". A damaged piece of cloth means that the connection between the piece of cloth and the four surrounding pieces of cloth is seriously affected. Even if the piece of cloth is intact, it will fall off, so it presents a completely damaged effect.

[0223] c. If there are sensitive cloth blocks around the damaged cloth block, call the joint damaged effect map and splice the joint damaged effect map onto the sensitive cloth block map based on the position of the sensitive cloth block.

[0224] If there are sensitive pieces around a damaged piece (non-damaged state), a specific joint damage effect map is called. The joint damage effect map is designed to show the visual effect of multiple pieces being damaged at the same time. Since sensitive pieces will not be completely damaged, the maps of adjacent non-sensitive pieces will be spliced ​​based on the position of the sensitive pieces to ensure a natural transition of the visual effect.

[0225] d. Based on the current durability of the sensitive cloth block, call the sensitive cloth block damage effect map to replace the map corresponding to the damage effect calculated by the pre-trained model, where the damage effect of the sensitive cloth block is at most a local damage effect.

[0226] S6. Select the corresponding damaged effect map and replace it on the corresponding piece of cloth of the virtual clothing.

[0227] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0228] In one embodiment, a virtual clothing interaction system based on AI is provided, and the virtual clothing interaction system based on AI corresponds to the virtual clothing interaction method based on AI in the above embodiment. Figure 5 As shown, the functional modules of the AI-based virtual clothing interaction system are described in detail as follows:

[0229] A cloth block division module is used to define a plurality of warps and a plurality of wefts based on the virtual clothing model, and divide the cloth of the virtual clothing model into a plurality of cloth blocks based on the warps and wefts, wherein the cloth blocks are divided into sensitive cloth blocks corresponding to sensitive areas of the torso and non-sensitive cloth blocks corresponding to non-sensitive areas of the torso;

[0230] Durability setting module, used to set the durability attribute of the cloth and set the initial durability and minimum durability. The minimum durability of sensitive cloth is greater than zero, and the minimum durability of non-sensitive cloth is zero. When the durability of the cloth is lower than the initial durability, there is a corresponding damage effect, and when the durability of the cloth is zero, there is a corresponding damage effect;

[0231] The attack damage setting module is used to determine the attack mode and its corresponding durability damage value, wherein the attack mode includes bullet damage, burning damage, and explosion damage. The durability damage value reflects the degree of damage to the fabric. Sensitive fabrics and non-sensitive fabrics will show different damage effects after being hit differently.

[0232] A strike recording and calculation module is used to record the strikes received by each piece of cloth and calculate the cumulative value of the durability damage received by each piece of cloth;

[0233] A damage effect determination module is used to input the order of attacks on each piece of cloth, the cumulative value of durability damage of various attack types and the initial durability of the corresponding piece of cloth into the pre-trained model, and obtain the damage effect; wherein the pre-trained model is trained based on the damage effects of actual pieces of cloth under different attack orders and degrees;

[0234] The texture usage module is used to select the corresponding damaged effect texture and replace it on the corresponding cloth piece of the virtual clothing.

[0235] For the specific definition of the AI-based virtual clothing interaction device, please refer to the definition of the AI-based virtual clothing interaction method above, which will not be repeated here. Each module in the above-mentioned AI-based virtual clothing interaction device can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0236] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 6 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used for data related to the AI-based virtual clothing interaction method. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a virtual clothing interaction method based on AI is implemented.

[0237] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the AI-based virtual clothing interaction method of the above embodiment is implemented. Alternatively, when the processor executes the computer program, the functions of each module / unit of the AI-based virtual clothing interaction device of the above embodiment are implemented. To avoid repetition, it will not be described here.

[0238] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the method for interacting with virtual clothing based on AI in the above embodiment is implemented. Alternatively, when the computer program is executed by a processor, the functions of each module / unit in the device for interacting with virtual clothing based on AI in the above device embodiment are implemented. To avoid repetition, it will not be described here.

[0239] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments of the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0240] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0241] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A virtual clothing interaction method based on AI, characterized in that: The following steps are involved: S1. Defining a plurality of warps and a plurality of wefts based on the virtual clothing model, and demarcating the fabric of the virtual clothing model into a plurality of cloth blocks based on the warps and wefts, wherein the cloth blocks are divided into sensitive cloth blocks corresponding to sensitive areas of the torso and non-sensitive cloth blocks corresponding to non-sensitive areas of the torso; S2. Set the durability attribute of the cloth and set the initial durability and the minimum durability. The minimum durability of the sensitive cloth is greater than zero, and the minimum durability of the non-sensitive cloth is zero. When the durability of the cloth is lower than the initial durability, there is a corresponding damaged effect. When the durability of the cloth is zero, there is a corresponding damaged effect. S3. Determine the attack mode and its corresponding durability damage value, wherein the attack mode includes bullet damage, burning damage, and explosion damage, and the durability damage value reflects the degree of damage to the fabric. Sensitive fabrics and non-sensitive fabrics will show different damage effects after being hit differently; S4. record the blows received by each piece of cloth, and calculate the cumulative value of the durability damage value received by each piece of cloth; S5. Input the order of attacks on each cloth piece, the cumulative value of durability damage of various attack types and the initial durability of the corresponding cloth piece into the pre-trained model, and obtain the damage effect; wherein the pre-trained model is trained based on the damage effect of the actual cloth piece under different attack orders and degrees; S6. Select the corresponding damaged effect map and replace it with the corresponding piece of cloth of the virtual clothing; The training steps of the pre-trained model include: S51. Collecting raw data, wherein the data set includes a plurality of different types and different orders of attacks on the actual cloth pieces, and the damage effects caused by the attacks of the various types in the order; S52. Calculate the durability damage value suffered by each actual cloth block, and add up the durability damage values ​​of different attack types to obtain the total durability damage value suffered by each cloth block; S53. organize the initial durability, attack type and order, and durability damage accumulation value of each real cloth piece into a structured input data format, and use the resulting damage effect as output to form a data set; S54. A deep learning neural network model is selected as a machine learning model, and a model structure is constructed, wherein the model structure includes an input layer, a hidden layer, and an output layer corresponding to the input data format, the output layer uses a Softmax activation function to convert the output result into a probability distribution, the optimizer uses adam, categorical_crossentropy is selected as the loss function, and accuracy is used as the evaluation indicator; S55. Divide the data set into a training set, a validation set, and a test set; S56. Use the training set data to batch train the model and optimize the network weights by gradient descent method; S57. Improve model performance by adjusting hyperparameters, where hyperparameters include learning rate, batch size, and regularization term; S58. Monitor model performance on the validation set and stop training when performance stops improving to prevent overfitting. S59. Use validation set data to evaluate model performance and adjust network structure or parameters.

2. The AI-based virtual clothing interaction method according to claim 1, characterized in that: The damaged effect map is designed with a number of universal connection anchor points corresponding to the warp and weft lines, and the universal connection anchor points are edge areas for providing connections for adjacent damaged effect maps and / or normal cloth maps.

3. The AI-based virtual clothing interaction method according to claim 1, characterized in that: The steps between S5 and S6 also include: A secondary adjustment is made to the combined damage effect of sensitive and non-sensitive areas.

4. The AI-based virtual clothing interaction method according to claim 3, characterized in that: The secondary adjustment of the combined damage effect of the sensitive area and the non-sensitive area includes the following steps: Traverse the cloth blocks and obtain the durability of the surrounding cloth blocks of each cloth block; If the durability of the surrounding cloth pieces is lower than the preset adjustment threshold, the cloth piece will be directly regarded as a damaged cloth piece; wherein, the damaged cloth piece corresponds to the damage effect; If there are sensitive cloth blocks around the damaged cloth block, the joint damaged effect map is called and the joint damaged effect map is spliced ​​onto the sensitive cloth block map based on the position of the sensitive cloth block; Based on the current durability of the sensitive cloth block, the sensitive cloth block damage effect map is called to replace the map corresponding to the damage effect calculated by the pre-trained model, wherein the damage effect of the sensitive cloth block is at most a local damage effect.

5. The AI-based virtual clothing interaction method according to claim 1, characterized in that: The S3 comprises the following steps: S31. Select a reference cloth piece; S32. Collect the number of shots when the reference cloth is broken after the simulated bullets of different selected calibers are randomly shot at the reference cloth for several times, and calculate the average number of shots based on the collected data, and calculate the durability damage value corresponding to the caliber = initial durability / average number of shots; S33. Collect the time required for the flame of the selected temperature to cause the reference cloth to burn at different preset distances, and calculate the average time required for burning based on the collected data, and calculate the durability damage value corresponding to the distance = initial durability / average time required for burning; S34. Collect the times required for the selected model of explosive to explode the reference cloth at different preset distances, and calculate the average number of explosions based on the collected data, and calculate the durability damage value corresponding to the distance = initial durability / average number of explosions.

6. A virtual clothing interactive system, characterized in that: The AI-based virtual clothing interaction method according to any one of claims 1 to 5 comprises: A cloth block division module is used to define a plurality of warps and a plurality of wefts based on the virtual clothing model, and divide the cloth of the virtual clothing model into a plurality of cloth blocks based on the warps and wefts, wherein the cloth blocks are divided into sensitive cloth blocks corresponding to sensitive areas of the torso and non-sensitive cloth blocks corresponding to non-sensitive areas of the torso; Durability setting module, used to set the durability attribute of the cloth and set the initial durability and minimum durability. The minimum durability of sensitive cloth is greater than zero, and the minimum durability of non-sensitive cloth is zero. When the durability of the cloth is lower than the initial durability, there is a corresponding damage effect, and when the durability of the cloth is zero, there is a corresponding damage effect; The attack damage setting module is used to determine the attack mode and its corresponding durability damage value, wherein the attack mode includes bullet damage, burning damage, and explosion damage. The durability damage value reflects the degree of damage to the fabric. Sensitive fabrics and non-sensitive fabrics will show different damage effects after being hit differently. A strike recording and calculation module is used to record the strikes received by each piece of cloth and calculate the cumulative value of the durability damage received by each piece of cloth; A damage effect determination module is used to input the order of attacks on each piece of cloth, the cumulative value of durability damage of various attack types and the initial durability of the corresponding piece of cloth into the pre-trained model, and obtain the damage effect; wherein the pre-trained model is trained based on the damage effects of actual pieces of cloth under different attack orders and degrees; The texture usage module is used to select the corresponding damaged effect texture and replace it on the corresponding cloth piece of the virtual clothing.

7. A computer device, characterized in that: It includes: one or more processors; Memory; One or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs are configured to: execute the virtual clothing interaction method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that: The storage medium stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement: the virtual clothing interaction method according to any one of claims 1 to 5.

Images