Fitting method for fitting wearing article model and human body model

Abstract

The invention discloses a fitting method for fitting a wearable article model and a human body model. The fitting method comprises the following steps: acquiring three-dimensional data of a user; generating a three-dimensional human body model; generating a wearing article model; fitting the human body model with the wearing article model, executing initial space matching in static fitting to enable the virtual wearing article to be worn on the three-dimensional human body model of the user and keep a preset region distance, and loading cloth physical parameters and environmental force field parameters in physical driving fitting; and trying on the virtual wearing article, displaying the virtual wearing article, the three-dimensional human body model of the user and the trying-on effect by the platform, selecting the virtual wearing article in the platform by the user, and wearing the virtual wearing article on the selected three-dimensional human body model of the user to complete trying-on. The time for the user to purchase the wearing articles is shortened, the suitability of the user to select the wearing articles is improved, and then the refund and change rate and the return rate of the platform are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of virtual wearable products, and particularly to a trial wearing method for fitting a wearable product model to a human body model. Background Art

[0002] Wearable product stores on online trading platforms provide users with a pre-sale wearable product preview function by using virtual wearable product technology. This function requires the store to measure physical wearable products in advance and upload their electronic data, and also requires users to measure their own three-dimensional data in advance. By generating 3D models from the electronic data of physical wearable products and the three-dimensional data of users, the pre-sale wearing preview effect can be achieved, thereby improving the accuracy of users' selection of online wearable products, ensuring that the sold online wearable products can achieve a suitable effect on users as much as possible, and ultimately reducing the return rate of the store and the time cost of users' selection.

[0003] Among them, the technology of generating and adjusting virtual model 3D models by users uploading three-dimensional data has been relatively mature, such as: Blender, Designdoll, and 3D Max, etc. Similarly, there are also mature software in the field of generating three-dimensional models of wearable products, such as: Marvelous Designer. The preview effect of virtual wearable products on virtual models also depends on the wearing settings and fitting process of virtual wearable products on virtual models. Moreover, it is urgent to solve the problem of allowing users to obtain three-dimensional data more quickly and avoiding direct data leakage on the platform.

[0004] When users use their own virtual models to try on virtual wearable products, it is also urgent to solve how to quickly let users judge whether the wearable products are suitable and how to assist users in matching, so as to improve the user experience. Summary of the Invention

[0005] To solve one or more problems existing in the prior art, the present invention provides a trial wearing method for fitting a wearable product model to a human body model. The technical solution adopted by the present invention to solve the above problems is: obtaining user three-dimensional data, obtaining the three-dimensional data stored in the user account, and the three-dimensional data is obtained by the user through offline measurement;

[0006] Generating a three-dimensional human body model, and the platform uses the three-dimensional data stored in the user account to generate a corresponding user three-dimensional human body model;

[0007] Generate a wearable product model. The platform obtains the wearable product data uploaded by the merchant side. The wearable product data at least includes three-dimensional data of the wearable product, fabric data, material data, and style data. The platform audits the wearable product data and performs corresponding virtual wearable product generation after passing the audit. Compare the generated virtual wearable product with the physical picture of the wearable product uploaded by the merchant side. If the comparison result is qualified, the generated virtual wearable product is qualified. Otherwise, verify the wearable product data and the physical picture of the wearable product, determine whether there is an error, and regenerate it;

[0008] The human body model fits with the wearable product model. In the static fit, perform an initial spatial match so that the virtual wearable product is worn on the user's three-dimensional human body model and maintains a preset regional distance. In the physically driven fit, load the fabric physical parameters and environmental force field parameters;

[0009] Virtual wearable product try-on. The platform displays the virtual wearable product, the user's three-dimensional human body model, and the try-on effect. The user selects the virtual wearable product on the platform and wears it on the selected user's three-dimensional human body model to complete the try-on.

[0010] In some embodiments, it further includes: data comparison. Obtain the three-dimensional data of the wearable product of the virtual wearable product selected by the user and the three-dimensional data of the user's three-dimensional human body model and perform a numerical comparison. Generate a corresponding status label according to the result of the numerical comparison. The status label is used to feedback the fit degree between the virtual wearable product and the user's three-dimensional human body model.

[0011] In some embodiments, the generation of the three-dimensional human body model further includes: topology optimization and standardized pose;

[0012] The topology optimization includes: adjusting the user's three-dimensional human body model to have a uniform quadrilateral topology and adding loops at the joints of the user's three-dimensional human body model;

[0013] The standardized pose includes: the initial pose adopts the A-type pose or the T-type pose, and perform bone binding and verify the rationality of the weights on the user's three-dimensional human body model for animation.

[0014] In some embodiments, the generation of the wearable product model further includes: cutting and splitting design before virtual wearable product generation and topology compatibility;

[0015] The cutting and splitting design includes: splitting the garment pieces according to the real sewing logic and leaving a seam allowance of 1-2 cm at the edges of the garment pieces;

[0016] The topological compatibility includes: adjusting the mesh density of the virtual wearable to be generated to be higher than that of the user's three-dimensional human model, so that the generated virtual wearable meets the set requirements for wrinkle details, and preferentially using quadrilaterals during the process of generating the wearable model.

[0017] In some embodiments, the static fitting in the fitting of the human model and the wearable model further includes: setting the user's three-dimensional human model as a static collider in the physics engine and loading a preset collision thickness, performing layer-by-layer simulation on the multi-layer virtual wearable, and freezing each layer after the simulation is completed and using it as the collider for the next layer.

[0018] In some embodiments, the fitting of the human model and the wearable model further includes: local penetration repair, wrinkle and draping enhancement;

[0019] The local penetration repair includes: performing push-pull adjustment on the penetrated area, adsorbing the vertices of the virtual wearable to the surface of the user's three-dimensional human model through a shrink wrap modifier, locking the first key positions through setting fixed points, where the first key positions include the neckline and cuffs, and simulating the elastic fabric of the virtual wearable through elastic constraints;

[0020] For the wrinkle and draping enhancement, perform subdivision on the virtual wearable through Catmull-Clark, perform displacement mapping on the virtual wearable, bake the high-poly details to the normal map of the virtual wearable, adjust the sewing thread tension of the virtual wearable to slow down or remove tearing, and perform double stitching of the sewing thread at the second key positions of the virtual wearable, where the second key positions include the crotch.

[0021] In some embodiments, the fitting of the human model and the wearable model further includes: cloth latency optimization, and the cloth latency optimization is: enabling interpolation to smooth the cloth movement trajectory and increasing the number of steps of the physics solver to 5 - 8 steps.

[0022] In some embodiments, the fitting of the human model and the wearable model further includes: multi-layer wearable penetration optimization, and the multi-layer wearable penetration optimization is: performing hierarchical setting of collision groups on the virtual wearable, enabling interaction for virtual wearables belonging to different groups, replacing the mesh collider with a distance field, and enabling continuous collision detection and increasing the number of steps of the physics solver according to the penetration amplitude when dealing with cloth penetration caused by high-speed movement.

[0023] In some embodiments, the fitting of the human model and the wearable model further includes: dynamic tearing optimization at the seams, and the dynamic tearing optimization at the seams is: adding elastic properties to the sewing threads at the seams and performing dynamic re-topology on the high-pressure areas of the virtual wearable.

[0024] In some embodiments, it further includes: customizing the beautification of a three-dimensional human model. The user customizes and uploads a real person's head portrait, and the system performs review and facial feature analysis on the real person's head portrait uploaded by the user, and then generates a corresponding personalized cartoon portrait. The personalized cartoon portrait is mapped onto the user's three-dimensional human model after customizing the skin color. If the user is a disabled person, the missing limbs of the user's three-dimensional human model are repaired and filled.

[0025] The technical effect achieved by the present invention is: through the above method, a virtual wearing item try-on preview with a better simulation effect is provided, ensuring a one-to-one correspondence between the three-dimensional human model, the virtual wearing item, the user, and the actual wearing item, making the virtual wearing item fit better with the three-dimensional human model to meet the fit requirements of actual try-on, and using the measuring instrument deployed offline to ensure the reliability and privacy of user data;

[0026] By comparing the above data to give corresponding status tags, thereby facilitating the user to quickly judge whether the virtual wearing item fits. At the same time, based on this status tag and the data of the selected virtual wearing item, the virtual wearing items for the remaining wearing positions are recommended and supplemented, and the recommended virtual wearing items are replaced and adjusted with one key, achieving the purpose of platform promotion and suggesting to the user, and providing reference support for subsequent one-key purchase;

[0027] Ultimately, it reduces the time for the user to select and purchase wearing items and improves the suitability of the wearing items selected by the user, thereby reducing the return and exchange rate of the platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a schematic block diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0029] In order to make the above objects, features, and advantages of the present invention more understandable, the following detailed description of the specific embodiments of the present invention is made in conjunction with the accompanying drawings. Many specific details are set forth in the following description in order to fully understand the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0030] The present invention discloses a try-on method for fitting a wearing item model with a human model, in combination with Figure 1 As shown, it includes: obtaining user three-dimensional data, obtaining the three-dimensional data stored in the user's account, and the three-dimensional data is obtained by the user through offline measurement;

[0031] For example: A user logs in to a measuring instrument deployed offline using a personal account. The measuring instrument guides (through voice or text instructions) the user to complete body measurement and stores the three-dimensional data in the corresponding user account.

[0032] Generate a three-dimensional human body model. The platform uses the three-dimensional data stored in the user account to generate a corresponding three-dimensional human body model of the user.

[0033] Generate a virtual wearable model. The wearable items include: clothing, shoes, hats, watches, jewelry, etc. The platform obtains the wearable item data uploaded by the merchant side. The wearable item data includes at least three-dimensional data of the wearable item, fabric data, material data, and style data. The platform reviews the wearable item data and performs corresponding virtual wearable item generation after the review passes. Compare the generated virtual wearable item with the physical picture of the wearable item uploaded by the merchant side. If the comparison result is qualified, the generated virtual wearable item is qualified; otherwise, verify the wearable item data and the physical picture of the wearable item to determine if there is an error and regenerate it.

[0034] Fit the human body model with the virtual wearable model. In static fitting, perform initial spatial matching so that the virtual wearable item is worn on the three-dimensional human body model of the user and maintains a preset regional spacing (e.g., the chest circumference line spacing is 3 - 5 cm). Use a local coordinate system to align components (e.g., left and right sleeves), and adopt symmetric processing to reduce the workload of unilateral adjustment. In physical drive fitting, load fabric physical parameters and environmental force field parameters.

[0035] Virtual wearable item try-on. The platform displays the virtual wearable item, the three-dimensional human body model of the user, and the try-on effect. The user selects the virtual wearable item on the platform and wears it on the selected three-dimensional human body model of the user to complete the try-on. Among them, the try-on interface provides a "one-click order placement" function, which can settle multiple items from different merchants at the same time (create and pay multiple orders at once).

[0036] Customize the beautification of the three-dimensional human body model. The user customizes and uploads a real person's head portrait. The system reviews and analyzes the facial features of the real person's head portrait uploaded by the user, and then generates a corresponding personalized cartoon head portrait. The personalized cartoon head portrait is mapped onto the three-dimensional human body model of the user after customizing the skin color. If the user is a disabled person, repair and fill in the missing limbs of the three-dimensional human body model of the user. In facial feature analysis, some facial features (usually facial birthmarks and scars) can be changed or removed according to the user's needs, realizing the mapping of the facial features and proportions of the model to those of the real person, and adjusting and removing facial skin color and facial birthmarks. Perform intelligent repair on the model of disabled persons.

[0037] It should be noted that the measuring instrument accurately measures the three-dimensional data of the user through terahertz waves, and then packs and encrypts it for storage in the user's account, meeting the requirements of measuring accurate three-dimensional data and user privacy. Additionally, the user's three-dimensional human model can use the avatar uploaded by the user to improve the simulation effect.

[0038] Specifically, it further includes: data comparison, obtaining the three-dimensional data of the virtual wearable item selected by the user and the three-dimensional data of the user's three-dimensional human model and performing a numerical comparison, generating corresponding status tags according to the result of the numerical comparison, where the status tags are used to feedback the fitness of the virtual wearable item to the user's three-dimensional human model. The status tags can be too large, loose, suitable, fitted, too tight, and specific fitness values. The numerical comparison is performed according to the measurement values of each body part of the user. The fitness of the virtual wearable item for each body part and the overall fitness of the user are calculated through the difference of the numerical comparison, and a prompt is given in the preview interface.

[0039] Filling in, filling in the remaining wearing positions of the user's three-dimensional human model according to the wearing item data and the status tags of the virtual wearable item selected by the user; using the style, color, and fabric data in the wearing item data to load into the recommendation algorithm to select appropriate virtual wearable items for filling in; using the status tags to select the size of the virtual wearable item, so that the automatically filled-in virtual wearable item meets the current selection size of the user. At the same time, when the user changes the size and attributes of the wearable item, a one-key change for automatic filling-in is provided. The platform provides "one-key purchase" support based on the user's selection or the automatically filled-in virtual wearable item.

[0040] Specifically, generating the three-dimensional human model further includes: topology optimization and standardized pose. The topology optimization includes: adjusting the user's three-dimensional human model to have a uniform quadrilateral topology, adding loops (meeting deformation support) at the joints (such as shoulders, knees) of the user's three-dimensional human model, and deleting unnecessary details (such as nails, pores); the standardized pose includes: using the A-type pose or T-type pose as the initial pose to avoid the distortion of the wearable item caused by the initial pose, performing bone binding and verifying the rationality of the weights on the user's three-dimensional human model for animation.

[0041] Specifically, generating the wearable item model further includes: cutting and splitting design before generating the virtual wearable item and topology compatibility.

[0042] The cutting and splitting design includes: splitting the garment pieces according to the real sewing logic (such as the front piece, back piece, sleeve piece) and leaving a seam allowance of 1-2 cm at the edge of the garment pieces. The garment pieces can be directly drawn into 2D patterns or imported into the block model from 3D software.

[0043] The topological compatibility includes: adjusting the mesh density of the virtual wearable to be generated to be higher than that of the user's three-dimensional human model (e.g., the number of faces of the user's three-dimensional human model is 50,000, and the number of faces of the virtual wearable is 80,000 - 100,000), so that the generated virtual wearable meets the set requirements for wrinkle details, and preferentially using quadrilaterals during the process of generating the wearable model.

[0044] Specifically, in the static fitting of the human model and the wearable model, it further includes: setting the user's three-dimensional human model as a static collider in the physics engine and loading a preset collision thickness (2 - 5 mm), performing layer-by-layer simulation on the multi-layer virtual wearable, and freezing and using it as the collider for the next layer after each layer of simulation is completed, so as to reduce the risk of the virtual wearable penetrating;

[0045] In the physically driven fitting, the cloth physical parameters include: stiffness (e.g., the stiffness of suit fabric is set to 500 - 1000 N / m, and that of silk is set to 50 - 100 N / m), ductility (e.g., knitted fabric allows 30% stretching, and denim is limited within 5%), and damping (e.g., set to 0.1 - 0.3 to reduce vibration). The environmental force field parameters are such as: the earth's gravity is 9.8 m / s 2 , and the moon simulation is set to 1.6 m / s 2 , the intensity and direction of the wind force field; in the dynamic simulation software, the solver: for high-precision solving, a time step of 0.001 seconds is adopted, and for real-time applications, it can be relaxed to 0.01 seconds. Self-collision can be selectively enabled to prevent the cloth from self-penetrating (which will increase the performance burden), and GPU acceleration is enabled to improve the face number solving efficiency;

[0046] The fitting of the human model and the wearable model further includes: local penetration repair, wrinkle and drape enhancement; the local penetration repair includes: performing push-pull adjustment on the penetrated area, adsorbing the vertices of the virtual wearable to the surface of the user's three-dimensional human model through a shrink wrap modifier, locking the first key positions through setting fixed points, and the first key positions include the neckline and cuffs, and simulating the elastic fabric (such as elastic band) of the virtual wearable through elastic constraints;

[0047] For the wrinkle and drape enhancement, the virtual wearable is subdivided through the Catmull-Clark subdivision algorithm, displacement mapping is performed on the virtual wearable, the details of the high-poly model are baked into the normal map of the virtual wearable, the sewing thread tension of the virtual wearable is adjusted (the intensity is increased to 80 - 100%) to slow down or remove tearing, and double stitching of the sewing thread is performed on the second key positions of the virtual wearable, and the second key positions include the crotch.

[0048] Specifically, the fitting of the human body model and the wearable product model further includes: cloth delay optimization, and the cloth delay optimization is: enabling interpolation to smooth the cloth movement trajectory, increasing the number of physical solver steps to 5-8 steps. An example of a tool instruction is:

[0049] cpp

[0050] / / Chaos Cloth configuration

[0051] ClothComponent->SetSubSteps(5);

[0052] ClothComponent->SetCollisionThickness(2.0f);

[0053] ClothComponent->EnableContinuousCollisionDetection(true);

[0054] Optimization for mutual penetration of multiple wearable products. The optimization for mutual penetration of multiple wearable products is: performing hierarchical setting of collision groups on virtual wearable products, enabling interaction for virtual wearable products belonging to different groups, and replacing the mesh collider with a distance field to reduce mutual penetration while optimizing the calculation efficiency. An example of a tool (Blender) instruction is: python

[0055] # Set collision groups

[0056] bpy.context.object.modifiers["Cloth"].settings.collision_group = 1bpy.context.object.modifiers["Cloth"].settings.collision_group_outer = 2;

[0057] When dealing with cloth penetration caused by high-speed movement, enable continuous collision detection and increase the number of physical solver steps according to the penetration amplitude;

[0058] Optimization for dynamic tearing at seams. The optimization for dynamic tearing at seams is: adding elastic properties to the sewing threads at the seams and performing dynamic retopology (using "DynaMesh" in ZBrush) on the high-pressure areas of virtual wearable products.

[0059] Specifically, it further includes animation adaptation, and the animation adaptation includes: automatic skinning, inheriting the influence of the human body bones through weight transfer from Mesh to Mesh (such as Copy Skin Weights in Maya), and using a third-party plugin (such as DeltaMush) to smooth the weight transition area;

[0060] Manually correct. In the weight painting mode, strengthen the influence of the shoulders and hips on the wearable, and weaken the influence of non-active areas such as the abdomen;

[0061] Hybrid shapes. Create corrective shapes for extreme poses (such as arms raised high), and smoothly transition through Driven Keys;

[0062] Secondary skeletons. Add Helper Joints to control complex deformations (such as skirt flapping).

[0063] Specifically, it also includes: rendering and performance optimization, and the rendering and performance optimization includes: seam alignment. Optimize the UV layout in RizomUV to ensure that the seams are located in hidden areas (such as side seams), and use UDIM multi-quadrant UV to support high-resolution textures;

[0064] PBR material construction. Paint wear and stain layers through Substance Painter to enhance realism;

[0065] LOD grading. Generate multiple levels of detail (such as 100,000 polygons for high-poly, 30,000 polygons for medium-poly, and 5,000 polygons for low-poly), and dynamically switch in Unity / Unreal according to the camera distance;

[0066] Collision body simplification. Create a low-polygon collision proxy for the human body to improve the efficiency of real-time interaction.

[0067] After completing the dynamic adaptation of the virtual wearable and the human body three-dimensional model, it is necessary to verify the stability and realism of the wearable under complex movements through systematic motion tests;

[0068] Test action design and environment configuration: 1. The benchmark action library includes basic actions: Walk, Run, Jump, Sit, Arm Swing; extreme actions: yoga poses (such as downward dog), Spin, and intense dance moves (such as Breaking); industry standards: use BVH format action data from the CMU motion capture database;

[0069] 2. Environmental variables, physical environment: simulate the influence of different gravity (space, underwater) and wind speed (0 - 15 m / s); interaction objects: add scene collision bodies (such as chairs, walls) to test the interaction between the wearable and the environment (such as the skirt being hooked);

[0070] Motion test execution process, layered testing strategy: 1. Unit testing, the purpose is to verify the local deformation during single-joint movement. The method is to drive a single bone in Maya (such as only bending the elbow) to check whether the sleeve wrinkles are as expected; 2. Integration testing, the purpose is the global stability of multi-joint collaborative movement. The method is to import BVH motion data, play it in real time in Unity, and observe whether there is penetration or tearing in the linkage areas such as the waist and shoulders; 3. Stress testing, the purpose is to test the performance limit under high-frequency / high-intensity movements. The method is to use a Python script to loop the jump-landing action (100 cycles) to monitor the stability of the cloth solution.

[0071] Data collection, deformation quantification: Use vertex distance detection tools (such as Maya's "Delta Mush") to record the distance changes between wearable products and the human body (the threshold is set to >2mm for penetration); performance indicators: record frame rate (FPS), GPU memory usage, and single frame solution time (<30ms is a real-time application standard). Automated script test pipeline, taking Python as an example:

[0072] #Maya batch action test script

[0073] import maya.cmds as cmds

[0074] motions=["walk","run","jump"]

[0075] for motion in motions:

[0076] cmds.bvh_import(f"{motion}.bvh")#Import motion data

[0077] cmds.playbackOptions(minTime=0,maxTime=100)

[0078] cmds.play(forward=True)

[0079] capture_cloth_penetration(f"output_{motion}.csv")

[0080] #Custom penetration detection function

[0081] CI / CD integration: Configure automated testing tasks in Jenkins / GitLab CI and perform regression testing every night; display the penetration rate and performance change trends of each version through visual reports (such as Plotly charts).

[0082] Parameter Tuning: 1. Problem: The fabric is too stiff. Adjustment parameter: Decrease the Bend Stiffness. Expected effect: Increase the natural draping. 2. Problem: High-frequency oscillation. Adjustment parameter: Increase the Damping to 0.5. Expected effect: Suppress the fabric jitter. 3. Problem: Local continuous penetration. Adjustment parameter: Increase the thickness of the collider to 10 mm. Expected effect: Force the separation of the models. 4. Problem: GPU memory overflow. Adjustment parameter: Enable the fabric simulation LOD (Levels 0 - 2). Expected effect: Dynamically reduce the number of faces according to the distance.

[0083] Case: Running test optimization. Problem found: The fabric under the armpit penetrates when swinging the arms quickly. Correction steps: 1. Add a Pin to the armhole in Marvelous Designer. 2. Set a local Sphere Collider for this area in Unity. 3. Temporarily increase the solving accuracy of this area when swinging the arms through an Animation Event. Verification result: The number of penetrating vertices decreased from 87 to 3, and the GPU consumption increased by 8%.

[0084] In summary, through the above methods, a virtual try-on preview of wearable products with better simulation effects is provided, ensuring a one-to-one correspondence between the 3D human model, virtual wearable products, users, and actual wearable products, making the virtual wearable products fit the 3D human model better to meet the requirements of actual try-on fit. The measuring instrument deployed offline is used to ensure the reliability and privacy of user data;

[0085] Based on the above data comparison, corresponding status labels are given, which facilitates users to quickly judge whether the virtual wearable products fit. At the same time, based on this status label and the data of the selected virtual wearable products, the recommendation and filling of virtual wearable products for the remaining wearing positions are completed, and the virtual wearable products are replaced and adjusted with one key, achieving the purpose of platform promotion and user recommendation, and providing reference support for subsequent one-key purchase;

[0086] Ultimately, it reduces the time for users to select wearable products and improves the suitability of the selected wearable products, thereby reducing the return and replacement rate of the platform.

[0087] The above-described embodiments merely represent one or more implementation manners of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation of the patent of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the appended claims.

Claims

1. A method for fitting a wearable article model to a human body model, characterized in that: Acquire user three-dimensional data, and acquire three-dimensional data stored in the user account, wherein the three-dimensional data is obtained by the user through offline measurement; Generate a three-dimensional human body model, the platform uses the three-dimensional data stored in the user account to generate a corresponding three-dimensional human body model of the user; Generate a wearable product model. The platform obtains wearable product data uploaded by the merchant, and the wearable product data at least includes wearable product three-dimensional data, fabric data, material data and style data. The platform reviews the wearable product data and generates corresponding virtual wearable products after the review is passed, and compares the generated virtual wearable products with the physical wearable product pictures uploaded by the merchant; The human body model is fitted with the wearable product model. In static fitting, initial spatial matching is performed so that the virtual wearable product is worn on the user's three-dimensional human body model and a preset area spacing is maintained. In physical driven fitting, fabric physical parameters and environmental force field parameters are loaded. Trying on virtual wearable products: the platform displays the virtual wearable products, the user's three-dimensional human body model and the try-on effect. The user selects the virtual wearable products on the platform and wears them on the selected user's three-dimensional human body model to complete the try-on.

2. The method for fitting a wearing article model to a human body model according to claim 1, characterized in that: Also includes: Data comparison, obtaining the three-dimensional data of the virtual wearable product selected by the user and the three-dimensional data of the user's three-dimensional human body model and performing numerical comparison, generating a corresponding status label according to the result of the numerical comparison, and the status label is used to feedback the fit between the virtual wearable product and the user's three-dimensional human body model.

3. The method for fitting a wearing article model to a human body model according to claim 1, characterized in that: The generating of the three-dimensional human body model further comprises: topology optimization and standardized posture; The topology optimization includes: adjusting the user three-dimensional human body model so that it has a uniform quadrilateral topology, and adding loops at the joints of the user three-dimensional human body model; The standardized posture includes: the initial posture adopts an A-type posture or a T-type posture, and performs bone binding and verifies the rationality of weights on the user's three-dimensional human body model used for animation.

4. The method for fitting a wearing article model to a human body model according to claim 1, characterized in that: The generating of the wearable article model further includes: topological compatibility before the generation of the virtual wearable article; The topological compatibility includes: adjusting the mesh density of the virtual wearable article to be generated to be higher than the mesh density of the user's three-dimensional human body model, so that the generated virtual wearable article meets the set wrinkle detail requirements, and giving priority to using quadrilaterals in the process of generating the wearable article model.

5. The method for fitting a wearing article model to a human body model according to claim 1, characterized in that: The static fitting in the fitting of the human body model and the wearable product model also includes: setting the user's three-dimensional human body model as a static collision body in the physical engine and loading a preset collision thickness, performing layer-by-layer simulation on the multiple layers of the virtual wearable product, and freezing each layer after completing the simulation and serving as the collision body of the next layer.

6. The method for fitting a wearing article model to a human body model according to claim 1, characterized in that: The fitting of the human body model and the wearable article model also includes: local penetration repair, wrinkle and drape enhancement; The local penetration repair includes: performing push-pull adjustment on the penetration area, adsorbing the vertices of the virtual wearable product to the surface of the user's three-dimensional human body model through a shrink wrap modifier, locking the first key position by setting a fixed point, the first key position includes a neckline and a cuff, and simulating the elastic fabric of the virtual wearable product through elastic constraints; The wrinkles and drape are enhanced, the virtual wearable product is subdivided through Catmull-Clark, a displacement map is performed on the virtual wearable product, high-poly details are baked into a normal map of the virtual wearable product, the sewing thread tension of the virtual wearable product is adjusted to slow down or eliminate tearing, and double stitching of the sewing thread is performed on a second key position of the virtual wearable product, wherein the second key position includes a crotch.

7. The method for fitting a wearing article model to a human body model according to claim 1, characterized in that: The fitting of the human body model and the wearable product model also includes: fabric delay optimization, and the fabric delay optimization is: enabling interpolation to smooth the fabric motion trajectory and increasing the number of physical operator steps to 5-8 steps.

8. The method for fitting a wearing article model to a human body model according to claim 1, characterized in that: The fitting of the human body model and the wearable product model also includes: optimizing the mutual penetration of multi-layer wearable products, wherein the mutual penetration optimization of multi-layer wearable products includes: performing hierarchical setting of collision groups for virtual wearable products, enabling interaction for virtual wearable products belonging to different groups, replacing mesh collision bodies with distance fields, and enabling continuous collision detection and increasing the number of physical solver steps according to the penetration amplitude when processing cloth penetration caused by high-speed motion.

9. The method for fitting a wearing article model to a human body model according to claim 1, characterized in that: The fitting of the human body model and the wearable product model also includes: dynamic tear optimization at the seams, and the dynamic tear optimization at the seams is: adding elastic properties to the sewing lines at the seams, and performing dynamic retopology on the high-pressure areas of the virtual wearable product.

10. The method for fitting a wearing article model to a human body model according to claim 1, characterized in that: Also includes: Customized 3D human body model beautification, users customize and upload real-life portraits. The system reviews and analyzes the facial features of the real-life portraits uploaded by users, and then generates a corresponding personalized cartoon portrait. The personalized cartoon portrait is mapped to the user's 3D human body model after customized skin color. If the user is a disabled person, the missing limbs of the user's 3D human body model will be repaired and filled.

Images