A dynamic dressing interaction system for a meta-universe virtual reality figure

Abstract

The application discloses a dynamic dressing interaction system for a metaverse virtual reality image, and relates to the technical field of virtual reality and computer graphics, comprising receiving virtual reality image data in a metaverse environment and outputting a spatial interaction tensor; generating a clothing 3D grid space through a thin plate spline interpolation algorithm, and performing light baking on the clothing 3D grid space to generate a static ambient light shadow map; mapping the clothing 3D grid space to a mathematical complex plane, generating a fractal base texture based on a complex seed point, a complex parameter and a physical interaction incremental term through iterative calculation, and synthesizing the fractal base texture and the static ambient light shadow map to obtain an intelligent fabric finally presented on the surface of the virtual reality image. The system realizes dynamic intelligent fabric with biological interaction characteristics under low computing power, and solves the technical pain points of virtual clothing fitting difficulty, no interaction and texture rigidity.

Description

Technical Field

[0001] The present invention relates to the technical field of a dynamic dress-up interactive system for a metaverse virtual reality avatar. Background Technology

[0002] With the rise of the metaverse concept, virtual reality avatars have become the core carriers for users in the digital world, and virtual clothing is a key element in expressing personality and identity. Traditional virtual try-on and dress-up technologies mainly rely on static texture mapping and physical fabric calculation. In traditional technologies, resolution limitations result in pixelation or blurring when viewed up close; interaction is lacking; textures are static and cannot dynamically respond to the user's touch or gaze; there is a disconnect from physics, as textures cannot respond to the special physical environment of the metaverse, such as the flow of patterns on clothing not changing when the direction of gravity changes; and resource consumption is high, as high-definition textures require a large amount of video memory and storage space.

[0003] Currently, Chinese invention application CN202311678137.8 discloses an interactive method, apparatus, device, and medium for changing virtual avatars. The method includes receiving a request from a first user to change the virtual avatar of a second user; obtaining the character information of the second user's first virtual avatar; modifying the character information according to the request to change the appearance of the first virtual avatar, thus obtaining a second virtual avatar; and storing the second virtual avatar and the modified character information. This application can modify the virtual avatars of other users and can be widely applied in the field of virtual avatar technology. However, existing technologies have significant drawbacks in metaverse VR interactive scenarios: traditional texture resolution is insufficient and video memory usage is high; virtual clothing lacks interactive feedback (e.g., no response to touch); and clothing textures, geometric deformations, and lighting effects are difficult to balance in real-time rendering. Summary of the Invention

[0004] The technical problem solved by this invention is that existing virtual dynamic dress-up interaction systems, in the metaverse environment, cannot generate virtual clothing textures with infinite resolution at low cost, cannot make real-time dynamic responses to user gaze and touch, and lack adapted complex physical rules.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a dynamic dress-up interaction system for a metaverse virtual reality avatar, comprising a perception module, a topology mapping module, and a fractal loom module:

[0006] The perception module is used to receive virtual reality image data in the metaverse environment, construct a virtual reality image semantic field based on the virtual reality image data, calculate the parallax angle and proximity sensing parameters based on the pose data of the VR headset and hand controller, and combine the parallax angle and proximity sensing parameters to output a spatial interaction tensor.

[0007] The topology mapping module is used to generate a 3D mesh space for clothing based on digital pattern atlases and virtual reality image semantic fields, using a thin plate spline interpolation algorithm, and to perform lighting baking on the 3D mesh space for clothing to generate a static ambient light occlusion map.

[0008] The fractal loom module is used to map the 3D mesh space of the garment to the mathematical complex plane, calculate the dynamically changing complex seed points and complex parameters based on the spatial interaction tensor, and calculate the physical interaction increment term in combination with environmental metadata. Based on the complex seed points, complex parameters and physical interaction increment term, the fractal basic texture is generated through iterative calculation. The fractal basic texture is synthesized with the static ambient light occlusion map to obtain the smart fabric finally presented on the surface of the virtual reality image.

[0009] Preferably, the perception module includes a data acquisition unit, a virtual reality image semantic field unit, and a spatial interaction unit:

[0010] The data acquisition unit includes:

[0011] The virtual reality character data of the current frame is obtained from the environment in which the metaverse VR is running. The virtual reality character data includes skeletal pose data and surface mesh data.

[0012] The skeletal posture data includes the real-time first three-dimensional position vector coordinates and first quaternion rotation data of key joints in the world coordinate system. The key joints include the cervical spine, shoulder joint, elbow joint, wrist joint, hip joint, knee joint, and ankle joint. The first three-dimensional position vector coordinates and first quaternion rotation data of all key joints are stored in matrix form to form a set of current posture transformation matrices of the virtual reality image.

[0013] All bone pose data are deformed in real time using a linear blending skinning algorithm to obtain surface mesh data, which includes vertex coordinates, normal vectors, and texture coordinates. The surface mesh data is stored in video memory in the form of vertex buffers and index buffers.

[0014] Preferably, the virtual reality image semantic field unit includes:

[0015] The construction process of building a semantic field for virtual reality images based on virtual reality image data includes:

[0016] During the import phase of the virtual reality avatar, semantic tags are predefined for each component of the virtual reality avatar. The semantic tags include the torso, left hand, right hand, head, and hair. The semantic tags include material ID information and vertex color information.

[0017] Off-screen rendering is performed on the virtual reality image data. The off-screen rendering process includes:

[0018] The virtual reality image is rendered by disabling lighting and textures, and only assigning a unique color to the corresponding component based on predefined semantic tags. The result of the secondary rendering is stored as a semantic ID texture.

[0019] The semantic ID texture is post-processed to extract and generate torso mask, hand mask, head mask and hair mask respectively. The torso mask, hand mask, head mask and hair mask correspond to each pixel region represented in the form of binary mask texture under the current VR view. Each pixel region includes torso region, hand region, head region and hair region.

[0020] The set of torso mask, hand mask, head mask, and hair mask, which are represented in the form of binary mask textures, is defined as the virtual reality image semantic field.

[0021] Preferably, the spatial interaction unit includes:

[0022] The pose data of the VR headset and the user's hand controller in the world coordinate system are obtained in real time from the SDK interface of the VR device. The pose data of the VR headset includes the real-time second three-dimensional position vector coordinates and the second quaternion rotation data of the VR headset. The pose data of the hand controller includes the real-time third three-dimensional position vector coordinates of the hand controller.

[0023] Using the pose data of the VR headset when the user begins to observe the virtual reality image as the initial reference pose data, the matrix product of the current VR headset's second quaternion rotation data and the second quaternion rotation data in the initial reference pose data is calculated to obtain the relative second quaternion rotation matrix. The relative second quaternion rotation matrix is ​​then converted into Euler angles to obtain the current VR headset's rotation increment. The three angular components of the Euler angles include pitch angle, yaw angle, and roll angle.

[0024] Calculate the first displacement difference between the current VR headset's second three-dimensional position vector coordinates and the second three-dimensional position vector coordinates in the initial reference pose data, and obtain the current VR headset's displacement increment;

[0025] The displacement increment and rotation increment are combined into a parallax angle;

[0026] The Boundary Volume Hierarchy (BVH) algorithm is used to optimize the search of the torso region and calculate the minimum Euclidean distance from the pose data of the hand controller to the surface mesh data corresponding to the torso region.

[0027] Obtain the nearest first three-dimensional position vector coordinates and the normal vector corresponding to the nearest first three-dimensional position vector coordinates from the surface mesh data corresponding to the torso region to the pose data of the hand controller; calculate the second vector difference between the pose data of the hand controller and the nearest first three-dimensional position vector coordinates.

[0028] Calculate the projection value of the second vector difference onto the direction of the normal vector corresponding to the nearest first three-dimensional position vector coordinates. The projection value is the dot product of the second vector difference and the normal vector corresponding to the nearest first three-dimensional position vector coordinates.

[0029] The proximity sensing parameters between the hand region and the torso region are calculated, and the mathematical expression for the proximity sensing parameters is as follows:

[0030] ;

[0031] in, For proximity sensing parameters, This represents the minimum Euclidean distance from the pose data of the hand controller to the surface mesh data corresponding to the torso region. The preset sensing radius, The second vector difference, The normal vector corresponding to the nearest three-dimensional position vector coordinates. This refers to the pressure sensitivity coefficient.

[0032] The disparity angle and proximity sensing parameters are combined into a spatial interaction tensor.

[0033] Preferably, the topology mapping module includes a mesh redirection unit and a lighting baking unit:

[0034] The mesh redirection unit is used to read the digital pattern atlas of the garment, calculate the transformation matrix using the thin-plate spline interpolation algorithm TPS, and map the vertex coordinates of the digital pattern atlas to the geometric topological surface of the current virtual reality image to generate a suitable 3D mesh space for the garment. The generation process includes:

[0035] Preset corresponding point groups of 2D digital template atlas and 3D virtual reality image surface mesh data as constraint points;

[0036] Based on constraint points, the transformation matrix is ​​calculated using the thin plate spline interpolation algorithm TPS. All vertices in the digital pattern atlas are mapped to the clothing 3D mesh space through the transformation matrix. The vertex connection relationship in the clothing 3D mesh space is completely consistent with the vertex connection relationship in the digital pattern atlas.

[0037] The lighting baking unit is used to separate the ambient occlusion channel in the rendering pipeline, convert the clothing 3D mesh space to the perceptual color space, extract the luminance component L, and generate a static ambient occlusion map that retains physical wrinkle information. The generation process includes:

[0038] In the rendering pipeline, a third rendering is performed on the clothing 3D mesh space. The third rendering includes calculating the number of beams of diffuse light from the surrounding environment that each point in the clothing 3D mesh space can receive, and outputting an ambient occlusion map, which is a grayscale image.

[0039] Convert the ambient occlusion map to the Lab color space, discard the A and B channels, keep only the luminance channel, and output a static ambient occlusion map.

[0040] Preferably, the fractal loom module includes a holographic complex plane projection unit, a biological state iterative solution unit, a potential energy field coloring unit, and a material synthesis unit.

[0041] The holographic complex plane projection unit includes:

[0042] The texture coordinates of the clothing's 3D mesh space are read and mapped to a mathematical complex plane. Based on this mathematical complex plane, a complex seed point is calculated using a phase offset function according to the disparity angle. The mathematical expression for calculating the complex seed point is as follows:

[0043] ;

[0044] in, For complex seed points, For space mapping functions, For texture coordinates, For parallax angle, This is the phase offset function.

[0045] Preferably, the biological state iterative solution unit includes:

[0046] A preset basic complex constant is established, and proximity sensing parameters are monitored in real time. Based on the influence rules of the dynamically changing proximity sensing parameters on the basic complex constant, dynamic complex parameters are obtained. The influence rules include:

[0047] When the proximity sensing parameter is 0 and negative, the fundamental complex constant remains unchanged;

[0048] When the proximity sensing parameter is positive, the imaginary part of the fundamental complex constant undergoes sinusoidal oscillation;

[0049] Obtain environmental metadata of the metaverse, and construct physical interaction constraint variables based on environmental metadata and spatial interaction tensor. The physical interaction constraint variables include physical rule state vector and interaction intensity field. The physical rule state vector includes the gravitational acceleration vector of the current region of the metaverse, the damping coefficient of the current environmental medium, and the time flow pattern. The interaction intensity field includes the collision energy, dragging energy, and object motion speed generated by the hand controller.

[0050] Calculate the physical interaction increment term based on the physical interaction constraint variables;

[0051] The dynamic fractal driving parameters are calculated based on dynamic complex parameters and physical interaction increment terms. The mathematical expression for the dynamic fractal driving parameters is as follows:

[0052] ;

[0053] in, For dynamic fractal driving parameters, For dynamic complex parameters, The feature extraction quantity represents the rule governing the influence of dynamically changing proximity sensing parameters on the underlying complex constant. For physical interaction increments, For physical interaction constraint variables, Texture coordinates , For time;

[0054] Based on complex seed points and dynamic fractal driving parameters, iterative calculations are performed on each pixel in the 3D mesh space of the clothing. The mathematical expression for the iterative calculation is as follows:

[0055] ;

[0056] Where, when n is 0, These are the initial coordinates of the corresponding pixel points in the 3D mesh space of the clothing. These initial coordinates are the coordinates obtained by mapping the texture coordinates using a spatial mapping function. The state of the pixel after n+1 iterations. The state of the pixel after n iterations;

[0057] Record the number of iterations and the final modulus required for each pixel to escape from the preset radius.

[0058] Preferably, the process of obtaining the physical interaction increment specifically includes:

[0059] Step S21: Through the physical interface of the VR or AR engine, read the environmental attributes of the coordinate area where the current virtual reality image is located in real time and obtain environmental metadata. The environmental attributes include normal physical area, anti-gravity area, zero gravity area and spatiotemporal turbulence area. The environmental metadata includes gravity acceleration vector, damping coefficient of the current environmental medium and the defined time elapsed ratio of the current scene.

[0060] The current environmental media include air, water, and vacuum, and the time elapsed factor is defined as 1 for normal and 0.1 for bullet time;

[0061] Step S22: Based on environmental metadata, perform localized mapping of physical rules on the clothing 3D mesh space. This localized mapping includes calculating the gravitational acceleration vector of the current region in the metaverse and setting the time flow pattern.

[0062] Calculate the gravitational acceleration vector of the current region of the metaverse. When in the normal physics region, the gravitational acceleration vector of the current region of the metaverse is equal to the standard free fall acceleration.

[0063] When in the anti-gravity region, the gravitational acceleration vector of the current region of the metaverse is equal to the negative standard free fall acceleration;

[0064] When in a zero-gravity region, the gravitational acceleration vector of the current region of the metaverse is equal to 0;

[0065] Set the identifier for the time flow mode: setting the identifier to 1 indicates that time flows forward, setting the identifier to -1 indicates that time flows backward, and setting the identifier to 0 indicates that time is frozen.

[0066] The gravitational acceleration vector of the current region of the metaverse, the damping coefficient of the current environmental medium, and the defined time elapsed factor of the current scene are encapsulated into a physical rule state vector;

[0067] Step S23: Obtain the spatial interaction tensor between the previous frame and the current frame from the spatial interaction unit, and extract dynamic features through differential calculation. The extraction process includes:

[0068] Calculate the ratio of the displacement difference of the hand controller relative to the torso region to the frame interval to obtain the interaction velocity vector;

[0069] Monitor proximity sensing parameters; when the proximity sensing parameter is greater than 0, it is determined to be a contact or collision state.

[0070] When the proximity sensing parameter is continuously positive and the corresponding texture coordinate position undergoes tangential displacement, it is determined to be a dragging state.

[0071] Step S24: Based on the kinetic energy theorem, calculate the collision energy generated by the hand controller at the instant the proximity sensing parameter changes from 0 to positive. The mathematical expression for the collision energy generated by the hand controller is:

[0072] ;

[0073] in, The collision energy generated by the hand controller. The square of the modulus of the interaction speed. For interactive velocity vectors, This is a preset virtual quality coefficient;

[0074] In drag mode, the drag energy is calculated, and the mathematical expression for the drag energy is:

[0075] ;

[0076] in, To drag energy, The coefficient of surface friction, This represents the tangential displacement length at the texture coordinate position. This refers to the velocity component of the user's hand controller rubbing along the 3D mesh surface of the garment when it comes into contact with the garment. The initial time at the moment of contact;

[0077] The collision energy, drag energy, and motion speed generated by the hand controller are encapsulated into a physical rule state vector;

[0078] Step S25: Merge the physical rule state vector and the interaction intensity field through channels to output the final physical interaction constraint variables, the mathematical expression of which is:

[0079] ;

[0080] in, For physical interaction constraint variables, This is the physical rule state vector. For the interaction intensity field, This is a vector concatenation symbol;

[0081] Step S26: Calculate the changes of each component of the gravitational acceleration vector and the interaction intensity field within a preset time period, and multiply the sum of the corresponding changes by the time flow rate mode to obtain the physical interaction increment term.

[0082] Preferably, the potential field coloring unit includes:

[0083] Based on the final iteration count and final modulus of each pixel, the normalized escape velocity is calculated. The mathematical expression for the normalized escape velocity is:

[0084] ;

[0085] in, To normalize the escape velocity, For the final module length, This represents the maximum number of iterations.

[0086] By using the programmatic color palette function Palette, all normalized escape velocities are mapped to RGB color values, and the combined mapping results are used to generate a fractal base texture.

[0087] Preferably, the material synthesis unit includes:

[0088] The fractal base texture is blended with a static ambient occlusion map using Multiply blending to obtain the final output smart fabric. The mathematical expression for Multiply blending is:

[0089] ;

[0090] in, It is a smart fabric. For fractal basic texture, This is a static ambient occlusion texture.

[0091] The beneficial effects of this invention are as follows: This invention introduces procedural fractal mathematics and complex plane holographic mapping theory. Utilizing the infinite self-similarity of complex dynamics and its sensitive dependence on initial conditions, it achieves pixel-level real-time generation of virtual fabric textures by adaptively adjusting complex seed points and phase offset parameters, completely abandoning traditional high-resolution static texture mapping. By employing a phase offset function driven by the parallax angle and a bio-state iterative operator driven by the proximity field, it constructs a holographic visual illusion mechanism that flows with the viewing angle and a bio-stress mechanism that responds nonlinearly with touch intensity, respectively. This imbues virtual clothing with digital vitality from a mathematical perspective, effectively solving the problems of rigid textures and lack of interactivity in traditional methods. Not only does it optimize visual performance, but it also utilizes a decoupled mapping strategy of thin-plate spline interpolation and static ambient light occlusion. While ensuring the clothing automatically and perfectly fits different topological human bodies, it preserves the physical-level wrinkle texture through pre-calculated lighting and shadows, eliminating the computational bottleneck of real-time ray tracing. The fractal driving parameters are determined by analytical calculation based on environmental metadata and interactive kinetic energy, replacing the expensive real-time cloth physics collision calculation. This greatly reduces the video memory usage and GPU computing load, enabling the dynamic dress-up interaction system with infinite resolution to run on VR all-in-one machines with limited computing power, meeting the high frame rate real-time rendering requirements of the metaverse immersive environment. Attached Figure Description

[0092] Figure 1 This is a basic flowchart of a dynamic dress-up interaction system for a metaverse virtual reality character, provided as an embodiment of the present invention. Detailed Implementation

[0093] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0094] Reference Figure 1 As one embodiment of the present invention, a dynamic dress-up interaction system for a metaverse virtual reality avatar is provided, comprising a perception module, a topology mapping module, and a fractal loom module:

[0095] The perception module is used to receive virtual reality image data in the metaverse environment, construct a virtual reality image semantic field based on the virtual reality image data, calculate the parallax angle and proximity sensing parameters based on the pose data of the VR headset and hand controllers, and output the parallax angle and proximity sensing parameters as a spatial interaction tensor.

[0096] The topology mapping module is used to generate a 3D mesh space for clothing based on digital pattern atlases and virtual reality image semantic fields, using a thin plate spline interpolation algorithm, and to perform lighting baking on the 3D mesh space of clothing to generate a static ambient light occlusion map.

[0097] The fractal loom module is used to map the 3D mesh space of clothing to the mathematical complex plane. It calculates the dynamically changing complex seed points and complex parameters based on the spatial interaction tensor, and calculates the physical interaction increment term in combination with environmental metadata. Based on the complex seed points, complex parameters and physical interaction increment term, it generates the fractal basic texture through iterative calculation. The fractal basic texture is then synthesized with the static ambient light occlusion map to obtain the smart fabric that is finally presented on the surface of the virtual reality image.

[0098] This invention constructs a semantic field and spatial interaction tensor through a perception module, transforming user gaze and touch into mathematical driving forces. Utilizing thin-plate spline interpolation and illumination baking in the topology mapping module, it achieves adaptive and perfect fit for different body types while preserving physical texture at low cost. The fractal loom module generates dynamic textures in real-time based on complex plane iterative calculations, eliminating static texture mapping. This system realizes dynamic intelligent fabrics with biological interaction characteristics under low computing power, solving the technical pain points of difficult virtual clothing adaptation, lack of interaction, and rigid textures.

[0099] The perception module includes a data acquisition unit, a virtual reality image semantic field unit, and a spatial interaction unit:

[0100] The data acquisition unit includes:

[0101] The virtual reality character data of the current frame is obtained from the environment in which the metaverse VR is running. The virtual reality character data includes skeletal pose data and surface mesh data.

[0102] Skeletal pose data includes the real-time first three-dimensional position vector coordinates and first quaternion rotation data of key joints in the world coordinate system. Key joints include cervical spine, shoulder joint, elbow joint, wrist joint, hip joint, knee joint and ankle joint. The first three-dimensional position vector coordinates and first quaternion rotation data of all key joints are stored in matrix form to form a set of current pose transformation matrices of virtual reality image.

[0103] All bone pose data are deformed in real time using a linear blending skinning algorithm to obtain surface mesh data. The surface mesh data includes vertex coordinates, normal vectors, and texture coordinates. The surface mesh data is stored in video memory in the form of vertex buffers and index buffers.

[0104] This unit generates real-time meshes by reading skeletal data and combining it with a linear hybrid skinning algorithm. The system not only knows the position of the virtual avatar but also how its surface deforms, which is the foundation for all subsequent texture bonding. It achieves high-precision capture of the dynamic deformation of virtual reality avatars in the VR environment, providing the most basic and real-time geometric and pose information for subsequent semantic field construction and spatial interaction analysis, ensuring the accuracy and immediacy of the system's perception of the virtual reality avatar's state. Storing mesh data in an index buffer avoids frequent large-scale data transfers between the CPU and GPU, ensuring smoothness under the high frame rate requirements of VR.

[0105] Virtual reality image semantic field units include:

[0106] The construction process of building a semantic field for virtual reality images based on virtual reality image data includes:

[0107] During the import phase of the virtual reality avatar, semantic tags are predefined for each component of the virtual reality avatar. The semantic tags include the torso, left hand, right hand, head, and hair. The semantic tags also include material ID information and vertex color information.

[0108] Off-screen rendering of virtual reality character data includes the following processes:

[0109] Disable lighting and textures from performing secondary rendering on virtual reality images, and assign unique colors to corresponding components only based on predefined semantic tags; store the secondary rendering results as semantic ID textures.

[0110] Assign a unique color, for example: render the torso as pure red, the left hand as pure green, and the hair as pure blue;

[0111] The semantic ID texture is post-processed to extract and generate torso mask, hand mask, head mask and hair mask respectively. The torso mask, hand mask, head mask and hair mask correspond to each pixel region represented in the form of binary mask texture under the current VR view. Each pixel region includes torso region, hand region, head region and hair region.

[0112] The set of torso mask, hand mask, head mask, and hair mask, which are represented in the form of binary mask textures, is defined as the virtual reality image semantic field.

[0113] This solution generates color-coded maps (semantic ID textures) through off-screen rendering, transforming complex 3D geometric judgments into extremely low-cost 2D texture sampling operations. It clearly distinguishes the positions of each part, allowing each part to be defined with different physical feedback parameters, greatly improving the finesse of the interaction logic.

[0114] Spatial interaction units include:

[0115] The pose data of the VR headset and the user's hand controller in the world coordinate system are obtained in real time from the SDK interface of the VR device. The pose data of the VR headset includes the real-time second three-dimensional position vector coordinates and the second quaternion rotation data of the VR headset. The pose data of the hand controller includes the real-time third three-dimensional position vector coordinates of the hand controller.

[0116] Using the VR headset's pose data when the user begins observing the virtual reality image as the initial reference pose data, the matrix product of the current VR headset's second quaternion rotation data and the second quaternion rotation data in the initial reference pose data is calculated to obtain the relative second quaternion rotation matrix. The relative second quaternion rotation matrix is ​​then converted into Euler angles to obtain the current VR headset's rotation increment. The three angular components of the Euler angles include pitch, yaw, and roll.

[0117] Calculate the first displacement difference between the current VR headset's second three-dimensional position vector coordinates and the second three-dimensional position vector coordinates in the initial reference pose data, and obtain the current VR headset's displacement increment;

[0118] The displacement increment and rotation increment are combined into a parallax angle;

[0119] In this embodiment, the parallax angle is a 4-dimensional vector, where the first three components are the three angular components of the Euler angle (rotation increment), and the fourth component is the first displacement difference (displacement increment).

[0120] The Boundary Volume Hierarchy (BVH) algorithm is used to optimize the search of the torso region and calculate the minimum Euclidean distance from the pose data of the hand controller to the surface mesh data corresponding to the torso region.

[0121] Obtain the nearest first three-dimensional position vector coordinates and the normal vector corresponding to the nearest first three-dimensional position vector coordinates from the surface mesh data corresponding to the torso region to the pose data of the hand controller; calculate the second vector difference between the pose data of the hand controller and the nearest first three-dimensional position vector coordinates.

[0122] Calculate the projection value of the second vector difference onto the direction of the normal vector corresponding to the nearest first three-dimensional position vector coordinates. The projection value is the dot product of the second vector difference and the normal vector corresponding to the nearest first three-dimensional position vector coordinates.

[0123] The proximity sensing parameters between the hand region and the torso region are calculated. The mathematical expression for the proximity sensing parameters is as follows:

[0124] ;

[0125] in, For proximity sensing parameters, This represents the minimum Euclidean distance from the pose data of the hand controller to the surface mesh data corresponding to the torso region. The preset sensing radius, The second vector difference, The normal vector corresponding to the nearest three-dimensional position vector coordinates. This refers to the pressure sensitivity coefficient.

[0126] A proximity sensing parameter value of 0 indicates no interaction (clothes are still), while a larger value indicates a stronger interaction and more intense texture fluctuations (turbulence, glow) on the clothes.

[0127] The preset sensing radius defines the aura range of the virtual clothing, set to 0.2 meters. The reaction calculation only begins when a hand enters within this distance.

[0128] For distance-weighted terms, a subtraction logic is used, with the closer the hand (…). The smaller the value, the larger the difference, and the stronger the reaction.

[0129] This is the intention multiplier factor. The base multiplier is 1 (i.e., just close). If a pressing intention is detected, the multiplier will be greater than 1, amplifying the reaction.

[0130] The pressure sensitivity coefficient is a preset constant (set to 5.0). The larger the value, the more exaggerated the response to the pressing action.

[0131] This is used to determine whether a hand is hovering outside the clothing or trying to press it inside. When the result is positive, the hand is on the outside. The result is negative; the hand is inside or squeezing inward.

[0132] Only retain negative values ​​(press component) and filter out positive values ​​(outer hover component).

[0133] The parallax angle and proximity sensing parameters are combined into a spatial interaction tensor.

[0134] Each frame updates the boundary volume hierarchy of the torso region based on the updated surface mesh data.

[0135] This step quantifies the core interactive behaviors of users in VR (changes in viewing angle and hand touch). Parallax angle drives dynamic changes in fractal textures based on viewing angle, increasing visual depth; proximity sensing parameters directly map the user's tactile intentions, driving the bio-responsive behavior of the fractal textures, greatly enhancing the interactivity and immersion of the virtual clothing. The calculation process ensures real-time performance and accuracy through optimized algorithms.

[0136] By fusing viewing (parallax angle) and touching (proximity) into a unified tensor data stream, the algorithm specifically distinguishes between hovering and pressing. Through calculation... Precisely determining whether a hand is floating on or sinking into the surface allows the virtual fabric to exhibit two distinct physical feedbacks: light when approached and ripples when pressed, enhancing the sense of immersion.

[0137] The perception module is the core front-end perception component of the Metaverse virtual reality avatar dynamic dressing interaction system. It aims to capture and accurately analyze the spatial state and interaction intentions of the user's virtual reality avatar within the Metaverse environment in real time. By acquiring high-fidelity skeletal posture and surface geometry data of the virtual reality avatar and performing semantic segmentation, it constructs spatial masks describing various parts of the avatar. Simultaneously, it accurately tracks the dynamics of the VR headset and hand controllers, extracting parallax angles reflecting the user's viewing angle and proximity sensing between the hands and torso.

[0138] The topology mapping module includes mesh redirection units and lighting baking units:

[0139] The mesh redirection unit is used to read the digital pattern atlas of the garment, calculate the transformation matrix using the thin-plate spline interpolation algorithm TPS, and map the vertex coordinates of the digital pattern atlas to the geometric topological surface of the current virtual reality avatar, generating a suitable 3D mesh space for the garment. The generation process includes:

[0140] Developers pre-set corresponding point groups of 2D digital template atlases and 3D virtual reality image surface mesh data as constraint points;

[0141] Based on constraint points, the transformation matrix is ​​calculated using the thin plate spline interpolation algorithm TPS. All vertices in the digital pattern atlas are mapped to the clothing 3D mesh space through the transformation matrix. The vertex connection relationship in the clothing 3D mesh space is completely consistent with the vertex connection relationship in the digital pattern atlas.

[0142] In this embodiment, the geometric topological surface of the current virtual reality image is a 3D human body model in .obj format. The digital pattern atlas of clothing includes the input 2D scanned pattern drawings of clothing. The 2D scanned pattern drawings of clothing define the shape and stitching relationship of each part of the clothing (such as the front piece, back piece, and sleeves) and are completely planar geometric figures.

[0143] Preset constraint points, for example:

[0144] Map the center point of the collar on the 2D digital pattern atlas to the point at the base of the neck on the 3D virtual reality avatar;

[0145] Align the end of the shoulder line with the apex of the shoulder;

[0146] Align the center of the cuff with the wrist;

[0147] These constraint points are used to tell the algorithm which key locations of the blueprint should be placed in the virtual reality avatar.

[0148] The topological structure (vertex connection relationship) of the 3D mesh space of the clothing is completely consistent with the original 2D digital pattern atlas of the clothing, but its three-dimensional shape has perfectly matched the body curves of the virtual reality image.

[0149] The mesh redirection unit allows a 2D clothing design blueprint to automatically and smoothly wrap around a 3D virtual reality avatar, achieving a perfect fit.

[0150] The lighting baking unit is used to separate the ambient occlusion channel in the rendering pipeline, convert the clothing 3D mesh space to the perceptual color space, extract the luminance component L, and generate a static ambient occlusion map that retains physical wrinkle information. The generation process includes:

[0151] In the rendering pipeline, a third rendering is performed on the clothing 3D mesh space. The third rendering includes calculating the number of diffuse light beams that each point in the clothing 3D mesh space can receive from the surrounding environment, and outputting an ambient occlusion map. The ambient occlusion map is a grayscale image, which represents all shadow information generated by geometry (folds, overlaps).

[0152] Convert the ambient occlusion map to the Lab color space, discard the A and B channels, keep only the luminance channel, and output a static ambient occlusion map.

[0153] Points deep within the folds receive very little light because they are blocked by the surrounding fabric surface, resulting in a low ambient light occlusion value (leaning towards black).

[0154] Points on a flat surface can receive light from all directions, resulting in a high ambient light occlusion value (leaning towards white).

[0155] Ambient occlusion is a technique that simulates the effect of gaps and corners being darker than a flat surface. Imagine the corner of a room; it's always darker than the center of the wall because it's blocked by other surfaces from receiving ambient light. In clothing, deep folds, below the collar, and under the armpits are typical areas that are occluded. Ambient occlusion maps contain lighting information, but sometimes this information is mixed with other color information (especially in more complex rendering workflows). The Lab color space separates image information into luma, A, and B channels. The luma channel includes information from pure black to pure white, the A channel includes the color spectrum from green to red, and the B channel includes the color spectrum from blue to yellow. The Lab color space is designed to closely resemble human visual perception, and it perfectly separates luminance from color.

[0156] The static ambient occlusion map is a grayscale image that contains only physical wrinkles and shadow / light relationships, perfectly preserving the three-dimensionality of the clothing.

[0157] The purpose of this unit is to give the well-fitting model natural light and shadow created by its geometry.

[0158] The topology mapping module outputs a 3D mesh space for the clothing and a static ambient occlusion map. The 3D mesh space for the clothing is a well-fitting 3D model, and the light map is a texture that records all the wrinkles and shadows.

[0159] When it's necessary to dress a virtual reality avatar in blue jeans, the rendering engine no longer needs to perform complex lighting calculations. The rendering engine's simplified steps include:

[0160] Step S11: Input a basic blue denim tile texture;

[0161] Step S12: Input the pre-baked static ambient occlusion map;

[0162] Step S13: In the shader, multiply the blue denim tile texture, fractal texture, and static ambient occlusion map.

[0163] In this way, the blue denim texture remains unchanged in the white areas (flat areas) of the static ambient occlusion map, while the black areas (deep wrinkles) of the static ambient occlusion map are darkened. The end result is a pair of blue jeans with realistic, physically accurate wrinkle shadows, and the computational cost is extremely low.

[0164] This decouples geometry from texture, providing a solid technical foundation for large-scale, high-performance digital fashion systems.

[0165] Thin plate spline interpolation is a deformation algorithm that minimizes energy. It ensures that when a 2D pattern is mapped onto a 3D human body of different sizes and heights, it can fit the surface while minimizing the stretching and distortion of the texture.

[0166] Fractal textures are mathematically calculated and have no thickness. By baking static ambient occlusion (AO), the shadows at the folds are pre-made into the texture, making the final dynamic texture look like realistic fabric undulations, without the need for expensive real-time ray tracing calculations.

[0167] The static ambient occlusion map in this unit records the inherent wrinkles and shadows created by the cutting and sewing of clothing in a standard pose. In dynamic costume changes, this map serves as a base shadow layer, overlaid with dynamic lighting and shadows in the real-time rendering pipeline, enhancing the volume of the clothing while maintaining low computational power.

[0168] This module solves the problem of clothing fit and cleverly decouples geometric deformation from texture generation. By extracting the luminance component as a lighting mask, it ensures that the subsequently generated mathematical fractal textures can accurately represent the fabric's inherent wrinkles and shadows.

[0169] The fractal loom module includes a holographic complex plane projection unit, a biological state iterative solution unit, a potential energy field coloring unit, and a material synthesis unit:

[0170] The holographic complex plane projection unit includes:

[0171] The texture coordinates of the clothing's 3D mesh space are read and mapped to the mathematical complex plane. Based on the mathematical complex plane, complex seed points are calculated using a phase offset function according to the disparity angle. The mathematical expression for calculating the complex seed points is as follows:

[0172] ;

[0173] in, For complex seed points, For space mapping functions, For texture coordinates, For parallax angle, This is the phase offset function.

[0174] Seed point of complex number in the mathematical complex plane This represents a specific point (in the form x+iy), which determines the specific position of the currently calculated pixel in the mathematical fractal pattern;

[0175] Space mapping function The function is a function in the prior art that transforms the geometric texture space to the mathematical complex plane. It is used to read the normalized texture coordinates (u,v) (ranging from 0 to 1) of the surface of the clothing model and non-linearly scale and translate them to a specific area on the complex plane. It ensures that every physical position of the clothing (such as the collar and cuffs) corresponds to a certain basic area of ​​the mathematical fractal pattern, ensuring that the texture is attached to the clothing.

[0176] Phase offset function Used to calculate minute displacements on the mathematical complex plane based on parallax angles.

[0177] ;

[0178] in, and The preset sensitivity coefficient, The pitch component of the parallax angle. The yaw angle component in the parallax angle. Using the imaginary unit, the phase offset function transforms the user's viewing angle into a coordinate offset on the complex plane, thereby achieving a holographic optical illusion where the texture moves with the viewing angle.

[0179] The real part sensitivity controls the amount of texture movement in the real axis (X-axis) direction of the complex plane when the user nods up or down (changes in pitch angle);

[0180] The vertical phase gain coefficient (or imaginary part sensitivity) controls the amount by which the texture moves along the imaginary axis (Y-axis) of the complex plane when the user shakes their head left or right (changes in yaw angle).

[0181] set up and It enables developers to simulate fabrics with a specific weave direction, if = If the material is isotropic (it looks the same from all directions), it is similar to ordinary plastic or glass. These two parameters are given to the designer as material properties and can be adjusted in the material panel of the development engine (Unity / Unreal).

[0182] The value range is a small floating-point number, between [0.0, 1.0].

[0183] The parallax angle is the input signal (how the user moves), and the periodic function ensures that the texture can smoothly return to its original state after the line of sight completes a full circle, without any abrupt changes;

[0184] and It is an amplifier. If the coefficient is 0, then Φ(θ)=0. At this time, no matter how the user looks at it, the texture is dead (degenerates into a traditional texture). The larger the coefficient, the stronger the parallax effect and the stronger the holographic effect.

[0185] static The part responsible for printing the pattern onto the clothing will have a static, unchanging pattern, just like a traditional texture, regardless of how the user turns their head to look at it, if no phase offset function is added. (Dynamic part) The user's viewing angle (parallax angle) is introduced as a variable to create a holographic illusion: when the user turns their head or moves in VR, the parallax angle changes. A tiny complex offset is calculated that propels the seed point z to slide gently across the complex plane.

[0186] The formula for complex seed points combines the geometric coordinates (u,v) of the garment surface with the user's viewing angle. By adding them together, the complex seed point z used to generate the final texture is calculated. When the user stares at the clothes and turns their head, the phase shift function causes the texture coordinates to shift slightly on the complex plane. The pattern on the clothes will not be completely still as it is printed on the fabric, but will produce subtle shifts and flickers according to the angle of the view, thus visually simulating the three-dimensional luster and texture of holographic anti-counterfeiting materials or high-grade silk.

[0187] This solution introduces a phase shift function. When the user turns their head, the texture undergoes a slight displacement on the complex plane. The human brain interprets this as depth or gloss, thus simulating the deep three-dimensional texture of holographic anti-counterfeiting labels, silk, or cat's eye stones.

[0188] This unit is used to establish the mapping from geometric space to mathematical space and introduces viewpoint-based dynamic perturbation.

[0189] The biological state iterative solution unit includes:

[0190] A pre-defined basic complex constant is used, and proximity sensing parameters are monitored in real time. Based on the rules governing the influence of the dynamically changing proximity sensing parameters on the basic complex constant, dynamic complex parameters are obtained. These rules include:

[0191] When the proximity sensing parameter is 0 and negative, the fundamental complex constant remains unchanged;

[0192] When the proximity sensing parameter is positive, the imaginary part of the fundamental complex constant undergoes sinusoidal oscillation;

[0193] Obtain environmental metadata of the metaverse, and construct physical interaction constraint variables based on environmental metadata and spatial interaction tensor. The physical interaction constraint variables include physical rule state vector and interaction intensity field. The physical rule state vector includes the gravitational acceleration vector of the current region of the metaverse, the damping coefficient of the current environmental medium and the time flow pattern. The interaction intensity field includes the collision energy, drag energy and object motion speed generated by the hand controller.

[0194] Calculate the physical interaction increment term based on the physical interaction constraint variables;

[0195] The dynamic fractal driving parameters are calculated based on dynamic complex parameters and physical interaction increment terms. The mathematical expression for the dynamic fractal driving parameters is as follows:

[0196] ;

[0197] in, For dynamic fractal driving parameters, For dynamic complex parameters, The feature extraction quantity represents the rule governing the influence of dynamically changing proximity sensing parameters on the underlying complex constant. For physical interaction increments, For physical interaction constraint variables, Texture coordinates , For time;

[0198] The feature extraction process includes:

[0199] The feature extraction amount is actually extracted from The intensity amplitude and duration characteristics, and the physical distance ( This is converted into a mathematical phase;

[0200] ;

[0201] Where A is the preset global gain. The term is a sinusoidal oscillation. The preset angular frequency;

[0202] The sinusoidal oscillation term gives life to the rhythm of breathing. If we don't use sin and directly use... So when the user presses and holds the button... If it's a constant, the fractal pattern will freeze in a deformed state, appearing as if it's frozen, multiplied by a sinusoidal oscillation term, even if the hand is held still. The value will also be in [+ , - The fabric swayed back and forth, its texture undulating like water being constantly disturbed. The speed at which the clothing texture fluctuates is indicated and is manually set by the developer in the initial state;

[0203] Global gain is a preset constant (0.1) that represents the material's sensitivity and is used to scale... The order of magnitude, because (Distance / pressure) may be measured in meters or centimeters (e.g., 0.2), while the complex constant ($C$) in fractal mathematical formulas is very sensitive to numerical values. Usually, a change of only 0.001 can cause a dramatic change. Without a preset A to scale down, a slight touch from the user may cause the entire screen to crash (become noise). A ensures that the numerical range of the physical world can be safely mapped to the effective parameter range of the mathematical world.

[0204] Using formula An amplitude modulation mechanism was implemented, in which proximity sensing parameters were used. As a modulation signal, it controls the amplitude of the oscillation, with the following effect:

[0205] When there is no interaction ( When =0, the perturbation is zero, and the basic texture shape is preserved;

[0206] When there is interaction, the texture undergoes continuous temporal evolution (driven by sin), avoiding static stagnation;

[0207] The intensity of the interaction is directly and linearly mapped to the intensity of the visual oscillation, thus achieving intuitive force-visual feedback.

[0208] This feature extraction value is fed into Directly modify the fractal iteration formula In The core logic of feature extraction includes threshold filtering and mapping. The positive pressing component is extracted by threshold filtering through the mathematical expression of the proximity sensing parameter, and the component is mapped to the amplitude feature of a sine wave.

[0209] The dynamic fractal driving parameters explain that the parameters at different locations on the clothing (such as the collar and hem) can be different. The fundamental complex constant is the static state of the fabric. When no one touches the clothing and the environment is normal, the appearance of the mathematical fractal pattern of the clothing is determined by the fundamental complex constant. For example, the original texture of an ordinary blue pair of jeans is determined by the fundamental complex constant when no external force is applied.

[0210] Based on complex seed points and dynamic fractal driving parameters, iterative calculations are performed on each pixel in the 3D mesh space of the clothing. The mathematical expression for the iterative calculation is as follows:

[0211] ;

[0212] Where, when n is 0, These are the initial coordinates of the corresponding pixels in the 3D mesh space of the clothing. The initial coordinates are the coordinates of the points after mapping the texture coordinates using a spatial mapping function. The state of the pixel after n+1 iterations. The state of the pixel after n iterations;

[0213] Record the number of iterations and the final modulus required for each pixel to escape from the preset radius.

[0214] When the proximity sensing parameter is 0 or negative, meaning the hand is not in contact, and when the proximity sensing parameter is positive, meaning the hand is approaching and moving towards the inside of the torso, the imaginary part of the fundamental complex constant undergoes a nonlinear perturbation with sinusoidal oscillation.

[0215] Used to create complex textures Used to make the texture fluctuate dynamically.

[0216] Perturbations of complex parameters driven by proximity sensing parameters can cause drastic changes in the iteration results, producing ripple diffusion or turbulence effects on the texture centered on the contact point.

[0217] Tiny changes in proximity parameters, after dozens of iterations, can trigger dramatic changes in texture morphology (such as changing from laminar to turbulent flow). This simulates the stress response of organisms when stimulated, and is more natural and unpredictable than simple linear interpolation animation.

[0218] The process of obtaining the physical interaction increment includes the following:

[0219] Step S21: Through the physical interface of the VR or AR engine, read the environmental attributes of the coordinate area where the current virtual reality image is located in real time, and obtain environmental metadata. The environmental attributes include the normal physical area, anti-gravity area, zero gravity area and spatiotemporal turbulence area. The environmental metadata includes the gravitational acceleration vector, the damping coefficient of the current environmental medium and the defined time elapsed ratio of the current scene.

[0220] The current environmental media include air, water, and vacuum, and the time elapsed factor is defined as 1 for normal and 0.1 for bullet time;

[0221] Step S22: Based on environmental metadata, perform localized mapping of physical rules on the clothing 3D mesh space. This localized mapping includes calculating the gravitational acceleration vector of the current region in the metaverse and setting the time flow pattern.

[0222] Calculate the gravitational acceleration vector of the current region of the metaverse. When in the normal physics region, the gravitational acceleration vector of the current region of the metaverse is equal to the standard free fall acceleration.

[0223] When in the anti-gravity region, the gravitational acceleration vector of the current region of the metaverse is equal to the negative standard free fall acceleration;

[0224] When in a zero-gravity region, the gravitational acceleration vector of the current region of the metaverse is equal to 0;

[0225] Set the identifier for the time flow mode: setting the identifier to 1 indicates that time flows forward, setting the identifier to -1 indicates that time flows backward, and setting the identifier to 0 indicates that time is frozen.

[0226] The gravitational acceleration vector of the current region of the metaverse, the damping coefficient of the current environmental medium, and the defined time elapsed factor of the current scene are encapsulated into a physical rule state vector;

[0227] Step S23: Obtain the spatial interaction tensor between the previous frame and the current frame from the spatial interaction unit, and extract dynamic features through differential calculation. The extraction process includes:

[0228] Calculate the ratio of the displacement difference of the hand controller relative to the torso region to the frame interval to obtain the interaction velocity vector;

[0229] Monitor proximity sensing parameters; when the proximity sensing parameter is greater than 0, it is determined to be a contact or collision state.

[0230] When the proximity sensing parameter is continuously positive and the corresponding texture coordinate position undergoes tangential displacement, it is determined to be a dragging state.

[0231] Step S24: Based on the kinetic energy theorem, calculate the collision energy generated by the hand controller at the instant the proximity sensing parameter changes from 0 to positive. The mathematical expression for the collision energy generated by the hand controller is:

[0232] ;

[0233] in, The collision energy generated by the hand controller. The square of the modulus of the interaction speed. For interactive velocity vectors, This is a preset virtual quality coefficient;

[0234] In this embodiment, m is a preset scalar parameter that defines the weight of the user's hand interaction. To make the interaction feel responsive, it is set to a moderate value (value is set to 1).

[0235] The developers adjusted this value in the engine until the ripple effect generated when the user quickly waves their hand to touch the clothes was neither too weak nor so weak that the entire texture would instantly collapse (turn into white noise).

[0236] The moment when the proximity sensing parameter changes from 0 to positive is when the first derivative of the proximity sensing parameter with respect to time is 0, and the value of the current proximity sensing parameter at the previous instant is less than or equal to 0 and the value at the next instant is greater than 0.

[0237] The collision energy generated by the hand controller produces an instantaneous high-frequency pulse signal;

[0238] In drag mode, the drag energy is calculated, and the mathematical expression for drag energy is:

[0239] ;

[0240] in, To drag energy, The coefficient of surface friction, This represents the tangential displacement length at the texture coordinate position. This refers to the velocity component of the user's hand controller rubbing along the 3D mesh surface of the garment when it comes into contact with the garment. The initial time at the moment of contact;

[0241] In this embodiment, the developer assigns values ​​based on the material characteristics that the virtual clothing wants to simulate. For example, when simulating rough cotton or linen... It is set to 0.7 (high friction, high drag energy, producing obvious drag marks);

[0242] When simulating smooth silk or latex, Set to 0.1 (low friction, less dragging energy, smoother texture sliding);

[0243] It determines the amount of "energy" generated when a user's finger slides across the surface of clothing, which in turn affects the degree of perturbation of the fractal texture.

[0244] If you only press the clothes vertically... If it is 0, then When the value is 0, the fractal pattern will vibrate due to collisions, but will not produce the stringing or smearing effects caused by dragging.

[0245] If you gently glide it along the surface of the clothing, although the pressure is not great, It's very big, so It will be very large, and the fractal pattern will follow the fingers to create a flowing trace.

[0246] The collision energy, drag energy, and motion speed generated by the hand controller are encapsulated into a physical rule state vector;

[0247] Step S25: Merge the physical rule state vector and the interaction intensity field through channels to output the final physical interaction constraint variables, the mathematical expression of which is:

[0248] ;

[0249] in, For physical interaction constraint variables, This is the physical rule state vector. For the interaction intensity field, This is a vector concatenation symbol;

[0250] The physical rule state vector component of the physical interaction constraint variables determines the flow direction and evolution logic (with / against gravity) of the fractal texture, while the interaction intensity field component determines the perturbation amplitude and luminescence intensity of the fractal texture.

[0251] Step S26: Calculate the changes of each component of the gravitational acceleration vector and the interaction intensity field within a preset time period, and multiply the sum of the corresponding changes by the time flow rate mode to obtain the physical interaction increment term.

[0252] The driving logic for the physical interaction increment includes:

[0253] When the collision energy in the interaction intensity field exceeds the preset collision threshold, or the drag energy exceeds the preset drag threshold (when the user's hand swings or touches the object violently), the physical interaction increment term outputs a high-frequency disturbance value, causing the constant term in the iteration formula to oscillate violently, resulting in a boiling or turbulent visual effect on the fractal texture of the clothing surface. The number of iterations will increase nonlinearly with the energy intensity.

[0254] The flow direction of the texture is constrained by the gravity vector information in the physical rule state vector. In the normal physical area, the texture flows in the same direction as gravity. In the anti-gravity area, the physical interaction increment introduces a reverse phase offset, driving the texture to flow upward against gravity.

[0255] Based on the time flow pattern in the physical rule state vector, if it is in the inverse time zone, the physical interaction increment term drives the time parameter in the iterative calculation to reverse, thus realizing the time reversal effect of texture animation;

[0256] The system is no longer limited to geophysics. If the metaverse scene is switched to a zero-gravity space capsule or a time reversal zone, the flow direction (gravity term) and evolution speed (time flow term) of clothing textures will automatically adapt, enhancing the consistency of the environmental narrative. Collision and drag energy are calculated through the kinetic energy theorem, making the interaction intensity proportional to the texture feedback intensity. A light touch produces ripples, and a heavy hit produces giant waves.

[0257] This unit is the core computing engine. By introducing physical interaction constraint variables, it directly maps the physical rules in the metaverse and user interaction behavior into the driving force of mathematical fractals, realizing the leap from static textures to physically anti-physical intelligent response fabrics.

[0258] The potential field coloring unit includes:

[0259] Based on the final iteration count and final modulus of each pixel, the normalized escape velocity is calculated. The mathematical expression for the normalized escape velocity is:

[0260] ;

[0261] in, To normalize the escape velocity, For the final module length, This represents the maximum number of iterations.

[0262] By using the programmatic color palette function Palette, all normalized escape velocities are mapped to RGB color values, and the combined mapping results are used to generate a fractal base texture.

[0263] Potential field shading units eliminate stepped color banding in color transitions. The generated fractal-based texture has infinite resolution and dynamic variation characteristics, but is flat in terms of lighting and shadow.

[0264] Fractal computation typically outputs an integer number of iterations, resulting in distinct color banding. The normalized escape velocity formula introduces logarithmic smoothing, which transforms discrete integers into continuous floating-point numbers, making color transitions as smooth as silk, so that even when viewed up to the millimeter level, no pixels or color levels are visible.

[0265] Material synthesis units include:

[0266] The fractal base texture is blended with a static ambient occlusion map using Multiply blending to obtain the final output smart fabric. The mathematical expression for Multiply blending is:

[0267] ;

[0268] in, It is a smart fabric. For fractal basic texture, This is a static ambient occlusion texture.

[0269] The generated virtual clothing not only has dynamic patterns that respond in real time to the user's gaze (parallax angle) and touch (proximity sensing parameters), but also retains an extremely realistic physical fold and three-dimensional feel (provided by static ambient occlusion maps), avoiding the problem of mathematical textures being flatly applied to the model surface.

[0270] This module is the artistic core of the invention. It uses mathematical formulas to weave fabric in real time, achieving infinite resolution and dynamic interactivity, which is completely different from traditional static image mapping technology.

[0271] The fractal loom module abandons the traditional approach of storing static image files as textures in game development. Instead, it uses mathematical formulas to calculate the color of each pixel in real time on the GPU. This not only achieves infinite resolution (without blurring when viewed up close), but also gives clothing a sense of life that responds to the environment and touch.

[0272] This invention introduces procedural fractal mathematics and complex plane holographic mapping theory. Utilizing the infinite self-similarity of complex dynamics and its sensitive dependence on initial conditions, it achieves pixel-level real-time generation of virtual fabric textures by adaptively adjusting complex seed points and phase offset parameters, completely abandoning traditional high-resolution static texture mapping. By employing a phase offset function driven by the parallax angle and a bio-state iterative operator driven by the proximity field, it constructs a holographic visual illusion mechanism that flows with the viewing angle and a bio-stress mechanism that responds nonlinearly to touch intensity, respectively. This mathematically endows virtual clothing with digital vitality, effectively solving the problems of rigid textures and lack of interactivity in traditional methods. Not only does it optimize visual performance, but it also utilizes a decoupled mapping strategy of thin-plate spline interpolation and static ambient light occlusion. This ensures that the clothing automatically and perfectly fits different topological human bodies while pre-computed lighting and shadows preserve the physical-level wrinkle texture, eliminating the computational bottleneck of real-time ray tracing. The fractal driving parameters are determined by analytical calculation based on environmental metadata and interactive kinetic energy, replacing the expensive real-time cloth physics collision calculation. This greatly reduces the video memory usage and GPU computing load, enabling the dynamic dress-up interaction system with infinite resolution to run on VR all-in-one machines with limited computing power, meeting the high frame rate real-time rendering requirements of the metaverse immersive environment.

[0273] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0274] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the protection scope of the present invention.

Claims

1. A dynamic dress-up interactive system for a metaverse virtual reality avatar, characterized in that, It includes a perception module, a topology mapping module, and a fractal loom module: The perception module is used to receive virtual reality image data in the metaverse environment, construct a virtual reality image semantic field based on the virtual reality image data, calculate the parallax angle and proximity sensing parameters based on the pose data of the VR headset and hand controller, and combine the parallax angle and proximity sensing parameters to output a spatial interaction tensor. The topology mapping module is used to generate a 3D mesh space for clothing based on digital pattern atlases and virtual reality image semantic fields, using a thin plate spline interpolation algorithm, and to perform lighting baking on the 3D mesh space for clothing to generate a static ambient light occlusion map. The fractal loom module is used to map the 3D mesh space of the garment to the mathematical complex plane, calculate the dynamically changing complex seed points and complex parameters based on the spatial interaction tensor, and calculate the physical interaction increment term in combination with environmental metadata. Based on the complex seed points, complex parameters and physical interaction increment term, the fractal basic texture is generated through iterative calculation. The fractal basic texture is synthesized with the static ambient light occlusion map to obtain the smart fabric finally presented on the surface of the virtual reality image. The fractal loom module includes a holographic complex plane projection unit, a biological state iterative solution unit, a potential energy field coloring unit, and a material synthesis unit. The holographic complex plane projection unit includes: The texture coordinates of the clothing's 3D mesh space are read and mapped to a mathematical complex plane. Based on this mathematical complex plane, a complex seed point is calculated using a phase offset function according to the disparity angle. The mathematical expression for calculating the complex seed point is as follows: ； in, For complex seed points, For space mapping functions, For texture coordinates, For parallax angle, This is the phase offset function; The biological state iterative solution unit includes: A preset basic complex constant is established, and proximity sensing parameters are monitored in real time. Based on the influence rules of the dynamically changing proximity sensing parameters on the basic complex constant, dynamic complex parameters are obtained. The influence rules include: When the proximity sensing parameter is 0 and negative, the fundamental complex constant remains unchanged; When the proximity sensing parameter is positive, the imaginary part of the fundamental complex constant undergoes sinusoidal oscillation; Obtain environmental metadata of the metaverse, and construct physical interaction constraint variables based on environmental metadata and spatial interaction tensor. The physical interaction constraint variables include physical rule state vector and interaction intensity field. The physical rule state vector includes the gravitational acceleration vector of the current region of the metaverse, the damping coefficient of the current environmental medium, and the time flow pattern. The interaction intensity field includes the collision energy, dragging energy, and object motion speed generated by the hand controller. Calculate the physical interaction increment term based on the physical interaction constraint variables; The dynamic fractal driving parameters are calculated based on dynamic complex parameters and physical interaction increment terms. The mathematical expression for the dynamic fractal driving parameters is as follows: ； in, For dynamic fractal driving parameters, For dynamic complex parameters, The feature extraction quantity represents the rule governing the influence of dynamically changing proximity sensing parameters on the underlying complex constant. For physical interaction increments, For physical interaction constraint variables, Texture coordinates , For time; Based on complex seed points and dynamic fractal driving parameters, iterative calculations are performed on each pixel in the 3D mesh space of the clothing. The mathematical expression for the iterative calculation is as follows: ； Where, when n is 0, These are the initial coordinates of the corresponding pixel points in the 3D mesh space of the clothing. These initial coordinates are the coordinates obtained by mapping the texture coordinates using a spatial mapping function. The state of the pixel after n+1 iterations. The state of the pixel after n iterations; Record the number of iterations and the final modulus required for each pixel to escape from the preset radius.

2. The dynamic dress-up interactive system for metaverse virtual reality avatars as described in claim 1, characterized in that, The perception module includes a data acquisition unit, a virtual reality image semantic field unit, and a spatial interaction unit: The data acquisition unit includes: The virtual reality character data of the current frame is obtained from the environment in which the metaverse VR is running. The virtual reality character data includes skeletal pose data and surface mesh data. The skeletal posture data includes the real-time first three-dimensional position vector coordinates and first quaternion rotation data of key joints in the world coordinate system. The key joints include the cervical spine, shoulder joint, elbow joint, wrist joint, hip joint, knee joint, and ankle joint. The first three-dimensional position vector coordinates and first quaternion rotation data of all key joints are stored in matrix form to form a set of current posture transformation matrices of the virtual reality image. All bone pose data are deformed in real time using a linear blending skinning algorithm to obtain surface mesh data, which includes vertex coordinates, normal vectors, and texture coordinates. The surface mesh data is stored in video memory in the form of vertex buffers and index buffers.

3. The dynamic dress-up interactive system for metaverse virtual reality avatars as described in claim 2, characterized in that, The virtual reality image semantic field unit includes: The construction process of building a semantic field for virtual reality images based on virtual reality image data includes: During the import phase of the virtual reality avatar, semantic tags are predefined for each component of the virtual reality avatar. The semantic tags include the torso, left hand, right hand, head, and hair. The semantic tags include material ID information and vertex color information. Off-screen rendering is performed on the virtual reality image data. The off-screen rendering process includes: The virtual reality image is rendered by disabling lighting and textures, and only assigning a unique color to the corresponding component based on predefined semantic tags. The result of the secondary rendering is stored as a semantic ID texture. The semantic ID texture is post-processed to extract and generate torso mask, hand mask, head mask and hair mask respectively. The torso mask, hand mask, head mask and hair mask correspond to each pixel region represented in the form of binary mask texture under the current VR view. Each pixel region includes torso region, hand region, head region and hair region. The set of torso mask, hand mask, head mask, and hair mask, which are represented in the form of binary mask textures, is defined as the virtual reality image semantic field.

4. The dynamic dress-up interactive system for metaverse virtual reality avatars as described in claim 3, characterized in that, The spatial interaction unit includes: The pose data of the VR headset and the user's hand controller in the world coordinate system are obtained in real time from the SDK interface of the VR device. The pose data of the VR headset includes the real-time second three-dimensional position vector coordinates and the second quaternion rotation data of the VR headset. The pose data of the hand controller includes the real-time third three-dimensional position vector coordinates of the hand controller. Using the pose data of the VR headset when the user begins to observe the virtual reality image as the initial reference pose data, the matrix product of the current VR headset's second quaternion rotation data and the second quaternion rotation data in the initial reference pose data is calculated to obtain the relative second quaternion rotation matrix. The relative second quaternion rotation matrix is ​​then converted into Euler angles to obtain the current VR headset's rotation increment. The three angular components of the Euler angles include pitch angle, yaw angle, and roll angle. Calculate the first displacement difference between the current VR headset's second three-dimensional position vector coordinates and the second three-dimensional position vector coordinates in the initial reference pose data, and obtain the current VR headset's displacement increment; The displacement increment and rotation increment are combined into a parallax angle; The Boundary Volume Hierarchy (BVH) algorithm is used to optimize the search of the torso region and calculate the minimum Euclidean distance from the pose data of the hand controller to the surface mesh data corresponding to the torso region. Obtain the nearest first three-dimensional position vector coordinates and the normal vector corresponding to the nearest first three-dimensional position vector coordinates from the surface mesh data corresponding to the torso region to the pose data of the hand controller; calculate the second vector difference between the pose data of the hand controller and the nearest first three-dimensional position vector coordinates. Calculate the projection value of the second vector difference onto the direction of the normal vector corresponding to the nearest first three-dimensional position vector coordinates. The projection value is the dot product of the second vector difference and the normal vector corresponding to the nearest first three-dimensional position vector coordinates. The proximity sensing parameters between the hand region and the torso region are calculated, and the mathematical expression for the proximity sensing parameters is as follows: ； in, For proximity sensing parameters, This represents the minimum Euclidean distance from the pose data of the hand controller to the surface mesh data corresponding to the torso region. The preset sensing radius, The second vector difference, The normal vector corresponding to the nearest three-dimensional position vector coordinates. This refers to the pressure sensitivity coefficient. The disparity angle and proximity sensing parameters are combined into a spatial interaction tensor.

5. The dynamic dress-up interactive system for metaverse virtual reality avatars as described in claim 4, characterized in that, The topology mapping module includes a mesh redirection unit and a lighting baking unit: The mesh redirection unit is used to read the digital pattern atlas of the garment, calculate the transformation matrix using the thin-plate spline interpolation algorithm TPS, and map the vertex coordinates of the digital pattern atlas to the geometric topological surface of the current virtual reality image to generate a suitable 3D mesh space for the garment. The generation process includes: Preset corresponding point groups of 2D digital template atlas and 3D virtual reality image surface mesh data as constraint points; Based on constraint points, the transformation matrix is ​​calculated using the thin plate spline interpolation algorithm TPS. All vertices in the digital pattern atlas are mapped to the clothing 3D mesh space through the transformation matrix. The vertex connection relationship in the clothing 3D mesh space is completely consistent with the vertex connection relationship in the digital pattern atlas. The lighting baking unit is used to separate the ambient occlusion channel in the rendering pipeline, convert the clothing 3D mesh space to the perceptual color space, extract the luminance component L, and generate a static ambient occlusion map that retains physical wrinkle information. The generation process includes: In the rendering pipeline, a third rendering is performed on the clothing 3D mesh space. The third rendering includes calculating the number of beams of diffuse light from the surrounding environment that each point in the clothing 3D mesh space can receive, and outputting an ambient occlusion map, which is a grayscale image. Convert the ambient occlusion map to the Lab color space, discard the A and B channels, keep only the luminance channel, and output a static ambient occlusion map.

6. The dynamic dress-up interactive system for metaverse virtual reality avatars as described in claim 5, characterized in that, The process of obtaining the physical interaction increment specifically includes: Step S21: Through the physical interface of the VR or AR engine, read the environmental attributes of the coordinate area where the current virtual reality image is located in real time and obtain environmental metadata. The environmental attributes include normal physical area, anti-gravity area, zero gravity area and spatiotemporal turbulence area. The environmental metadata includes gravity acceleration vector, damping coefficient of the current environmental medium and the defined time elapsed ratio of the current scene. The current environmental media include air, water, and vacuum, and the time elapsed factor is defined as 1 for normal and 0.1 for bullet time; Step S22: Based on environmental metadata, perform localized mapping of physical rules on the clothing 3D mesh space. This localized mapping includes calculating the gravitational acceleration vector of the current region in the metaverse and setting the time flow pattern. Calculate the gravitational acceleration vector of the current region of the metaverse. When in the normal physics region, the gravitational acceleration vector of the current region of the metaverse is equal to the standard free fall acceleration. When in the anti-gravity region, the gravitational acceleration vector of the current region of the metaverse is equal to the negative standard free fall acceleration; When in a zero-gravity region, the gravitational acceleration vector of the current region of the metaverse is equal to 0; Set the identifier for the time flow mode: setting the identifier to 1 indicates that time flows forward, setting the identifier to -1 indicates that time flows backward, and setting the identifier to 0 indicates that time is frozen. The gravitational acceleration vector of the current region of the metaverse, the damping coefficient of the current environmental medium, and the defined time elapsed factor of the current scene are encapsulated into a physical rule state vector; Step S23: Obtain the spatial interaction tensor between the previous frame and the current frame from the spatial interaction unit, and extract dynamic features through differential calculation. The extraction process includes: Calculate the ratio of the displacement difference of the hand controller relative to the torso region to the frame interval to obtain the interaction velocity vector; Monitor proximity sensing parameters; when the proximity sensing parameter is greater than 0, it is determined to be a contact or collision state. When the proximity sensing parameter is continuously positive and the corresponding texture coordinate position undergoes tangential displacement, it is determined to be a dragging state. Step S24: Based on the kinetic energy theorem, calculate the collision energy generated by the hand controller at the instant the proximity sensing parameter changes from 0 to positive. The mathematical expression for the collision energy generated by the hand controller is: ； in, The collision energy generated by the hand controller. The square of the modulus of the interaction speed. For interactive velocity vectors, This is a preset virtual quality coefficient; In drag mode, the drag energy is calculated, and the mathematical expression for the drag energy is: ； in, To drag energy, The coefficient of surface friction, This represents the tangential displacement length at the texture coordinate position. This refers to the velocity component of the user's hand controller rubbing along the 3D mesh surface of the garment when it comes into contact with the garment. The initial time at the moment of contact; The collision energy, drag energy, and motion speed generated by the hand controller are encapsulated into a physical rule state vector; Step S25: Merge the physical rule state vector and the interaction intensity field through channels to output the final physical interaction constraint variables, the mathematical expression of which is: ； in, For physical interaction constraint variables, This is the physical rule state vector. For the interaction intensity field, This is a vector concatenation symbol; Step S26: Calculate the changes of each component of the gravitational acceleration vector and the interaction intensity field within a preset time period, and multiply the sum of the corresponding changes by the time flow rate mode to obtain the physical interaction increment term.

7. The dynamic dress-up interactive system for metaverse virtual reality avatars as described in claim 6, characterized in that, The potential energy field coloring unit includes: Based on the final iteration count and final modulus of each pixel, the normalized escape velocity is calculated. The mathematical expression for the normalized escape velocity is: ； in, To normalize the escape velocity, For the final module length, This represents the maximum number of iterations. By using the programmatic color palette function Palette, all normalized escape velocities are mapped to RGB color values, and the combined mapping results are used to generate a fractal base texture.

8. The dynamic dress-up interactive system for metaverse virtual reality avatars as described in claim 7, characterized in that, The material synthesis unit includes: The fractal base texture is blended with a static ambient occlusion map using Multiply blending to obtain the final output smart fabric. The mathematical expression for Multiply blending is: ； in, It is a smart fabric. For fractal basic texture, This is a static ambient occlusion texture.

Images