Body accessory automatic adaptation method for personalized body type of digital human

Abstract

The invention discloses a body accessory automatic adaptation method for a personalized body type of a digital human, and the method comprises the steps: constructing an SMPL human body model, setting a standard SMPL human body model in a reference form, and generating a target SMPL human body model according to the change state of a set target body type; making a corresponding body accessory model based on the standard SMPL human body model; constructing an RBF (Radial Basis Function) interpolation function according to the displacement change of the vertexes of the related parts of the human body from the standard SMPL human body model to the target SMPL human body model, inputting the coordinates of the vertexes of the body accessory model into the interpolation function to obtain new vertex coordinates of the body accessory model, and outputting the new vertex coordinates of the body accessory model to the target SMPL human body model. The body accessory model and the skeleton joint points are matched with the target SMPL human body model. According to the method provided by the invention, automatic adaptation of the body accessory and the digital human body shape change is efficiently and accurately realized, and the sense of reality and interactivity of the digital human are effectively improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of 3D digital humans, and specifically relates to a method for automatically adapting body accessories to the personalized body shape of a digital human. Background Art

[0002] With the booming development of digital technologies in many fields such as film and television, games, virtual reality, and the metaverse, digital humans have increasingly become core elements. Users have higher and higher requirements for the realism, personalized customization, and real-time interactivity of digital humans. In the process of constructing digital humans, the adaptation of body accessories to the body shape of digital humans is a key problem. Traditionally, when the body shape of a digital human needs to be changed, for example, from a standard body shape to a fatter or thinner body shape, the adjustment of body accessories such as clothing, hairpieces, bone joint points, and oral models mostly relies on cumbersome manual operations or semi-automated processes. This is not only inefficient, consuming a large amount of human and time costs, but also extremely likely to result in poor adaptation effects. Clothing may show unnatural stretching, ill-fitting wrinkles, or penetration phenomena; hairpieces cannot fit the new contour of the head and appear obtrusive; if the bone joint points cannot be accurately adapted to the body shape change, it will cause the digital human to have stiff and uncoordinated movements, or even distorted limbs; if the oral model does not fit well with the face, it will show dislocation and a lack of realism during facial expression changes, seriously affecting the visual perception and user experience of the digital human, and hindering the in-depth application of digital human technology in various industries.

[0003] Therefore, there is an urgent need for an efficient, accurate, and automated method to solve the problem of adapting body accessories to the body shape change of digital humans. Summary of the Invention

[0004] Aiming at the above-mentioned drawbacks, the present invention aims to provide a method for automatically adapting body accessories such as clothing models, hairpiece models, model bone joint points, and oral models to the body shape change of digital humans based on the SMPL (Skinned Multi-Person Linear) human model and using the Radial Basis Function (RBF) interpolation technology, thereby improving the realism and interactivity of digital humans.

[0005] According to the above purpose, in the first aspect of the present invention, a method for automatically adapting body accessories to the personalized body shape of a digital human is provided, and the method includes:

[0006] Construct an SMPL human model, set a standard SMPL human model in the reference state, and generate a target SMPL human model according to the set change state of the target body shape;

[0007] Make corresponding body accessory models based on the standard SMPL human model;

[0008] Construct an RBF interpolation function based on the displacement changes of the vertices of the relevant parts of the human body from the standard SMPL human model to the target SMPL human model. Input the vertex coordinates of the body attachment model into the interpolation function to obtain the new vertex coordinates of the body attachment model, so as to achieve the adaptation of the body attachment model and the bone joints to the target SMPL human model.

[0009] Further, the body attachment model includes a clothing model, a hairpiece model, and an oral cavity model, and the relevant parts of the human body include the body part, the head, the bone joints, and the face.

[0010] Further, the process of the clothing model following the deformation of the human body and adapting to the target SMPL human model includes: According to the set position relationship between the vertices of the clothing model and the standard SMPL human model, construct a first interpolation function based on RBF through the displacement changes of the body vertices from the standard SMPL human model to the target SMPL human model. Input the vertex coordinates of the clothing model into the first interpolation function to obtain the vertex coordinates of the new clothing model, and achieve the adaptation of the clothing model to the target SMPL human model.

[0011] Further, the process of the hairpiece model following the deformation of the human body and adapting to the target SMPL human model includes: Construct a second interpolation function based on RBF through the displacement changes of the head vertices from the standard SMPL human model to the target SMPL human model. Input the vertex coordinates of the hairpiece model into the second interpolation function to obtain the vertex coordinates of the new hairpiece model, and achieve the adaptation of the hairpiece model to the target SMPL human model.

[0012] Further, the process of the bone joints following the deformation of the human body and adapting to the target SMPL human model includes: According to the bone hierarchy of the standard SMPL human model and the position relationship between the coordinate positions of each bone joint and the surrounding adjacent human model vertices, construct a third interpolation function based on RBF through the displacement changes of the adjacent human model vertices of the bone joints from the standard SMPL human model to the target SMPL human model, and obtain the coordinates of the new bone joints, so as to achieve the adaptation of the bone joints to the target SMPL human model.

[0013] Further, the process of the oral cavity model following the deformation of the human body and adapting to the target SMPL human model includes: Construct a fourth interpolation function based on RBF according to the displacement changes of the facial vertices from the standard SMPL human model to the target SMPL human model. Input the vertex coordinates of the oral cavity model into the fourth interpolation function to obtain the vertex coordinates of the new oral cavity model, and achieve the adaptation of the oral cavity model to the target SMPL human model.

[0014] Further, the selected facial vertices are the vertices adjacent to the periphery of the mouth.

[0015] Further, the standard SMPL human model is the model state when both the shape parameters and pose parameters of the model are equal to zero.

[0016] In a second aspect of the present invention, there is provided a computer storage medium storing a computer program thereon, and when the computer program is run by a processor, it is configured to execute the automatic adaptation method as described in the first aspect above.

[0017] The automatic adaptation method for body accessories of a digital human's personalized body shape disclosed in the present invention achieves highly automated and precise adaptation effects. It abandons the inefficient way of traditional manual adaptation, and the entire process is based on advanced algorithms and model construction to build an automated adaptation process, greatly saving labor and time costs. The adaptation work that originally took several hours or even days is shortened to several minutes, greatly improving the creation and application efficiency of digital humans. At the same time, relying on the accuracy of the SMPL human model and the powerful fitting ability of the RBF interpolation algorithm, it deeply considers the characteristics of each body accessory, and realizes the precise synchronization of clothing, hairpieces, bone joint points, oral models, etc. with the body shape changes of the digital human, avoiding adaptation defects, and effectively improving the realism and visual quality of the digital human. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the standard SMPL human model in an embodiment of the present invention.

[0019] FIG. 2(a) is a schematic diagram of the first target SMPL human model in an embodiment of the present invention.

[0020] FIG. 2(b) is a schematic diagram of the second target SMPL human model in an embodiment of the present invention.

[0021] FIG. 3(a) is a schematic diagram of body accessories and bone movements adapted according to the standard SMPL human model in an embodiment of the present invention.

[0022] FIG. 3(b) is a schematic diagram of body accessories and bone movements adapted according to the first target SMPL human model in an embodiment of the present invention.

[0023] FIG. 3(c) is a schematic diagram of body accessories and bone movements adapted according to the second target SMPL human model in an embodiment of the present invention.

[0024] Figure 4 It is a schematic flow chart of the automatic adaptation method for body accessories of a digital human's personalized body shape in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not used to limit the present invention.

[0026] It should be noted that the illustrations provided in this embodiment only schematically illustrate the basic concept of the present invention. The structures, ratios, sizes, etc. shown in the drawings of this specification are only used to cooperate with the content disclosed in the specification for those skilled in this technology to understand and read, and are not used to limit the limiting conditions under which the present invention can be implemented. Any modification of the structure, change of the proportional relationship or adjustment of the size, without affecting the effects that the present invention can produce and the objectives that can be achieved, should still fall within the scope that can be covered by the technical content disclosed in the present invention.

[0027] The orientation or positional relationship indicated by terms such as "front", "rear", "left", "right", "middle", "longitudinal", "transverse", "horizontal", "inner", "outer", etc. cited in this specification is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus cannot be construed as a limitation of the present invention. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance.

[0028] An embodiment of the present invention provides a method for automatically adapting body attachments to the personalized body shape of a digital human, which can automatically and accurately adapt the body attachments to the body shape changes of the digital human.

[0029] Refer to Figures 1 to 4 As shown, a method for automatically adapting body attachments to the personalized body shape of a digital human disclosed in an embodiment of the present invention specifically includes the following processes:

[0030] The first step is the construction of the SMPL human model.

[0031] The SMPL (Skinned Multi-Person Linear) human body model was published at SIGGRAPH Asia in 2015 under the leadership of Michael J. Black of the Max Planck Institute. It is a skeleton-driven parametric human body model. The human body can be understood as the sum of a basic model and the deformations on this model. On the basis of the deformations, PCA is performed to obtain low-dimensional parameters that characterize the shape - shape parameters (shape). At the same time, a kinematic tree is used to represent the pose of the human body, that is, the rotation relationship between each joint point and its parent node in the kinematic tree, which can be represented as a three-dimensional vector. Eventually, the local rotation vectors of each joint point constitute the pose parameters (pose) of the SMPL model. The SMPL model is obtained through training and is defined by the function M(θ, β). Among them, β is the shape parameter (shape parameters), θ is the pose parameter (pose parameters). β represents 10 - dimensional parameters such as the height, weight, and head - to - body ratio of the human body, which can control the shape change of the human body. θ represents a total of 75 parameters including the overall movement pose of the human body and the relative angles of 24 joints.

[0032] Step 2: Initialize the standard SMPL human body model.

[0033] Load the SMPL human body model described by the average vertex set and weights as the reference form, that is, the model state where both the shape parameter β and the pose parameter θ are equal to zero, as Figure 1 shown in the model state with a T - pose of a person with a height of 165 cm and a weight of 50 kg as the reference.

[0034] Step 3: Generate the target SMPL human body model.

[0035] Based on the standard form of the SMPL human body model, set the target body shape state according to the shape parameter β, and use the changed model as the target form. As shown in Figure 2, Figure 2(a) shows the human body state of a person with a height of 165 cm and a weight of 70 kg, that is, the first target SMPL human body model. Figure 2(b) shows the human body model state of a person with a height of 170 cm and a weight of 50 kg, that is, the second target SMPL human body model. They are Figure 1 shown in the reference form and have changes in height or weight, resulting in changes in the human body shape.

[0036] Step 4: Construct a body shape deformation model based on RBF interpolation

[0037] To achieve displacement interpolation, the present invention uses Radial Basis Function (RBF). The radial basis function interpolation method has the advantages of high numerical accuracy and independence from grid topology, and is a general interpolation method suitable for realizing bidirectional displacement transfer on arbitrary grid topologies.

[0038] The basic form of radial basis function interpolation is:

[0039]

[0040] where F(x) is the interpolation function, N is the total number of radial basis functions (total number of control points) used in the interpolation problem, is the general form of the adopted radial basis function, ||γ - γ i || is the Euclidean distance between two position vectors, γ i is the position of the support point of the i-th radial basis function, and ω i is the weight coefficient corresponding to the i-th radial basis function. There are many types of radial basis functions. For grid deformation, Wendland's C 2 function is often used. The interpolation conditions of the radial basis function can be described in matrix form. After solving for the weight coefficients, the displacements of any grid nodes within the computational domain can be obtained. It should be noted that the interpolation principle of the radial basis function is prior art and will not be elaborated here.

[0041] The present invention uses the reference SMPL human model as the source and the SMPL model corresponding to the transformed target body shape as the target to construct an RBF interpolation model architecture and obtain the interpolation function F(x). RBF interpolation effectively connects the source model and the target model with its excellent local approximation performance and fitting ability for complex deformations. Considering factors such as the smoothness of the Gaussian function and the adaptability of the thin plate spline function to large deformations, the optimal type of radial basis function is selected according to the type of body attachment to be adapted and the different body parts, ensuring that the interpolation model accurately captures all-round geometric deformation details from the standard body shape to the target body shape, and constructs a continuous, stable and high-precision RBF interpolation function for deformation mapping, providing accurate guidance for body attachment adaptation.

[0042] Step 5, the intelligent adaptation process of body attachments.

[0043] First, the system provides a set of body attachment models made based on the standard SMPL human model. These body attachment models include clothing models, hairpiece models, oral models, and accurate bone bindings. These body attachment models, as model materials for adapting to various body shapes, can all be directly extracted from the model library of the brother.

[0044] Next, the adaptation processes of the clothing model, hairpiece model, oral model, and skeletal joint points will be described in detail.

[0045] (1) Clothing model adaptation:

[0046] Based on the principle of vertex mapping and deformation conduction, the clothing model pre-worn on the standard SMPL human model is deformed into a clothing model adapted to the target human model's body shape. According to the relationship between the clothing model and the SMPL standard human model, the positional relationship between the vertices of the clothing model set by the system and the vertices of the standard SMPL human model, that is, the relative distance and direction, through the displacement change of the selected body vertices from the standard body shape to the target body shape, constructs the first interpolation function F1(x) based on RBF. Using RBF interpolation calculation, the coordinates of each vertex of the clothing model are input into this first interpolation function F1(x) to obtain the new coordinate positions of the vertices of the clothing model, completing the deformation of each vertex of the SMPL target human model and seamlessly mapping it to the corresponding vertices of the clothing model, realizing the clothing following the human body deformation and obtaining a clothing model adapted to the target human model.

[0047] (2) Hairpiece model adaptation

[0048] Adopt the method of fitting the head contour. Through the displacement change of the selected head vertices from the standard body shape to the target body shape, construct the second interpolation function F2(x) based on RBF. Using RBF interpolation calculation, the coordinates of each vertex of the hairpiece model are input into this second interpolation function F2(x) to obtain the new coordinate positions of the vertices of the hairpiece model, realizing the precise adjustment of the root position of the hairpiece model, the hair strand direction, and the overall coverage range of the hairpiece, making the hairpiece closely fit the changed head contour and avoiding problems such as the hairpiece being suspended or having too large a gap with the scalp.

[0049] (3) Adaptation of skeletal joint points

[0050] Model bone binding is a key link in 3D animation production. Model bone binding is the process of associating a virtual bone system with the mesh vertices of a 3D model. Through this association, the movement of the bones can drive the deformation of the model mesh, enabling the model to present natural and smooth animation effects, allowing the originally static model to make various actions and expressions. The present invention is based on the method of recalculating joint point positions. According to the fine bone hierarchical structure of the SMPL standard human model and the positional relationship between the coordinate positions of each bone joint point and the surrounding adjacent human model vertices, through the displacement change of the human model vertices adjacent to the bone joint points from the standard body shape to the target body shape, construct the third interpolation function F3(x) based on RBF. Using RBF interpolation calculation, the coordinates of the bone joint points are input into this third interpolation function F3(x) to obtain the new coordinate positions of the bone joint points and ensure the consistency of the bone system and the overall deformation of the human body.

[0051] (4) Oral model adaptation

[0052] The adaptation method based on spatial positioning focuses on the coordinated changes of the oral model and the face of the SMPL human model. According to the displacement changes of facial vertices from the standard body shape to the target body shape, especially the changes of vertices of the oral model adjacent to the mouth of the SMPL human model, the fourth interpolation function F4(x) based on RBF is constructed. The coordinates of the vertices of the oral model are input into the fourth interpolation function F4(x) using RBF interpolation calculation to obtain a new coordinate position of the vertices of the oral model, so that the position of the oral model perfectly matches the overall deformation of the face, and the misalignment of the oral cavity and the face is eliminated.

[0053] The present invention uses the same set of clothing, hair, oral models and skeletal movements based on the standard SMPL standard body shape, and automatically adapts to different body shapes of SMPL human models through RBF interpolation. The result example is shown in Figure 3. The model shown in Figure 3 (a) on the far left is the standard SMPL human model and the clothing, hair, oral model and skeletal movements made by the modeler. The two models shown in Figures 3 (b) and 3 (c) on the right are examples of the results after automatic adaptation of clothing, hair, oral models and skeletal movements under target human models of different weights and heights, namely, the schematic diagram of the body accessories and skeletal movements adapted to the first target SMPL human model and the schematic diagram of the body accessories and skeletal movements adapted to the second target SMPL human model. It can be seen from the accompanying drawings that the automatic adaptation method based on the present invention is more realistic and accurate in effect and quality.

[0054] The method for automatically adapting the body accessories of a digital human's personalized body shape provided by the embodiment of the present invention abandons the inefficient traditional manual adaptation method, and builds an automated adaptation process based on advanced algorithms and models throughout the process, which greatly saves manpower and time costs, shortens the adaptation work that originally took several hours or even days to several minutes, and greatly improves the efficiency of digital human creation and application. In addition, relying on the accuracy of the SMPL human body model and the powerful fitting ability of the RBF interpolation algorithm, the characteristics of each body accessory are deeply considered to achieve accurate synchronization of clothing, hair pieces, bone joints, oral models, etc. with the changes in the digital human body shape, avoiding adaptation defects, and effectively improving the realism and visual quality of the digital human.

[0055] Another embodiment of the present invention further provides a computer storage medium on which a computer program is stored, wherein the computer program is executed by a processor to implement the automatic adaptation method described in the above embodiment.

[0056] If the above functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs.

[0057] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0058] The above-described embodiments merely represent several implementation manners of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. 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 invention patent should be subject to the appended claims.

Claims

1. An automatic adaptation method for body attachments of a personalized body shape of a digital human, characterized in that, The method includes: Construct an SMPL human body model, set the standard SMPL human body model in the reference form, and generate a target SMPL human body model according to the change state of the set target body shape; Produce a corresponding body accessory model based on the standard SMPL human body model; Construct an RBF interpolation function according to the displacement change of the vertices of the relevant human body parts from the standard SMPL human body model to the target SMPL human body model, input the vertex coordinates of the body accessory model into the interpolation function, and obtain the new vertex coordinates of the body accessory model, so as to realize that the body accessory model and the bone joints follow the human body deformation and adapt to the target SMPL human body model.

2. The automatic adaptation method according to claim 1, wherein The body accessory model includes a clothing model, a hairpiece model, and an oral cavity model, and the relevant human body parts include the body part, the head, the bone joints, and the face.

3. The automatic adaptation method according to claim 2, wherein The clothing model following the human body deformation includes: according to the set position relationship between the vertices of the clothing model and the standard SMPL human body model, construct a first interpolation function based on RBF through the displacement change of the body vertices from the standard SMPL human body model to the target SMPL human body model, input the vertex coordinates of the clothing model into the first interpolation function, and obtain the vertex coordinates of the new clothing model, so as to realize the adaptation of the clothing model to the target SMPL human body model.

4. The automatic adaptation method according to claim 2, characterized in that The hairpiece model following the human body deformation includes: construct a second interpolation function based on RBF through the displacement change of the head vertices from the standard SMPL human body model to the target SMPL human body model, input the vertex coordinates of the hairpiece model into the second interpolation function, and obtain the vertex coordinates of the new hairpiece model, so as to realize the adaptation of the hairpiece model to the target SMPL human body model.

5. The automatic adaptation method according to claim 2, wherein The bone joints following the human body deformation includes: according to the bone hierarchy of the standard SMPL human body model and the position relationship between the coordinate positions of each bone joint and the surrounding adjacent human body model vertices, construct a third interpolation function based on RBF through the displacement change of the bone joint adjacent human body model vertices from the standard SMPL human body model to the target SMPL human body model, and obtain the coordinates of the new bone joints, so as to realize the adaptation of the bone joints to the target SMPL human body model.

6. The automatic adaptation method according to claim 2, wherein The oral cavity model following the human body deformation includes: construct a fourth interpolation function based on RBF according to the displacement change of the face vertices from the standard SMPL human body model to the target SMPL human body model, input the vertex coordinates of the oral cavity model into the fourth interpolation function, and obtain the vertex coordinates of the new oral cavity model, so as to realize the adaptation of the oral cavity model to the target SMPL human body model.

7. The automatic adaptation method according to claim 6, wherein The selected face vertices are the vertices adjacent to the mouth periphery.

8. The automatic adaptation method according to claim 1, wherein The standard SMPL human body model is the model state when both the shape parameter and the pose parameter of the model are equal to zero.

9. A computer storage medium having a computer program stored thereon, characterized in that, When the computer program is run by a processor, it is used to execute the method for automatically adapting the body accessories of the personalized body shape of the digital human as described in any one of claims 1 to 8 above.

Images