Digital human model-wearing detection method and device

By constructing a directed distance field for each limb part of a digital person, and automatically detecting the mold penetration between limb parts, it solves the problem of time-consuming, labor-intensive and error-prone problems in the prior art, and realizes efficient and accurate mold penetration detection. Improves the realism and visual effect of the digital human model.

CN118918094APending Publication Date: 2024-11-08ALIPAY (HANGZHOU) INFORMATION TECH CO LTD
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202411103710.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-12
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

The existing digital human-wired model inspection mainly relies on manual visual inspection, which is time-consuming and labor-intensive and prone to omissions and errors, and cannot achieve automated and accurate model inspection.

Method used

By constructing a directed distance field for each limb of digital human, the mode of penetration between limbs is calculated based on the coordinates and joint transformation data of surface vertices, and the mode of penetration between limbs is automatically detected.

Benefits of technology

It realizes the automation and accuracy of digital human model inspection, simplifies the inspection process, timely discovers and solves the model inspection problem, and improves the reality and visual effect of the digital human model.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118918094A_ABST
    Figure CN118918094A_ABST
Patent Text Reader

Abstract

The embodiment of the invention provides a digital human model-crossing detection method and device. The method comprises the following steps: constructing a directed distance field of each limb part according to a first coordinate of a surface vertex of each limb part of the digital human in a current posture; for a first limb part of the digital human, determining a first model penetration degree of the first limb part relative to a second limb part according to a value of a directed distance field of a surface vertex of the first limb part relative to the second limb part; and determining whether the first limb part and the second limb part are subjected to mold penetration or not according to the first mold penetration degree.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The embodiments of this specification relate to the field of computer technology, and in particular, to a method and device for detecting digital human penetration. Background Art

[0002] With the development of technology and computer graphics, virtual humans are widely used in video games, film and television production, virtual reality and other fields. The realistic performance of digital humans is crucial to improving user experience and visual effects. However, in the process of designing digital human clothing and movements, due to various factors, there may be a phenomenon of penetration between digital humans and clothing or in movements.

[0003] Model penetration refers to the situation where the clothes worn by the digital human appear to be visible through the model in some actions or postures, resulting in visual unreality and damaging the user experience. At present, the detection of model penetration of digital humans mainly relies on manual visual inspection and experience judgment, which is not only time-consuming and labor-intensive, but also prone to omissions and errors.

[0004] Therefore, a reasonable and reliable solution is needed to automatically perform efficient and accurate mold penetration detection on digital humans. Summary of the invention

[0005] The embodiments of this specification provide a digital human penetration detection solution, which can automatically perform efficient and accurate penetration detection on the digital human.

[0006] In a first aspect, an embodiment of the present specification provides a method for detecting penetration of a digital human, comprising: constructing a signed distance field for each limb part of the digital human according to the first coordinates of the surface vertices of each limb part of the digital human in the current posture; for a first limb part of the digital human, determining a first penetration degree of the first limb part with respect to the second limb part according to the value of the signed distance field of the surface vertices of the first limb part with respect to the second limb part; and determining whether the first limb part penetrates the second limb part according to the first penetration degree.

[0007] In some embodiments, before constructing the signed distance fields of the respective limb parts, it also includes: obtaining the vertex weight distribution of the digital human, the first joint transformation data and the vertex coordinate data in the reference posture, and the second joint transformation data in the current posture; the first joint transformation data and the second joint transformation data both include the joint transformation matrices of the multiple joints of the digital human; the vertex weight distribution includes the weight distribution of each surface vertex of the digital human with respect to the associated joints; and determining the first coordinates of each surface vertex of the digital human in the current posture according to the first joint transformation data, the vertex coordinate data, the second joint transformation data and the vertex weight distribution.

[0008] In some embodiments, the determining of the first coordinates of each surface vertex of the digital human in the current posture includes: for each target vertex among the surface vertices of the digital human, according to the second coordinate of the target vertex in the reference posture, and the first joint transformation matrix of each associated joint in the reference posture and the second joint transformation matrix in the current posture, determining the third coordinate of the target vertex relative to each of the joints; according to the weight distribution of the target vertex with respect to the joints, performing weighted summation on the third coordinates of the target vertex to obtain the first coordinate of the target vertex in the current posture.

[0009] In some embodiments, determining the third coordinate of the target vertex relative to each of the joints includes: determining the relative coordinates of the target vertex relative to the local coordinate systems of each of the joints based on the first joint transformation matrix and the second coordinates; determining the third coordinates of the target vertex relative to each of the joints based on the second joint transformation matrix and the relative coordinates determined for the target vertex.

[0010] In some embodiments, each surface vertex of the digital human is associated with a number of joints, and a weight distribution of the associated joints is preset; before constructing the signed distance fields of each limb part, it also includes: for each surface vertex of the digital human, determining the limb part to which the joint corresponding to the maximum weight value in the weight distribution of the surface vertex belongs, and dividing the surface vertex into the limb part.

[0011] In some embodiments, determining the first penetration of the first limb part with respect to the second limb part includes: for each surface vertex of the first limb part, determining the second penetration of the surface vertex with respect to the second limb part according to the value of the directed distance field of the surface vertex with respect to the second limb part; and determining the average value of the second penetration of the surface vertices of the first limb part as the first penetration.

[0012] In some embodiments, the method also includes: determining whether each surface vertex of the first limb part penetrates with the second limb part based on the second penetration degree of each surface vertex of the first limb part; determining the penetration ratio of the first limb part with respect to the second limb part based on the first number of surface vertices in the first limb part that penetrate with the second limb part, and the second number of each surface vertices of the first limb part.

[0013] In some embodiments, for any surface vertex of the first limb part, when the surface vertex is located inside the second limb part, the value of the signed distance field of the surface vertex with respect to the second limb part is positive and proportional to the edge distance to the second limb part; when the surface vertex is located outside the second limb part, the value is 0.

[0014] In some embodiments, determining whether the first limb part and the second limb part are interpenetrating based on the first interpenetration degree includes: if the first interpenetration degree exceeds a preset interpenetration degree threshold, determining that the first limb part and the second limb part are interpenetrating.

[0015] In a second aspect, an embodiment of the present specification provides a digital human penetration detection device, comprising: a construction unit, configured to construct a signed distance field of each limb part of the digital human according to the first coordinates of the surface vertices of each limb part of the digital human in the current posture; a first determination unit, configured to determine, for a first limb part of the digital human, a first penetration degree of the first limb part with respect to the second limb part according to the value of the signed distance field of the surface vertices of the first limb part with respect to the second limb part; a second determination unit, configured to determine whether the first limb part penetrates the second limb part according to the first penetration degree.

[0016] In a third aspect, an embodiment of the present specification provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed in a computer, the computer is caused to execute the method described in any implementation manner in the first aspect.

[0017] In a fourth aspect, an embodiment of the present specification provides a computing device, comprising a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, the method described in any implementation manner in the first aspect is implemented.

[0018] In a fifth aspect, an embodiment of the present specification provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the method described in any implementation manner in the first aspect.

[0019] The solution provided by the above-mentioned embodiment of this specification can construct a signed distance field (SDF) of each limb part according to the first coordinates of the surface vertices of each limb part of the digital human in the current posture. Then, for the first limb part of the digital human, the first degree of penetration of the first limb part with respect to the second limb part can be determined according to the value of the signed distance field of the surface vertex of the first limb part with respect to the second limb part, and according to the first degree of penetration, it is determined whether the first limb part and the second limb part have penetrated each other. By constructing a dedicated signed distance field for each limb part, these signed distance fields can accurately describe and quantify the spatial relationship between the limb parts, providing an accurate measurement tool for penetration detection. Therefore, the signed distance field technology can be combined to automatically and efficiently and accurately detect the penetration of the digital human after changing clothes or performing actions, greatly simplifying the process of penetration detection, timely discovering the problem of digital human penetration, so that the problem of digital human penetration can be solved in time, thereby improving the realism and visual effect of the digital human. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the multiple embodiments disclosed in this specification, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only the multiple embodiments disclosed in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0021] Figure 1 is an exemplary schematic diagram of a digital human's head;

[0022] Figure 2 It is a schematic diagram of an application scenario in which the embodiments of this specification can be applied;

[0023] Figure 3 is a flow chart of a process of determining the first coordinates of each surface vertex of a digital human in a current posture in an embodiment of this specification;

[0024] Figure 4 is a flow chart of a digital human penetration detection method in an embodiment of this specification;

[0025] Figure 5 It is a schematic diagram of the positional relationship between the vertices in space and the head of the digital human;

[0026] Figure 6 is another flow chart of the digital human penetration detection method in the embodiment of this specification;

[0027] Figure 7 It is a structural schematic diagram of a digital human penetration detection device in an embodiment of this specification. DETAILED DESCRIPTION

[0028] The solution provided in this specification is described below in conjunction with the drawings and embodiments.

[0029] As mentioned above, digital human penetration will lead to visual unreality and damage the user experience. Therefore, it is necessary to detect digital human penetration. The current digital human penetration detection method mainly relies on manual visual inspection and experience judgment, which is not only time-consuming and labor-intensive, but also prone to omissions and errors.

[0030] The embodiments of this specification provide a digital human penetration detection solution, which can automatically perform efficient and accurate penetration detection on digital humans, greatly simplify the penetration detection process, and promptly discover digital human penetration problems so that digital human penetration problems can be solved in a timely manner, thereby improving the realism and visual effects of digital humans.

[0031] The solution provided in the embodiments of this specification relates to a digital human including multiple limb parts. The multiple limb parts include at least four limbs of the digital human, and the four limbs may be a left lower limb, a right lower limb, a left upper limb, and a right upper limb. Furthermore, the multiple limb parts may also include a torso and / or a head of the digital human.

[0032] In practice, the surface of a digital human has a surface mesh. The surface mesh of a digital human is the basic structure of a digital human model, which is composed of many small triangles. Multiple small triangles are connected to form the outer surface of the digital human. Correspondingly, the surface of each limb of a digital human also has a surface mesh. Take the head of a digital human as an example. Figure 1 As shown in , the surface mesh of the head is composed of many small triangles, and the multiple small triangles are connected to each other to form the outer surface of the head. Figure 1 is an exemplary schematic diagram of a digital human head. The vertices of the small triangles are the vertices of the surface mesh. The vertices of the surface mesh of a digital human are called the surface vertices of the digital human. The vertices of the surface mesh of a limb are called the surface vertices of the limb.

[0033] After the digital human changes clothes or actions, Figure 2 As shown in , according to the coordinates of the surface vertices of each limb part of the digital human in the current posture (hereinafter referred to as the first coordinates), the signed distance field of each limb part is constructed. Figure 2 is a schematic diagram of an application scenario in which the embodiments of this specification can be applied. A signed distance field is a spatial representation method used to describe the distance between a point in space and the surface (edge) of an object. Afterwards, for a certain limb part among the limb parts, such as the right upper limb of a digital human, when it is necessary to detect whether the right upper limb penetrates the left upper limb, the following method can be used. Figure 2As shown in , according to the value of the directed distance field of the surface vertex of the right upper limb with respect to the left upper limb, the degree of penetration of the right upper limb with respect to the left upper limb is determined, and penetration detection is performed based on the degree of penetration to obtain the detection result of whether the right upper limb and the left upper limb have penetrated. Among them, the degree of penetration can intuitively reflect the existence and degree of penetration. Similarly, when it is necessary to detect whether the right upper limb has penetrated with other limbs except the left upper limb, for example, when it is necessary to detect whether the right upper limb has penetrated with the head, it can be as follows Figure 2 As shown in , according to the values ​​of the directed distance field of the surface vertices of the right upper limb with respect to the head, the degree of penetration of the right upper limb with respect to the head is determined, and penetration detection is performed based on the degree of penetration to obtain a detection result of whether the right upper limb penetrates the head.

[0034] As mentioned above, the first coordinates can be preset or determined on site before starting to construct the signed distance fields for each limb part.

[0035] Specifically, in one implementation, the first coordinates of the surface vertices of each limb part of the digital human in the current posture can be pre-stored in a target file (such as an FBX file). FBX (Filmbox) is a 3D file format used to exchange 3D models and animation data between different 3D software. When constructing a signed distance field for each limb part, the first coordinates of the surface vertices of each limb part in the current posture can be obtained from the target file, and then the signed distance field of each limb part can be constructed according to the obtained first coordinates.

[0036] Since the number of vertices on each surface of the digital human is relatively large, the number of coordinates of each surface vertex of the digital human in multiple different postures will be more. If all these coordinates are stored, the target file will become very large. Considering that each surface vertex is associated with several joints, the number of joints of the digital human is much less than the number of vertices. In order to save storage space, in another embodiment, a reference posture can be set for the digital human, for example, the T pose (posture) or A pose of the digital human is set as the reference posture, so that the target file stores the vertex weight distribution of the digital human, the joint transformation data of the digital human in the reference posture (hereinafter referred to as the first joint transformation data) and the vertex coordinate data, and the joint transformation data of the digital human in other postures (hereinafter referred to as the second joint transformation data). Among them, the first joint transformation data and the second joint transformation data both include the joint transformation matrices of multiple joints of the digital human. The vertex weight distribution includes the weight distribution of each surface vertex of the digital human with respect to the associated several joints. The vertex coordinate data includes the coordinates of each surface vertex in space.

[0037] Based on this, the first coordinates of each surface vertex of the digital human in the current posture can be calculated according to the vertex weight distribution, the first joint transformation data, the vertex coordinate data and the second joint transformation data, and then the directed distance field of each limb part of the digital human can be constructed according to the calculated first coordinates. In one example, the first coordinates can be calculated using the Linear Blend Skinning (LBS) algorithm. The LBS algorithm is a technology used to simulate character animation, which achieves animation effects by associating one or more bones with the character's skin.

[0038] Furthermore, one can Figure 3 The coordinate determination process shown in the figure determines the first coordinates of each surface vertex of the digital human in the current posture. Figure 3 This is a flowchart of the process of determining the first coordinates of each surface vertex of the digital human in the current posture in the embodiment of this specification. The determination process can be performed by any platform or server or device cluster with computing and processing capabilities, and includes steps S301-S305 as shown below.

[0039] like Figure 3 As shown, in step S301, the vertex weight distribution of the digital human, the first joint transformation data and the vertex coordinate data in the reference posture, and the second joint transformation data in the current posture are obtained; the first joint transformation data and the second joint transformation data both include the joint transformation matrices of multiple joints of the digital human; the vertex weight distribution includes the weight distribution of each surface vertex of the digital human with respect to the associated several joints.

[0040] Specifically, the vertex weight distribution of the digital human, the first joint transformation data and vertex coordinate data in the reference pose, and the second joint transformation data in the current pose can be obtained from the target file as described above. Furthermore, when the target file is an FBX file, data can be extracted from the FBX file by calling the FBX SDK (Software Development Kit).

[0041] Next, in step S303, for each target vertex among the surface vertices of the digital human, the third coordinate of the target vertex relative to each of the joints is determined according to the second coordinate of the target vertex in the reference posture, and the first joint transformation matrix of each associated joint in the reference posture and the second joint transformation matrix in the current posture.

[0042] Specifically, the relative coordinates of the target vertex relative to the local coordinate system of each joint can be determined according to the first joint transformation matrix and the second coordinates. Afterwards, the third coordinates of the target vertex relative to each of the joints can be determined according to the second joint transformation matrix and the relative coordinates determined for the target vertex.

[0043] Next, in step S305, weighted sum is performed on each third coordinate of the target vertex according to the weight distribution of the target vertex with respect to each associated joint, so as to obtain the first coordinate of the target vertex in the current posture.

[0044] The target vertex is vertex e, vertex e is associated with joints j1, j2, and j3, the second coordinate of vertex e in the reference posture is P, the weight distribution of vertex e with respect to joints j1, j2, and j3 is w1, w2, and w3, the first joint transformation matrix of joints j1, j2, and j3 in the reference posture is θ1, θ2, and θ3, respectively, and the second joint transformation matrix of joints j1, j2, and j3 in the current posture is M1, M2, and M3, respectively. When determining the first coordinate of vertex e in the current posture, the relative coordinate rei=θi*P of vertex e relative to the local coordinate system of joint ji can be calculated first, i=1, 2, 3. After that, the third coordinate ei=Mi*rei of vertex e relative to joint ji can be calculated. Then, e1, e2, and e3 can be weighted summed according to the weight distribution to calculate the first coordinate E=w1*e1+w2*e2+w3*e3 of vertex e in the current posture.

[0045] In the foregoing, it is introduced that the first coordinates of each surface vertex of the digital human in the current posture can be preset or determined on the spot. In practice, in addition to the first coordinates of each surface vertex being preset or determined on the spot, the limb parts to which each surface vertex belongs can also be preset or determined on the spot.

[0046] Specifically, as an example, a configuration file recording the correspondence between each surface vertex of the digital human and the limb part to which it belongs can be generated in advance. When performing mold penetration detection on the digital human, the surface vertices of each limb part of the digital human can be known by reading the configuration file.

[0047] As another example, each surface vertex of the digital human is associated with several joints, and a weight distribution of the associated joints is preset. When performing mold penetration detection on the digital human, the limb parts to which each surface vertex of the digital human belongs can be determined based on the weight distribution. For example, for each surface vertex of the digital human, the limb part to which the joint corresponding to the maximum weight value in the weight distribution of the surface vertex belongs can be determined, and the surface vertex is divided into the limb part. It should be understood that the limb parts to which each joint of the digital human belongs are pre-configured.

[0048] Next, combine Figure 4 , the digital human penetration detection method is further explained. Figure 4 This is a flow chart of a digital human penetration detection method in an embodiment of this specification. The method can be executed by any platform or server or device cluster with computing and processing capabilities, and includes steps S401-S405 as shown below.

[0049] like Figure 4 As shown, in step S401, a directed distance field of each limb part is constructed according to the first coordinates of the surface vertices of each limb part of the digital human in the current posture.

[0050] Among them, the signed distance field is a spatial representation method used to describe the distance between a point in space and the surface (edge) of an object. For each limb part of a digital human, the signed distance field technology can be used to construct the signed distance field of the limb part according to the first coordinates of each surface vertex of the limb part in the current posture. For example, each surface vertex can be voxelized into an N×N×N voxel grid as a signed distance field.

[0051] Furthermore, in order to better adapt to the scene, the solution provided in the embodiment of this specification adopts an improved signed distance field technology. The characteristic of the improved signed distance field technology is that, for the signed distance field of any limb part, and any target vertex in the space, when the target vertex is located inside the limb part, the signed distance value of the target vertex is a positive number and is proportional to the edge distance to the limb part; when the target vertex is located outside the limb part, the signed distance value of the target vertex is 0. Among them, the edge distance from the target vertex to the limb part can be the shortest distance from the target vertex to each point on the edge (surface) of the limb part. As shown in the example Figure 5 Take the head of the digital person shown in FIG. 1 and vertices a and b in the space as an example. Vertex a is located inside the head, and the value of the signed distance field of vertex a with respect to the head is a positive number and is proportional to the edge distance to the head; vertex b is located outside the head, and the value of the signed distance field of vertex b with respect to the head is 0. Among them, Figure 5 It is a schematic diagram of the positional relationship between the vertices in space and the head of the digital human.

[0052] Next, in step S403, for the first limb part of the digital human, the first penetration of the first limb part with respect to the second limb part is determined according to the value of the directed distance field of the surface vertex of the first limb part with respect to the second limb part.

[0053] In one example, the first limb part can be any one of the limbs of the digital human. Considering that the limbs of the digital human are prone to interpenetration with other limbs, in another example, the first limb part can be any one of the limbs of the digital human. The second limb part is any one of the limbs of the digital human except the first limb part.

[0054] Taking the detection of whether the limbs of the digital human are interpenetrating with other limbs as an example, each limb part of the limbs of the digital human can be used as a first limb part. For each first limb part, at least one limb part of the limb parts of the digital human except the first limb part is used as a second limb part, and according to the value of the directed distance field of the surface vertex of the first limb part with respect to each second limb part, the first interpenetration degree of the first limb part with respect to the second limb part is determined.

[0055] According to the description of the signed distance field in the previous text, it can be known that for any surface vertex of the first limb part, when the surface vertex is located inside the second limb part, the value of the signed distance field of the surface vertex with respect to the second limb part can be a positive number and is proportional to the edge distance to the second limb part; when the surface vertex is located outside the second limb part, the value can be 0. In this case, the first penetration of the first limb part with respect to the second limb part is positively correlated with the value of the signed distance field of the surface vertex of the first limb part with respect to the second limb part.

[0056] In one embodiment, when determining the first penetration of a first limb part with respect to a second limb part, the second penetration of each surface vertex of the first limb part with respect to the second limb part may be determined according to the value of the directed distance field of the surface vertex with respect to the second limb part. Afterwards, the average value of the second penetration of each surface vertex of the first limb part may be determined as the first penetration of the first limb part with respect to the second limb part.

[0057] Furthermore, when the above-mentioned improved signed distance field technology is used to construct a signed distance field for each limb part of the digital human, for each surface vertex of the first limb part, the value of the signed distance field of the surface vertex with respect to the second limb part can be determined as the second penetration degree of the surface vertex with respect to the second limb part.

[0058] It should be noted that by constructing exclusive signed distance fields for each limb part, these signed distance fields can accurately describe and quantify the spatial relationship between limb parts, providing an accurate measurement tool for penetration detection. In addition, the adopted signed distance field is an interpretable strategy. By improving the definition of the signed distance field, it can be clearly known whether a vertex is in a penetration state. This method is more reasonable and accurate.

[0059] Next, in step S405, it is determined whether the first limb part penetrates the second limb part according to the first degree of penetration.

[0060] As an example, a penetration threshold may be preset, so that the first penetration may be compared with the preset penetration threshold. When the first penetration exceeds the penetration threshold, it may be determined that penetration occurs between the first limb part and the second limb part. When the first penetration does not exceed the penetration threshold, it may be determined that no penetration occurs between the first limb part and the second limb part.

[0061] As another example, an evaluation rule may be preset, so that the evaluation rule is used to determine whether the first limb part and the second limb part penetrate each other according to the first penetration degree. The content of the evaluation rule may be set according to actual needs and is not specifically limited here.

[0062] It should be noted that after determining whether the first limb part and the second limb part have interpenetration, a detection result indicating whether the first limb part and the second limb part have interpenetration can be obtained. Afterwards, the detection result can be included in the detection result information of the digital human. The detection result information can provide quantitative reference data to assist the designer of the digital human to make efficient corrections.

[0063] Figure 4 The solution provided by the corresponding embodiment can combine the directed distance field technology to automatically and efficiently and accurately detect the penetration of the digital human after changing clothes or performing actions, greatly simplifying the penetration detection process, and timely discovering the problem of digital human penetration, so that the problem of digital human penetration can be solved in time, thereby improving the realism and visual effects of the digital human.

[0064] In one embodiment, in addition to the penetration detection of limb parts, the penetration ratio of the limb parts can also be evaluated. By incorporating the penetration ratio into the detection result information of the digital human, it can further assist the designer of the digital human to make efficient corrections.

[0065] Next, combine Figure 6 , introduces the process of model penetration detection and model penetration ratio evaluation of digital human limbs. Figure 6This is another flow chart of the digital human penetration detection method in the embodiment of this specification. The method can be executed by any platform or server or device cluster with computing and processing capabilities, and includes steps S601-S611 as shown below.

[0066] like Figure 6 As shown, in step S601, a directed distance field of each limb part is constructed according to the first coordinates of the surface vertices of each limb part of the digital human in the current posture.

[0067] In step S603, for each surface vertex of the first limb part, a second penetration of the surface vertex with respect to the second limb part is determined according to the value of the signed distance field of the surface vertex with respect to the second limb part.

[0068] In step S605, an average value of the second penetration of each surface vertex of the first limb part is determined as the first penetration of the first limb part with respect to the second limb part.

[0069] In step S607, it is determined whether the first limb part penetrates the second limb part according to the first degree of penetration.

[0070] For the explanation of steps S601-S607, please refer to the relevant description in the previous text, which will not be repeated here.

[0071] In step S609, based on the second penetration degree of each surface vertex of the first limb part, it is determined whether each surface vertex penetrates the second limb part.

[0072] The specific implementation process of step S609 is similar to that of step S405 in the previous text. Please refer to the relevant description of step S405 and will not be repeated here.

[0073] It should be noted that, in one embodiment, in order to improve the overall execution efficiency of the digital human penetration detection process, step S609 can be executed after step S603 is executed, for example, it can be executed in parallel with step S605. In another embodiment, step S609 can be executed when the determination result of step S607 is yes. It should be understood that the embodiments of this specification do not specifically limit the execution time of step S609.

[0074] In step S611, the penetration ratio of the first limb part with respect to the second limb part is determined based on a first number of surface vertices in the first limb part that penetrate with the second limb part and a second number of each surface vertices of the first limb part.

[0075] Specifically, the ratio of the first number to the second number can be determined as the penetration ratio of the first limb part with respect to the second limb part.

[0076] Figure 6 The solution provided by the corresponding embodiment can combine the directed distance field technology to automatically and efficiently and accurately detect the penetration of the digital human after changing clothes or performing actions, and provide the penetration ratio as an evaluation indicator. This solution has the characteristics of automation, efficiency and accuracy, and can greatly simplify the penetration detection process, timely discover the problem of digital human penetration, so that the problem of digital human penetration can be solved in time, thereby improving the realism and visual effect of the digital human.

[0077] Need to explain, Figure 3 , Figure 4 , Figure 6 In the corresponding embodiments, for the calculation process based on vertex coordinates, in order to improve the calculation efficiency, the vertex coordinates can be merged into tensors, and the calculation process can be quickly implemented in a batch processing manner.

[0078] Figure 7 It is a structural diagram of a digital human penetration detection device in an embodiment of this specification. The device can be applied to any platform or server or device cluster with computing and processing capabilities. The device includes: a construction unit 701, configured to construct a signed distance field of each limb part of the digital human according to the first coordinates of the surface vertices of each limb part of the digital human in the current posture; a first determination unit 702, configured to determine the first penetration degree of the first limb part with respect to the second limb part according to the value of the signed distance field of the surface vertex of the first limb part with respect to the second limb part for the first limb part of the digital human; a second determination unit 703, configured to determine whether the first limb part penetrates the second limb part according to the first penetration degree.

[0079] In one embodiment, the above-mentioned device may also include: an acquisition unit (not shown in the figure), configured to acquire the vertex weight distribution of the digital human, the first joint transformation data and the vertex coordinate data in the reference posture, and the second joint transformation data in the current posture; the first joint transformation data and the second joint transformation data both include the joint transformation matrices of multiple joints of the digital human; the vertex weight distribution includes the weight distribution of each surface vertex of the digital human with respect to the associated several joints; a third determination unit (not shown in the figure), configured to determine the first coordinates of each surface vertex of the digital human in the current posture according to the first joint transformation data, the vertex coordinate data, the second joint transformation data and the vertex weight distribution.

[0080] In one embodiment, the third determination unit can be further configured to: for each target vertex among the surface vertices of the digital human, determine the third coordinate of the target vertex relative to each of the joints according to the second coordinate of the target vertex in the reference posture, and the first joint transformation matrix of each associated joint in the reference posture and the second joint transformation matrix in the current posture; and perform weighted summation on the third coordinates of the target vertex according to the weight distribution of the target vertex with respect to each joint to obtain the first coordinate of the target vertex in the current posture.

[0081] In one embodiment, the third determination unit can be further configured to: determine the relative coordinates of the target vertex relative to the local coordinate systems of each of the above-mentioned joints based on the first joint transformation matrix and the second coordinates; determine the third coordinates of the target vertex relative to each of the above-mentioned joints based on the second joint transformation matrix and the relative coordinates determined for the target vertex.

[0082] In one embodiment, each surface vertex of the digital human is associated with a number of joints, and a weight distribution of the associated joints is preset; the above-mentioned device may also include: a division unit (not shown in the figure), which is configured to determine, for each surface vertex of the digital human, the limb part to which the joint corresponding to the maximum weight value in the weight distribution of the surface vertex belongs, and divide the surface vertex into the limb part.

[0083] In one embodiment, the first determination unit 702 can be further configured to: for each surface vertex of the first limb part, determine the second penetration of the surface vertex with respect to the second limb part according to the value of the directed distance field of the surface vertex with respect to the second limb part; and determine the average value of the second penetration of each surface vertex of the first limb part as the first penetration.

[0084] In one embodiment, the second determination unit 703 can also be configured to: determine whether each surface vertex of the first limb part penetrates with the second limb part based on the second penetration degree of each surface vertex of the first limb part; determine the penetration ratio of the first limb part with respect to the second limb part based on the first number of surface vertices in the first limb part that penetrate with the second limb part, and the second number of each surface vertices of the first limb part.

[0085] In one embodiment, for any surface vertex of a first limb part, when the surface vertex is located inside a second limb part, the value of the signed distance field of the surface vertex with respect to the second limb part is a positive number and is proportional to the edge distance to the second limb part; when the surface vertex is located outside the second limb part, the value is 0.

[0086] In one embodiment, the second determination unit 703 may be further configured to: determine that the first limb part and the second limb part are penetrating if the first penetration degree exceeds a preset penetration degree threshold.

[0087] The present specification also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed in a computer, the computer is caused to execute the following Figure 3 , Figure 4 , Figure 6 Describe the method.

[0088] The embodiment of the present specification also provides a computing device, including a memory and a processor, wherein the memory stores an executable code, and when the processor executes the executable code, the following is implemented: Figure 3 , Figure 4 , Figure 6 Describe the method.

[0089] The embodiments of the present specification also provide a computer program product, including a computer program / instruction, which is executed by a processor to implement the following Figure 3 , Figure 4 , Figure 6 Describe the method.

[0090] Those skilled in the art should be aware that in one or more of the above examples, the functions described in the multiple embodiments disclosed in this specification can be implemented by hardware, software, firmware, or any combination thereof. When implemented by software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium.

[0091] In some cases, the actions or steps described in the claims may be performed in a different order than in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0092] The specific implementation methods described above further illustrate in detail the purposes, technical solutions and beneficial effects of the multiple embodiments disclosed in this specification. It should be understood that the above description is only the specific implementation methods of the multiple embodiments disclosed in this specification, and is not used to limit the protection scope of the multiple embodiments disclosed in this specification. Any modifications, equivalent substitutions, improvements, etc. made on the basis of the technical solutions of the multiple embodiments disclosed in this specification should be included in the protection scope of the multiple embodiments disclosed in this specification.

Claims

1. A digital human penetration detection method, comprising: Constructing the directed distance field of each limb part according to the first coordinates of the surface vertices of each limb part of the digital human in the current posture; For a first limb part of the digital human, determining a first penetration degree of the first limb part with respect to the second limb part according to the value of the directed distance field of the surface vertex of the first limb part with respect to the second limb part; According to the first penetration degree, it is determined whether the first limb part penetrates the second limb part.

2. The method according to claim 1, wherein: Before constructing the signed distance fields of the limb parts, the method further includes: Acquire the vertex weight distribution of the digital human, the first joint transformation data and the vertex coordinate data in the reference posture, and the second joint transformation data in the current posture; the first joint transformation data and the second joint transformation data both include the joint transformation matrices of the multiple joints of the digital human; the vertex weight distribution includes the weight distribution of each surface vertex of the digital human with respect to the associated multiple joints; The first coordinates of each surface vertex of the digital human in the current posture are determined according to the first joint transformation data, the vertex coordinate data, the second joint transformation data and the vertex weight distribution.

3. The method according to claim 2, wherein: The determining of the first coordinates of each surface vertex of the digital human in the current posture includes: For each target vertex among the surface vertices of the digital human, determine the third coordinate of the target vertex relative to each of the joints according to the second coordinate of the target vertex in the reference posture, and the first joint transformation matrix of each associated joint in the reference posture and the second joint transformation matrix in the current posture; According to the weight distribution of the target vertex with respect to each joint, weighted sum is performed on each third coordinate of the target vertex to obtain the first coordinate of the target vertex in the current posture.

4. The method according to claim 3, wherein: The determining of the third coordinate of the target vertex relative to each of the joints comprises: Determine the relative coordinates of the target vertex relative to the local coordinate systems of the joints according to the first joint transformation matrix and the second coordinates; A third coordinate of the target vertex relative to each of the joints is determined according to the second joint transformation matrix and the relative coordinates determined for the target vertex.

5. The method according to claim 1, wherein: Each surface vertex of the digital human is associated with a number of joints, and a weight distribution of the associated joints is preset; Before constructing the signed distance fields of the limb parts, the method further includes: For each surface vertex of the digital human, the limb part to which the joint corresponding to the maximum weight value in the weight distribution of the surface vertex belongs is determined, and the surface vertex is divided into the limb part.

6. The method according to claim 1, wherein: The determining a first degree of penetration of the first limb part with respect to the second limb part comprises: For each surface vertex of the first limb part, determining a second penetration degree of the surface vertex with respect to the second limb part according to a value of the signed distance field of the surface vertex with respect to the second limb part; The average value of the second penetration of each surface vertex of the first limb part is determined as the first penetration.

7. The method according to claim 6, further comprising: Determining, according to the second penetration degree of each surface vertex of the first limb part, whether each surface vertex penetrates the second limb part; The penetration ratio of the first limb part with respect to the second limb part is determined based on a first number of surface vertices in the first limb part that penetrate with the second limb part and a second number of each surface vertices of the first limb part.

8. The method according to claim 1, wherein: For any surface vertex of the first limb part, when the surface vertex is located inside the second limb part, the value of the signed distance field of the surface vertex with respect to the second limb part is positive and proportional to the edge distance to the second limb part; when the surface vertex is located outside the second limb part, the value is 0.

9. The method according to any one of claims 1 to 8, wherein: The step of determining whether the first limb part and the second limb part penetrate each other according to the first degree of penetration includes: If the first penetration degree exceeds a preset penetration degree threshold, it is determined that the first limb part and the second limb part penetrate each other.

10. A digital human penetration detection device, comprising: A construction unit, configured to construct a signed distance field of each limb part according to the first coordinates of the surface vertices of each limb part of the digital human in the current posture; A first determination unit is configured to determine, for a first limb part of the digital human, a first penetration degree of the first limb part with respect to the second limb part according to a value of a directed distance field of a surface vertex of the first limb part with respect to the second limb part; The second determination unit is configured to determine whether the first limb part penetrates the second limb part according to the first penetration degree.

11. A computer-readable storage medium having a computer program stored thereon, wherein: When the computer program is executed in a computer, the computer is caused to execute the method according to any one of claims 1 to 9.

12. A computing device comprising a memory and a processor, wherein: The memory stores executable codes, and when the processor executes the executable codes, the method according to any one of claims 1 to 9 is implemented.

Citation Information

Patent Citations

  • Method, system and equipment for binding clothing model and human body model

    CN116187051A

  • Virtual image driving method and device, electronic equipment, storage medium and product

    CN118115636A

  • Contact-aware retargeting of motion

    US20230037339A1