Rapid 3D digital human face skeleton binding and transferring method

By extracting and transferring the bone binding information of the 3D digital human facial model, using deformation function and skin weight transfer technology, the problem of time-consuming and inrelative bone binding in traditional methods is solved, and a fast and efficient 3D digital human is achieved. Facial bone binding transfer.

CN119516058AInactive Publication Date: 2025-02-25ZHEJIANG VERSATILE MEDIA
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411614736.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-11-13
Publication Date
2025-02-25
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Traditional methods of creating 3D digital humans and creating bone bindings require high abilities and time-consuming to artists, and cannot reuse the binding relationship of the face model, resulting in inefficient productivity.

Method used

By extracting the bone binding information of existing digital humans, the re-topology artist creates a new facial model, calculates the deformation function, obtains the existing bone structure through the deformation function, and transfers the skin weight.

Benefits of technology

Automatic transfer of bone binding is achieved, reducing artists' workload on carving, improving productivity, and allowing reusing of binding relationships.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119516058A_ABST
    Figure CN119516058A_ABST
Patent Text Reader

Abstract

The invention discloses a rapid 3D digital human face skeleton binding and transferring method, which is characterized in that a skeleton is transplanted and bound to a new digital human face model according to a rule existing between the face model and the skeleton, and comprises the following steps: S1, extracting skeleton binding information of an existing digital human; s2, making a carved new face model by a re-topology artist; s3, calculating a deformation function according to the existing face model and the new face model; s4, enabling the existing skeleton structure to pass through the deformation function to obtain a new skeleton structure; and S5, skin weight transfer. Compared with a traditional 3D digital face driven binding process, the rapid 3D digital human face skeleton binding transfer method has the advantages that skeleton binding is automatically transferred through a deformation function according to an implicit relationship between the skeleton and the face model, a large amount of engraving by artists is not needed, and a large amount of manpower work is reduced.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of 3D digital human, and in particular to a fast 3D digital human face skeleton binding transfer method. Background Art

[0002] In the fields of film, television, games and entertainment, 3D digital humans are often used to have realistic 3D images and rich content presentation capabilities. Face driving is the core technology. At present, a common solution is bone driving, which is to establish the relationship between bones and the 3D model of the face. By controlling the bones, the face model can be driven to make various expressions. Establishing the relationship between bones and face models is bone binding, which is usually designed or carved by artists in digital content creation software.

[0003] The main drawbacks of the existing solutions are that the traditional method of making 3D digital humans and creating skeleton rigs requires high artist skills and takes a lot of time, often taking weeks or months. In addition, the modification and replacement of the face model cannot reuse a set of binding relationships, resulting in low production efficiency. To this end, we propose a fast 3D digital human face skeleton rig transfer method. Summary of the invention

[0004] The purpose of the present invention is to provide a fast 3D digital human face skeleton binding transfer method to solve the problems raised in the above background technology.

[0005] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a method for fast 3D digital human face skeleton binding transfer, a method for fast 3D digital human face skeleton binding transfer, the method transplants the skeleton to bind to the new digital human face model according to the law between the face model and the skeleton, and comprises the following steps:

[0006] S1, extracting the skeleton binding information of the existing digital human;

[0007] S2, retopology artist creates a new facial model for carving;

[0008] S3, calculating a deformation function according to the existing facial model and the new facial model;

[0009] S4, obtaining a new bone structure by applying a deformation function to the existing bone structure;

[0010] S5. Skin weight transfer.

[0011] Preferably, in S1, the existing digital human is produced by digital content creation software, and the digital human skeleton binding information can be obtained through the script;

[0012] A single bone is defined as:

[0013] Joint={id,name,offset,parent};

[0014] id is the bone number, name is the bone name, It is the relative offset of the parent bone, and parent is the parent bone number.

[0015] Preferably, the M facial bones form a skeleton tree according to a hierarchical relationship, which is defined as:

[0016] Tree = {Joint 0 ,Joint 1 ,…Joint M};

[0017] The face model is simplified as a set of N surface vertices, defined as:

[0018] Mesh (N) = {P 0 ,P 1 ,…P N},

[0019] Preferably, the facial model surface vertices are associated with the skeleton tree through the "skinning" technology, and the deformation and animation of the model are controlled by moving the bones, that is:

[0020]

[0021] Where T represents the bone transformation function, which can obtain the bones after movement. For skin weight, the weight relationship between the skeleton tree and the model is recorded. The existing bound digital face can be expressed as:

[0022] Face = {Tree (ref,M) ,Mesh ref,(N) ,W ref}.

[0023] Preferably, the retopology in S2 is used to simplify a complex, high-polygon model into a low-polygon model that is simpler, easier to use and manage.

[0024] Preferably, the high-poly model sculpted by the artist is converted into a low-poly model using DDC software retopology, so that the wiring and structure of the model are more regular and the topology is the same as the existing model, that is:

[0025] Mesh target,(N) =Retopology(Mesh raw,(K) ,Mesh ref,(N) ),(K>>N).

[0026] Preferably, in the 3D modeling, the deformation function is used to geometrically deform the model to produce a new shape, that is:

[0027] Mesh deform,(N) =Deform(Mesh ref,(N) ).

[0028] Preferably, a radial basis function is selected to create a smooth deformation, namely:

[0029] RBF: Y = GX;

[0030] Among them, X is the original control point, Y is the deformation control point, and a large number of control point pairs are used to list a linear equation group to calculate G and obtain the deformation function.

[0031] Preferably, in S4, it is assumed that the vertices and bones are distributed in the same hidden surface, so the deformation function can be reused, and the deformed bones are:

[0032] Tree deform,(M) =RBF(Tree ref,(M) ).

[0033] Preferably, in S5, DDC software is used to transfer the skin to obtain a facial model corresponding to the topology and a skeleton tree matching it.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] Compared with the traditional 3D digital face driven binding process, the present invention automatically transfers the bone binding through the deformation function according to the implicit relationship between the bone and the facial model, which does not require a lot of carving by the artist and reduces a lot of manpower work. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 The pattern that controls the model deformation for the default state;

[0037] Figure 2 A pattern that controls the deformation of the model for the motion state of the mouth bones;

[0038] Figure 3 This is a schematic diagram showing the influence of the selected ear bones on the model range;

[0039] Figure 4 This is a schematic diagram showing the influence of the selected nose bridge bone on the model range;

[0040] Figure 5 This is a diagram showing the influence of the selected facial parent bone on the model range;

[0041] Figure 6-8 The figure shows the spatial distribution of bones and the regularity of model vertices. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] The present invention provides a technical solution: a method for fast 3D digital human face bone binding transfer, which transplants bones and binds them to a new digital human face model according to the rules between the face model and the bones, and includes the following steps:

[0044] S1, extracting the skeleton binding information of the existing digital human;

[0045] S2, retopology artist creates a new facial model for carving;

[0046] S3, calculating a deformation function according to the existing facial model and the new facial model;

[0047] S4, obtaining a new bone structure by applying a deformation function to the existing bone structure;

[0048] S5. Skin weight transfer.

[0049] As a preferred implementation, in S1, the existing digital human is produced by digital content creation (DDC) software, and the digital human skeleton binding information can be obtained through the script;

[0050] A single bone is defined as:

[0051] Joint={id,name,offset,parent};

[0052] id is the bone number, name is the bone name, It is the relative offset of the parent bone, and parent is the parent bone number.

[0053] As a preferred implementation, the M facial bones form a skeleton tree according to a hierarchical relationship, which is defined as:

[0054] Tree = {Joint 0 ,Joint 1 ,…Joint M};

[0055] The face model is simplified as a set of N surface vertices, defined as:

[0056] Mesh(N) = {P 0 ,P 1 ,…P N},

[0057] As a preferred implementation, the vertices on the face model surface are associated with the skeleton tree through the "skinning" technology, and the deformation and animation of the model are controlled by moving the skeleton, that is:

[0058]

[0059] Where T represents the bone transformation function, which can obtain the bones after movement. For skin weight, the weight relationship between the skeleton tree and the model is recorded. The existing bound digital face can be expressed as:

[0060] Face = {Tree (ref,M) ,Mesh ref,(N) ,W ref}.

[0061] As a preferred implementation, retopology is an important step that involves optimizing the mesh structure of the model to make it more suitable for animation and real-time rendering. The definition of retopology can be understood as follows: retopology is the process of simplifying a complex, high-polygon model (high-poly) into a simpler, easier to use and manage low-polygon model (low-poly). This process not only reduces the complexity of the model, but also ensures its efficient performance in a real-time rendering environment.

[0062] As a preferred implementation, the high-poly model sculpted by the artist is converted into a low-poly model by using DDC software retopology, so that the wiring and structure of the model are more regular and the topology structure is the same as that of the existing model, that is:

[0063] Mesh target,(N) =Retopology(Mesh raw,(K) ,Mesh ref,(N) ),(K>>N).

[0064] As a preferred implementation, in the 3D modeling, the deformation function (Deform) of S3 is used to geometrically deform the model to produce a new shape, that is:

[0065] Mesh deform,(N) =Deform(Mesh ref,(N) ).

[0066] As a preferred implementation, there are multiple deformation functions, and we select one of them: Radial Basis Function, which can be used to create smooth deformation, namely:

[0067] RBF: Y = GX;

[0068] Among them, X is the original control point, and Y is the deformation control point. By listing a linear equation system through a large number of control point pairs, G can be calculated, that is, the deformation function is obtained.

[0069] As a preferred implementation, after a lot of observations, we found that the spatial distribution of well-bound bones and model vertices has a regularity, so it can be assumed that the vertices and bones are distributed in the same hidden surface, and the deformation function can also be reused, then the deformed bones are:

[0070] Tree deform,(M) =RBF(Tree ref,(M) ).

[0071] As a preferred implementation, in S5, the facial model corresponding to the topology and the skeleton tree matching it have been obtained according to the above steps, and the skinning can be transferred using the DDC software.

[0072] Please refer to Figure 1-2 , you can see Figure 1 is the default state. Figure 2 The mouth bones move, causing the model to deform and make the mouth appear open.

[0073] The degree to which bones affect the model is also referred to as skin weights, such as Figure 3 As shown, select the ear bone, and you can see the affected model range as follows. The degree of influence increases from blue to red, which means that the movement of the ear bone will cause the model in the colored area to deform. Figure 4 This is a schematic diagram showing the influence of the selected nose bridge bone on the model range; Figure 5 A diagram showing the extent to which the selected facial parent bone affects the model.

[0074] like Figure 6-8 As shown, there is a regularity between the spatial distribution of well-bound bones and the model vertices. It can be seen that the positions of the bone vertices (green dots) are generally distributed on the facial model.

[0075] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0076] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A fast 3D digital human face skeleton binding transfer method, characterized by: The method transplants bones and binds them to a new digital human face model according to the rules existing between the face model and the bones, and includes the following steps: S1, extracting the skeleton binding information of the existing digital human; S2, retopology artist creates a new face model for carving; S3, calculating a deformation function according to the existing facial model and the new facial model; S4, obtaining a new bone structure by applying a deformation function to the existing bone structure; S5. Skin weight transfer.

2. A fast 3D digital human face skeleton binding transfer method according to claim 1, characterized in that: In S1, the existing digital human is produced by digital content creation software, and the digital human skeleton binding information can be obtained through the script; A single bone is defined as: Joint={id,name,offset,parent}; id is the bone number, name is the bone name, It is the relative offset of the parent bone, and parent is the parent bone number.

3. The method for fast 3D digital human face skeleton binding transfer according to claim 2, characterized in that: The M facial bones form a skeleton tree according to the hierarchical relationship, which is defined as: Tree={Joint0,Joint1,…Joint M }; The face model is simplified as a set of N surface vertices, defined as:

4. The method for fast 3D digital human face skeleton binding transfer according to claim 3, characterized in that: The facial model surface vertices are associated with the bone tree through the "skinning" technology, and the deformation and animation of the model are controlled by moving the bones, namely: Where T represents the bone transformation function, which can obtain the bones after movement. For skin weight, the weight relationship between the skeleton tree and the model is recorded. The existing bound digital face can be expressed as: Face={Tree (ref,M) ,Mesh ref,(N) ,W ref }。 5. The method for fast 3D digital human face skeleton binding transfer according to claim 4, characterized in that: The retopology in S2 is used to simplify a complex, high-polygon model into a lower-polygon model that is cleaner and easier to use and manage.

6. The method for fast 3D digital human face skeleton binding transfer according to claim 5, characterized in that: Use DDC software to retopology to convert the artist's high-poly model into a low-poly model, so that the wiring and structure of the model are more regular, and the topology structure is the same as the existing model, that is: Mesh target(N) =Retopology(Mesh raw,(K) ,Mesh ref,(N) ),(K>>N)。 7. A fast 3D digital human face skeleton binding transfer method according to claim 6, characterized in that: In 3D modeling, the deformation function is used to geometrically deform the model to produce a new shape, namely: Mesh deform,(N) =Deform(Mesh ref,(N) )。 8. The method for fast 3D digital human face skeleton binding transfer according to claim 7, characterized in that: Select the radial basis function to create a smooth deformation, that is: RBF: Y = GX; Among them, X is the original control point, Y is the deformation control point, and a large number of control point pairs are used to list a linear equation group to calculate G and obtain the deformation function.

9. The method for fast 3D digital human face skeleton binding transfer according to claim 8, characterized in that: In S4, it is assumed that the vertices and bones are distributed in the same hidden surface, so the deformation function can be reused, and the deformed bones are: Tree deform,(M) =RBF(Tree ref,(M) )。 10. The method for fast 3D digital human face skeleton binding transfer according to claim 9, characterized in that: In S5, DDC software is used to transfer the skin to the facial model corresponding to the topology and the matching skeleton tree.

Citation Information

Patent Citations

  • Game role face model processing method, device, processor and terminal

    CN109395390A

  • Digital human generation method and device, storage medium and electronic equipment

    CN114693876A

  • Facial skeleton binding automatic migration method and system based on fbx file

    CN114820917A

  • Virtual object skeleton binding migration method and device, equipment and storage medium

    CN115131476A