CN112102470A - Linear microminiaturible parametric clothing model manufacturing method and parameter optimization method thereof - Google Patents
Linear microminiaturible parametric clothing model manufacturing method and parameter optimization method thereof Download PDFInfo
- Publication number
- CN112102470A CN112102470A CN202010828692.4A CN202010828692A CN112102470A CN 112102470 A CN112102470 A CN 112102470A CN 202010828692 A CN202010828692 A CN 202010828692A CN 112102470 A CN112102470 A CN 112102470A
- Authority
- CN
- China
- Prior art keywords
- clothes
- parameters
- model
- human body
- garment
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000005457 optimization Methods 0.000 title claims abstract description 20
- 238000004519 manufacturing process Methods 0.000 title claims description 4
- 230000008569 process Effects 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 7
- 230000003068 static effect Effects 0.000 abstract description 3
- 238000013135 deep learning Methods 0.000 abstract description 2
- 238000005516 engineering process Methods 0.000 abstract description 2
- 238000003709 image segmentation Methods 0.000 abstract description 2
- 230000006872 improvement Effects 0.000 abstract description 2
- 238000002360 preparation method Methods 0.000 abstract 1
- 230000006870 function Effects 0.000 description 12
- 238000001514 detection method Methods 0.000 description 5
- 230000014509 gene expression Effects 0.000 description 5
- 230000009471 action Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 230000003247 decreasing effect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000037237 body shape Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000013386 optimize process Methods 0.000 description 1
- 238000009877 rendering Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/12—Cloth
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Geometry (AREA)
- Multimedia (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Human Computer Interaction (AREA)
- General Engineering & Computer Science (AREA)
- Computer Graphics (AREA)
- Software Systems (AREA)
- Processing Or Creating Images (AREA)
Abstract
The invention has provided a kind of linear very little parametric clothes model preparation method, utilize the video sequence including personage, rebuild the rough model of the target, and on this basis, through the image segmentation technology, cut apart human body part and clothes part, separate the model building to clothes and human body separately, and then get the double-deck model with higher precision and higher sense of reality, the invention proposes an optimization method based on the above-mentioned scheme at the same time, and a parametric scheme to clothes, trousers, can realize according to three-dimensional data and two-dimensional image and optimize the geometric parameter of the clothes, trousers dynamically in real time; the scheme of the invention fully utilizes the robustness provided by the existing deep learning scheme of static reconstruction, and completes the work of parameterization, semantization and the like in later-stage improvement and optimization so as to complete dynamic reconstruction.
Description
Technical Field
The invention relates to computer vision, three-dimensional reconstruction and computer graphics.
Background
At present, in the fields of human body identification, segmentation and the like, mature algorithms exist, for example, the current human body picture segmentation algorithms of the optimal performance, i.e., the CIHP _ PGN and the LIP _ JPPNet, can segment a human body part from a background through a two-dimensional picture or a video and the like, and can segment the human body into a plurality of parts, i.e., hair, face, clothes, arms, trousers, feet and the like, so that the accuracy of most common scenes can be higher, but for a few challenging scenes, the algorithm fails, and causes low precision and low sense of reality, and therefore, the algorithm can only be applied to a certain extent as general application.
Meanwhile, in the aspect of human body static three-dimensional reconstruction, algorithms with high robustness, such as PiFu and PAIR, are available, the former completes a single RBG picture as input through training a depth network, and a reconstructed three-dimensional model is used as output. The latter is improved on the basis of the former, and the estimation of the SMPL human posture and shape is added, so that the algorithm is more suitable for the three-dimensional reconstruction of the human body.
The project is a research result based on the basic maturity of the related algorithm.
Disclosure of Invention
The first purpose of the invention is to provide a method for making a linear differentiable parameterized garment model with higher precision and higher sense of realism;
a second object of the invention is to provide a method for optimizing parameters of a method for making a linearly differentiable parameterized garment model that optimizes three-dimensional geometry and corresponding two-dimensional geometric parameters.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
the method for making the linear micro parametric clothes model comprises the following steps:
s1, the plate sheet of the clothes is completely determined by 11 parameters, the parameters completely determine 16 nodes, the symmetry is met, the symmetry about a vertical central axis is ensured, the 16 nodes divide the whole plate sheet into 8 sub-plate sheets, namely, 11 parameters in total, and the geometric size and the node position of the clothes are completely determined;
s2, the plate pieces of the trousers are completely determined by 7 parameters, the parameters completely determine 10 nodes, the symmetry is met, the symmetry about a vertical central axis is ensured, the 10 nodes divide the whole plate piece into 4 sub-plate pieces, namely, the total 7 parameters, and the geometric size and the node position of the trousers are completely determined;
s3, filling the encrypted contour with a triangular surface patch by using a Delaunay triangulation algorithm to obtain a final clothes plate;
s4, calculating corresponding (u, v) coordinates and shape function values (N1, N2, N3 and N4) of each patch vertex.
The parameter optimization method of the linear micro parametric clothing model manufacturing method comprises the following steps:
s1, designing a loss function on the basis of obtaining a reasonable inner-layer human body model, and optimizing three-dimensional geometry (coordinates of all vertexes) of the garment through gradient descent;
s2, designing a loss function while optimizing three-dimensional geometry of the garment, and optimizing two-dimensional geometric parameters of the garment through gradient descent;
and S3, in the optimization process, rapidly detecting whether the clothes and the inner-layer human body collide or not, and specially processing the collision part.
Further, the collision detection described in S3 above calculates the minimum normal distance between the point to be processed and the inner layer human body, and when the minimum normal distance is smaller than a given threshold, adds a set gradient to the point to be processed, the direction of which is the corresponding normal direction, and acts on the final gradient descent.
The invention has the beneficial effects that: the method utilizes a video sequence containing a character to reconstruct a rough model of an object, on the basis, a human body part and a clothes part are segmented by an image segmentation technology, and the clothes and the human body are respectively modeled to obtain a double-layer model with higher precision and higher reality, thereby realizing the first purpose of the method; meanwhile, the invention provides a new optimization method and a parameterization scheme for clothes and trousers, fully utilizes the robustness provided by the existing deep learning scheme of static reconstruction, and completes the work of parameterization, semantization and the like in later improvement and optimization to complete dynamic reconstruction, thereby realizing real-time dynamic optimization of geometric parameters of the clothes and the trousers according to three-dimensional data and two-dimensional images, and further realizing the second aim of the invention.
Detailed Description
Example one
The method for making the linear micro parametric clothes model comprises the following steps:
s1, the plate sheet of the clothes is completely determined by 11 parameters, the parameters completely determine 16 nodes, the symmetry is met, the symmetry about a vertical central axis is ensured, the 16 nodes divide the whole plate sheet into 8 sub-plate sheets, namely, 11 parameters in total, and the geometric size and the node position of the clothes are completely determined;
s2, the plate pieces of the trousers are completely determined by 7 parameters, the parameters completely determine 10 nodes, the symmetry is met, the symmetry about a vertical central axis is ensured, the 10 nodes divide the whole plate piece into 4 sub-plate pieces, namely, the total 7 parameters, and the geometric size and the node position of the trousers are completely determined;
s3, filling the encrypted contour with a triangular surface patch by using a Delaunay triangulation algorithm to obtain a final clothes plate;
s4, calculating corresponding (u, v) coordinates and shape function values (N1, N2, N3 and N4) of each patch vertex.
The parameter optimization method of the linear micro parametric clothing model making method comprises the following steps:
s1, designing a loss function on the basis of obtaining a reasonable inner-layer human body model, and optimizing three-dimensional geometry (coordinates of all vertexes) of the garment through gradient descent;
s2, designing a loss function while optimizing three-dimensional geometry of the garment, and optimizing two-dimensional geometric parameters of the garment through gradient descent;
and S3, in the optimization process, rapidly detecting whether the clothes and the inner-layer human body collide or not, and specially processing the collision part.
Further, the collision detection at S3 is: and calculating the minimum normal distance between the point to be processed and the inner layer human body, and adding a set gradient when the minimum normal distance is smaller than a given threshold value, wherein the direction is the corresponding normal direction, and the set gradient acts on final gradient descent.
Example two
Firstly, a video sequence of the same person is input, a total of P frames is used as an input image, the video needs to contain the whole body of the person, the whole body can be the front, the side or the back, and the action of the person can be various actions in a simple standing posture, but complex actions such as heel turning, falling down and the like are not included. For each frame, a corresponding coarse model is output through the neural network via the PAIR, and parameters corresponding to the SMPL model are output. I.e. a total of P models M _ i (i ═ 1,2, 3.., P). P sets of parameters (β, θ, r, t, s) _ i (i ═ 1,2, 3.., P), where β, θ correspond to shape and position parameters in the SMPL parameters, respectively, r, t correspond to the amount of rotation and translation of the model, respectively, and s corresponds to the scale.
The SMPL model is then optimized and a model M _ i (i ═ 1,2, 3.., P) is fitted. Firstly, based on an SMPL parameter output by PAIR as a priori, optimizing (beta, theta, r, t, s) _ i (i is 1,2,3,.. multidot., P) to obtain an SMPL model with approximately proper posture and body type, defining an offset b (6890 multiplied by 3) for each point on the SMPL model, namely the offset of each vertex relative to the original position, and independently optimizing b to obtain the fitted SMPL model, so that a topologically consistent expression m _ i (i is 1,2, 3.. multidot., P) is obtained for all P models, wherein the P models are expressed by the SMPL parameter and the offset b.
Meanwhile, for each frame of the input video, the corresponding clothes/trousers, and parts of the human body are divided using the CIHP _ PGN and LIP _ JPPNet. Combining the P models M _ i (i 1,2, 3.. so, P) and the corresponding SMPL expressions M _ i (i 1,2, 3.. so, P), correspondingly segmenting on the corresponding SMPL models, counting the classification of the final vertices by marking the vertices of the models and "voting" the markers, and then obtaining the clothes parts Mc _ i (i 1,2, 3.., P) and the human body parts Ms _ i (i 1,2, 3.., P) expressed by the SMPL by finding the closest point of the fitted SMPL models and the rough models and directly transferring the classification labels, so as to obtain the clothes parts Mc _ i (i 1,2, 3.., P) expressed by the models and the human body parts Ms _ i (i 1,2, 3.. so, P).
The SMPL expression is then optimized again in a targeted manner by the segmented model to describe the inner body part S _ i (i ═ 1,2, 3.., P) of the model. And fitting corresponding clothes and trousers parts by using the parameterized models of the clothes and the trousers, simultaneously processing the problems of collision and contact, optimizing corresponding parameters of the clothes and the trousers, further optimizing to obtain final clothes and trousers models, and carrying out parameterized expression on the clothes and the trousers parts to obtain C _ i (i is 1,2,3,.. multidot.P). The method comprises the following specific steps:
first, to ensure that the garment, pants can be properly and reasonably optimized, "put on" the mannequin, and properly "sewn on," proper initialization work must be done. In the present invention, the manikin is first kept in the T-position and default body shape, and one piece of clothes or trousers is placed in front of and behind the manikin. Such a garment would consist of two panels of the same parameters, one front and one back, and the portions that should be "stitched" would be joined in an optimized process.
And then, optimizing the vertex coordinates of four plates of the clothes and the trousers, optimizing the corresponding geometric parameters of the plates, and gradually changing the posture and the body type of the human body model in the optimizing process. In addition, because the geometric parameters of clothes and trousers can only determine the vertex coordinates of the plate under the two-dimensional condition, and the model is three-dimensional in the actual optimization process, the invention provides a new optimization scheme to solve the problem, so that the gradient can be transmitted back to the geometric parameters of the plate. Firstly, a two-dimensional mesh needs to be additionally defined, and a vertex coordinate V is recordedc_2d，Vp_2dRespectively corresponding to the results of the clothes and the trousers models which are tiled on a two-dimensional plane. And simultaneously, adding an additional term into the loss function:
Ld_edge＝∑(||edge(Vc)-edge(Vc_2d)||2)+∑(||edge(Vp)-edge(Vp_2d)||2) (3.16)
the term is that the three-dimensional model can be the same as the corresponding two-dimensional grid as far as possible after being tiled, and the specific operation is that the edges of all the surface patches of the three-dimensional clothes and the trousers model are taken in sequence and are subjected to one-to-one difference with the edges of the corresponding two-dimensional grid, so that the total sum of the lengths of the models is obtained. The term can ensure that the tangential telescopic deformation is restrained by the shape of the corresponding two-dimensional plate when the clothes and trousers model in the three-dimensional space is continuously deformed, and the normal deformation is not restrained. The final result is that all the tangential expansion deformation of clothes and trousers models in the three-dimensional space is totally reflected on the two-dimensional plate, and the phenomena of folds and the like can be generated.
To optimize the garment, pant model, the overall loss function is then defined as follows:
Loss＝wcdLcd+wconstrLconstr+wd_edgeLd_edge+wlsLls+weLe+wnLn (3.17)
Lcd＝CD(clothes(mi),Vc)+CD(pants(mi),Vp) (3.18)
Lconstr＝∑(||constrait_front(Vc,Vp)-constrait_back(Vc,Vp)||2) (3.19)
Lcdis the chamfer distance of the garment and pants pattern and the garment and pants portion corresponding to the SMPL pattern, LconstrIt is a constraint on the point where the front and rear panels should be sewn.
The loss function is then minimized, using a random gradient descent. The object to be optimized now includes the vertex coordinates V of the model of the clothes and trousersc，VpAnd also contains the corresponding geometric parameter paramc，paramp. In the practical process, the two parameters should be optimized according to different gradient sizes, that is, assuming that the vertex coordinates of the optimized model are optimized, the gradient decreasing proportion is 1 each time, and when the corresponding geometric parameters are optimized, the gradient decreasing proportion is lr each timeg。
Meanwhile, since the clothes and the human body are necessarily in contact, collision detection and collision processing are necessarily required. Since the optimization is based on gradient descent to minimize the loss function, the invention provides a simple and fast collision detection and processing which can be integrated into the gradient descent process.
The invention assumes that the collision thickness r exists on the surface of the human body model, when optimizing each time, firstly, the projection t of the distance between the point on the clothing to be optimized and the nearest point in the human body model along the normal direction is judged, when t is less than r, the collision is considered to occur, and an additional gradient d along the normal direction needs to be added. Therefore, in the re-optimization process, each point coordinate to be optimized not only obtains the gradient of the back propagation of the loss function, but also has an additional gradient, and the expression is as follows:
λ(X)＝if(NN_normal(X)·(X-NN(X)))＜r)
wherein NN _ normal () represents the normal vector for finding the point corresponding to the nearest point, and X represents a point to be optimized in the clothes and trousers model.
And optimizing according to the gradient after collision detection and processing to obtain the result of the final clothes and trousers model, and obtaining the optimized geometric dimension of the clothes through corresponding two-dimensional grids.
Optimizing the garment model individually for each frame consumes a lot of time, so the solution of the present invention is to select one of the frames, typically the first frame, as a reference frame, and the process of wearing clothes and trousers and the process of optimizing the geometric parameters of clothes and trousers are done only on the reference frame. For other frames, the optimization of the residual frame can be quickly completed only by taking the optimization result of the reference frame as a priori and initializing. Because the character poses in a video sequence are gradual over time (number of frames), the best solution is to optimize frame by frame in the order of the video sequence. In order to ensure the continuity between two continuous frames. Only one of the following needs to be added to the overall loss function:
Lnn＝||mi(NN_index(mi-1,Vc))-Vc||2+||mi(NN_index(mi-1,Vp))-Vp||2 (3.21)
where NN _ index (X, Y) represents the set of indices of the closest point of each point of the set of points Y in X. The item can quickly and uniformly change the clothing model correspondingly with the change of the posture of the human body model.
And finally, combining the parameterized inner-layer human body part and the parameterized clothes part to obtain the geometric representation of the reconstructed model. And then, the texture of the model is estimated by utilizing the input video and monitoring the image obtained by directly rendering the model, and finally, a vivid reconstruction effect can be realized when the model is rendered again, and the input viewpoint is accurately reproduced.
The above description is not intended to limit the technical scope of the present invention, and any modification, equivalent change and modification of the above embodiments according to the technical spirit of the present invention are still within the technical scope of the present invention.
Claims (3)
1. The method for manufacturing the linear differentiable parameterized garment model is characterized by comprising the following steps of: the method comprises the following steps:
s1, the plate sheet of the clothes is completely determined by 11 parameters, the parameters completely determine 16 nodes, the symmetry is met, the symmetry about a vertical central axis is ensured, the 16 nodes divide the whole plate sheet into 8 sub-plate sheets, namely, 11 parameters in total, and the geometric size and the node position of the clothes are completely determined;
s2, the plate pieces of the trousers are completely determined by 7 parameters, the parameters completely determine 10 nodes, the symmetry is met, the symmetry about a vertical central axis is ensured, the 10 nodes divide the whole plate piece into 4 sub-plate pieces, namely, the total 7 parameters, and the geometric size and the node position of the trousers are completely determined;
s3, filling the encrypted contour with a triangular surface patch by using a Delaunay triangulation algorithm to obtain a final clothes plate;
s4, calculating corresponding (u, v) coordinates and shape function values (N1, N2, N3 and N4) of each patch vertex.
2. Method for the optimization of parameters of a linear differentiable parametric garment modeling method according to claim 1, characterized in that: the method comprises the following steps:
s1, designing a loss function on the basis of obtaining a reasonable inner-layer human body model, and optimizing three-dimensional geometry (coordinates of all vertexes) of the garment through gradient descent;
s2, designing a loss function while optimizing three-dimensional geometry of the garment, and optimizing two-dimensional geometric parameters of the garment through gradient descent;
and S3, in the optimization process, rapidly detecting whether the clothes and the inner-layer human body collide or not, and specially processing the collision part.
3. Method for the optimization of parameters of a linear differentiable parametric garment modeling method according to claim 2, characterized in that: and S3, calculating the minimum normal distance between the point to be processed and the inner layer human body, adding a set gradient when the minimum normal distance is smaller than a given threshold value, wherein the direction is the corresponding normal direction, and the set gradient acts on the final gradient descent.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010828692.4A CN112102470A (en) | 2020-08-18 | 2020-08-18 | Linear microminiaturible parametric clothing model manufacturing method and parameter optimization method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010828692.4A CN112102470A (en) | 2020-08-18 | 2020-08-18 | Linear microminiaturible parametric clothing model manufacturing method and parameter optimization method thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112102470A true CN112102470A (en) | 2020-12-18 |
Family
ID=73753790
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010828692.4A Pending CN112102470A (en) | 2020-08-18 | 2020-08-18 | Linear microminiaturible parametric clothing model manufacturing method and parameter optimization method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112102470A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112652057A (en) * | 2020-12-30 | 2021-04-13 | 北京百度网讯科技有限公司 | Method, device, equipment and storage medium for generating human body three-dimensional model |
| CN113223152A (en) * | 2021-05-14 | 2021-08-06 | 浙江大学 | Method for automatic pose and wrinkle migration for three-dimensional garment model aesthetic display |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101630417A (en) * | 2009-08-25 | 2010-01-20 | 东华大学 | Rapid posture-synchronizing method of three-dimensional virtual garment |
| US20100328311A1 (en) * | 2009-06-24 | 2010-12-30 | International Business Machines Corporation | system and method for establishing correspondence, matching and repairing three dimensional surfaces of arbitrary genus and arbitrary topology in two dimensions using global parameterization |
| CN102982578A (en) * | 2012-10-31 | 2013-03-20 | 北京航空航天大学 | Estimation method for dressed body 3D model in single character image |
| CN111311751A (en) * | 2020-02-12 | 2020-06-19 | 叠境数字科技（上海）有限公司 | Three-dimensional clothes model reconstruction method based on deep neural network |
-
2020
- 2020-08-18 CN CN202010828692.4A patent/CN112102470A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100328311A1 (en) * | 2009-06-24 | 2010-12-30 | International Business Machines Corporation | system and method for establishing correspondence, matching and repairing three dimensional surfaces of arbitrary genus and arbitrary topology in two dimensions using global parameterization |
| CN101630417A (en) * | 2009-08-25 | 2010-01-20 | 东华大学 | Rapid posture-synchronizing method of three-dimensional virtual garment |
| CN102982578A (en) * | 2012-10-31 | 2013-03-20 | 北京航空航天大学 | Estimation method for dressed body 3D model in single character image |
| CN111311751A (en) * | 2020-02-12 | 2020-06-19 | 叠境数字科技（上海）有限公司 | Three-dimensional clothes model reconstruction method based on deep neural network |
Non-Patent Citations (1)
| Title |
|---|
| STEPHEN MILLER等: ""Parametrized shape models for clothing"", 《2011 IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND AUTOMATION》, 18 August 2011 (2011-08-18), pages 4861 - 4868 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112652057A (en) * | 2020-12-30 | 2021-04-13 | 北京百度网讯科技有限公司 | Method, device, equipment and storage medium for generating human body three-dimensional model |
| CN112652057B (en) * | 2020-12-30 | 2024-05-07 | 北京百度网讯科技有限公司 | Method, device, equipment and storage medium for generating human body three-dimensional model |
| CN113223152A (en) * | 2021-05-14 | 2021-08-06 | 浙江大学 | Method for automatic pose and wrinkle migration for three-dimensional garment model aesthetic display |
| CN113223152B (en) * | 2021-05-14 | 2022-07-12 | 浙江大学 | Method for automatic pose and wrinkle migration for three-dimensional garment model aesthetic display |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10546433B2 (en) | Methods, systems, and computer readable media for modeling garments using single view images | |
| CN109147048B (en) | Three-dimensional mesh reconstruction method by utilizing single-sheet colorful image | |
| US20220189095A1 (en) | Method and computer program product for producing 3 dimensional model data of a garment | |
| US10867453B2 (en) | Method and system for generating an image file of a 3D garment model on a 3D body model | |
| Wang et al. | Design automation for customized apparel products | |
| De Aguiar et al. | Marker-less deformable mesh tracking for human shape and motion capture | |
| Starck et al. | Model-based multiple view reconstruction of people | |
| US20180197331A1 (en) | Method and system for generating an image file of a 3d garment model on a 3d body model | |
| CN105006014B (en) | The realization method and system that virtual clothing Fast simulation is tried on | |
| Wang et al. | Virtual human modeling from photographs for garment industry | |
| CN109003325A (en) | A kind of method of three-dimensional reconstruction, medium, device and calculate equipment | |
| CN111951381B (en) | Three-dimensional face reconstruction system based on single face picture | |
| CN111951384A (en) | Three-dimensional face reconstruction method and system based on single face picture | |
| CN112784621A (en) | Image display method and apparatus | |
| CN108564619B (en) | Realistic three-dimensional face reconstruction method based on two photos | |
| CN112102470A (en) | Linear microminiaturible parametric clothing model manufacturing method and parameter optimization method thereof | |
| Zheng et al. | Editablenerf: Editing topologically varying neural radiance fields by key points | |
| CN103049929A (en) | Multi-camera dynamic scene 3D (three-dimensional) rebuilding method based on joint optimization | |
| CN116416407A (en) | Method, system, equipment and medium for processing collision and self-collision in clothes simulation | |
| Li et al. | Robust 3D human motion reconstruction via dynamic template construction | |
| Zell et al. | Elastiface: Matching and blending textured faces | |
| Cushen et al. | Markerless real-time garment retexturing from monocular 3d reconstruction | |
| CN107895398B (en) | Relief editing method combined with visual saliency | |
| Chen et al. | Character animation creation using hand-drawn sketches | |
| Guo et al. | Mesh-guided optimized retexturing for image and video |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 518000 717, building r2-a, Gaoxin industrial village, No. 020, Gaoxin South seventh Road, Gaoxin community, Yuehai street, Nanshan District, Shenzhen, Guangdong Applicant after: Yiyun Zhixing (Shenzhen) Technology Co.,Ltd. Address before: 518000 1403a-1005, east block, Coast Building, No. 15, Haide Third Road, Haizhu community, Yuehai street, Nanshan District, Shenzhen, Guangdong Applicant before: Yiyun Zhixing (Shenzhen) Technology Co.,Ltd. |