A Defect Detection System for Anime Character Modeling

Abstract

The present invention relates to the technical field of three-dimensional modeling, and particularly relates to a defect detection system for anime character modeling, including: a storage module for editing the motion coding of the character model and storing the motion coding of the character model; a capture module for selecting the motion coding of the character model stored in the storage module and driving the motion of the character model based on the motion coding of the character model to capture the three-dimensional animation of the character model; a disassembly module for receiving the three-dimensional animation of the character model captured by the capture module and disassembling the character model in the three-dimensional animation of the character model. The present invention provides the driving conditions for the constructed character model by editing the motion coding of the character model, enabling defect detection of the character model in a motion state. Compared with the existing static defect detection technology for character modeling, its detection accuracy is higher, the detection effect is more comprehensive, it can detect character modeling defects that cannot be detected by static detection, and effectively improves the compatibility of character modeling.

Description

Technical Field

[0001] The present invention relates to the technical field of 3D modeling, and particularly relates to a defect detection system for anime character modeling. Background Art

[0002] Anime character modeling constructs a 3D image based on character settings through professional software and technologies. Starting from building an accurate bone framework, carefully carving the facial features, body proportions and other appearance features of the character, and then endowing rich materials and colors to make it have a realistic texture and unique style. Whether it is a cute and adorable character or a cool and domineering one, it can be vividly presented in the virtual world.

[0003] The invention patent with the application number 202311001222.0 discloses a method for detecting defects in a real-scene 3D model, which is characterized in that it includes: obtaining a real-scene 3D model; the real-scene 3D model includes real-scene 3D model data with a half-edge data structure and texture pictures; obtaining the external camera parameters, and performing off-screen rendering on the real-scene 3D model based on the external camera parameters to obtain corresponding two-dimensional images and depth maps; obtaining a deep learning data set, preprocessing the deep learning data set to obtain a training set, and training a preset deep learning model based on the training set to obtain a defect detection model; the deep learning model is a Unetformer network model, the Unetformer network includes an Encoder part and a Decoder part, the Encoder part is a Resnet-18 network, and the Decoder part adopts an attention mechanism based on Transformer; inputting the two-dimensional image and the depth map into the defect detection model to obtain the two-dimensional coordinates of the image plane corresponding to the defect point cloud, converting the two-dimensional coordinates of the image plane corresponding to the defect point cloud into three-dimensional coordinates in the real-scene 3D model and coloring the defect point cloud to obtain a defect patch: the obtaining of the external camera parameters includes: performing Poisson disk sampling on the real-scene 3D model to obtain corresponding sampling points; determining the external camera parameters based on the positions of the sampling points, the rendering distance and the normal unit vectors of the sampling points: the external camera parameters include the position and rendering direction of the rendering camera.

[0004] This application aims to solve the problems that: "Due to the complex format and large amount of data of the real-scene 3D model, there are the following deficiencies in manual quality inspection tasks: 1) The workload is huge, causing a great burden on vision; 2) Different staff have different standards for defect detection; 3) It is easy to produce the phenomenon of missed detection. At present, most are for the texture quality evaluation and defect detection in the model modeling and texturing process, and there is less research on defect detection of the models automatically generated by modeling software. Therefore, the current method for detecting defects in real-scene 3D models has the problems of huge detection workload, low efficiency and easy missed detection."

[0005] However, in the process of constructing a 3D character model, it is difficult to take into account the actions that the character needs to complete in the later stage. Therefore, it is difficult for the current character modeling defect detection technology to achieve dynamic defect detection.

[0006] For this reason, a defect detection system for anime character modeling is proposed. Summary of the Invention

[0007] Aiming at the above-mentioned shortcomings of the existing technology, the present invention provides a defect detection system for anime character modeling, which solves the technical problems proposed in the above background technology.

[0008] To achieve the above objectives, the present invention is realized through the following technical solutions:

[0009] A defect detection system for anime character modeling, comprising:

[0010] A storage module, used for editing the motion encoding of the character model and storing the motion encoding of the character model; a capture module, used for selecting the motion encoding of the character model stored in the storage module, driving the motion of the character model based on the motion encoding of the character model to capture the 3D animation of the character model; a disassembly module, used for receiving the 3D animation of the character model captured by the capture module, disassembling the character model in the 3D animation of the character model, so that the limbs of the character model in the 3D animation of the character model are separated at each joint position; a hiding module, used for selecting a limb model as a non-hidden target on the disassembled character model, taking the remaining unselected limb models as hidden targets, and hiding them in the 3D animation of the character model; an acquisition module, used for obtaining the 3D animation of the character model with only the non-hidden target limb model, playing the 3D animation of the character model, and acquiring the motion space defined by the motion trajectory of the non-hidden target limb model based on the played 3D animation of the character model; an analysis module, used for receiving the motion space collected by the acquisition module, placing all the received motion spaces in the 3D space where the 3D animation of the character model is located, and analyzing the modeling defect tendency of the character model based on the placement result of the motion space in the 3D space; a determination module, used for receiving the analysis result of the modeling defect tendency of the character model in the analysis module, and determining whether the character model has modeling defects by applying the analysis result; a feedback module, used for receiving the determination result in the determination module and feedbacking it to the system terminal user.

[0011] Furthermore, the motion encoding of the character model stored in the storage module is manually edited by the system terminal user, and the motion encoding of the character model includes: several groups of continuous timestamps, several groups of position coordinates are bound to each group of timestamps, and each group of position coordinates is set with a distinguishing mark;

[0012] Each difference marker corresponds to a limb joint on the human model. The human model is in a three-dimensional space and is placed based on the set of human model motion encodings with the earliest timestamp in the human model motion encoding. Further, based on the timestamp order of the human model motion encoding, continuous translation transformation is performed on the position coordinates of the same marker to capture the three-dimensional animation of the human model;

[0013] Among them, after the capture module captures the three-dimensional animation of the human model, it synchronously stores the three-dimensional animation of the human model. The difference marker set for the position coordinates is either a number or the literal name of a limb joint.

[0014] Furthermore, a sub-module is set under the storage module, including:

[0015] An upload unit for uploading the human model;

[0016] A marking unit for marking each human model motion encoding, and the marking content is another set of human model motion encodings in the storage module that has the highest similarity with the human model motion encoding;

[0017] Among them, the human model uploaded by the upload unit is stored synchronously in the storage module. During the operation stage of the marking unit, the similarity between each human model motion encoding is synchronously identified, and further based on the similarity identification result, a marking operation is performed.

[0018] Furthermore, the similarity identification logic for the human model motion encoding in the marking unit is as follows:

[0019] Receive the human model motion encodings stored in the storage module, select any two human model motion encodings to perform similarity identification, so that all combinations of pairwise human model motion encodings perform similarity identification operations:

[0020]

[0021] In the formula: SIMM(A,B) is the similarity between human model motion encoding A and human model motion encoding B; n is the set of vectors in the human model motion encoding; A i is the value of the i-th vector in human model motion encoding A; B i is the value of the i-th vector in human model motion encoding B;

[0022] Among them, the larger the value of SIMM(A,B), the higher the similarity between the two sets of human model motion encodings.

[0023] Furthermore, when the capture module operates and selects the human model motion encoding, it follows:

[0024] S1: Set the number of selected character model motion encodings, and select an equal number of character model motion encodings from the character model motion encodings stored in the storage module;

[0025] S2: Traverse the marked character model motion encodings of each character model motion encoding, and identify whether the marked character model motion encoding of the character model motion encoding exists in the previously selected character model motion encodings;

[0026] S3: Discard the character model motion encodings determined to exist, and select the number of discarded character model motion encodings from the storage module again;

[0027] S4: Jump to S2;

[0028] S5: Repeat S2 - S4 until the number of selected character model motion encodings is equal to the set number of selected character model motion encodings and then end.

[0029] Furthermore, in the construction stage of the character model, by independently constructing each limb model of the character model and then splicing the independently constructed limb models, the character model is obtained;

[0030] The acquisition module runs synchronously with the hidden module. The hidden module runs continuously in the system. The number of continuous runs of the hidden module in the system is equal to the number of limb models. During the continuous running of the hidden module, the non - hidden targets selected each time are different;

[0031] The motion space defined by the non - hidden target limb model in the acquisition module is obtained in the 3D animation of the character model. After the acquisition module acquires the motion space of the non - hidden target limb model, it synchronously stores the motion space of the non - hidden target limb model;

[0032] Among them, the motion space of the non - hidden target limb model is a set of three - dimensional space models.

[0033] Furthermore, the modeling defect tendency analysis logic of the character model in the analysis module is expressed as:

[0034]

[0035] In the formula: F is the modeling defect tendency value; j, j + 1 are two adjacent motion spaces;

[0036] Among them, when the intersection of two adjacent motion spaces j, j + 1 is an empty set, the modeling defect tendency value F is zero. In other cases, the modeling defect tendency value F is 1. Based on the above formula, all adjacent two motion spaces are calculated.

[0037] Further, a sub-module is provided under the determination module, including:

[0038] A sniffing unit for sniffing defective limb models in the character model;

[0039] Among them, during the operation stage of the sniffing unit, the motion spaces with all modeling defect tendency values F being 1 are obtained, and further, the limb model to which the intersection source motion space of the two motion spaces belongs is determined as the defective limb model, that is, the defective limb model in the sniffed character model.

[0040] Further, the feedback module is connected to the computer device that plays the three-dimensional animation of the character model through a wireless network, with the computer device as the feedback target;

[0041] Among them, the content fed back by the feedback module to the system-side user also includes the limb model pointed to by the sniffing result of the sniffing unit.

[0042] Further, an upload unit and a marking unit are connected to the lower level of the storage module through wireless network interaction. The storage module is connected to a capture module, a decomposition module, a hiding module, an analysis module, and a determination module through wireless network interaction. The determination module is connected to a sniffing unit through wireless network interaction at the lower level, and the determination module is connected to a feedback module through wireless network interaction.

[0043] Adopting the technical solution provided by the present invention, compared with the known public technology, it has the following beneficial effects:

[0044] The present invention provides a defective detection system for anime character modeling. During the operation of the system, the driving conditions of the character model are provided by editing the motion codes of the character model, so that the character model is defect-detected in the motion state. Compared with the existing static defective detection technology for character modeling, its detection accuracy is higher, the detection effect is more comprehensive, and it can detect the defective character modeling that cannot be detected by static detection. Thus, the compatibility of character modeling is improved, providing a greater optimization space for the subsequent motions and actions of character modeling. At the same time, the system can also sniff the limb models on the character model with modeling defects, thereby assisting the modeling user to improve the character model more quickly. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0046] Figure 1It is a schematic structural diagram of an anime character modeling defect detection system. Specific implementation manners

[0047] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0048] The present invention will be further described below with reference to embodiments.

[0049] Embodiment 1:

[0050] An anime character modeling defect detection system according to this embodiment, as Figure 1 shown, includes:

[0051] A storage module, configured to edit the motion codes of the character models and store the motion codes of the character models;

[0052] A sub-module is provided below the storage module, including:

[0053] An upload unit, configured to upload the character models;

[0054] A marking unit, configured to mark the motion codes of each character model, and the marking content is another set of motion codes of the character models in the storage module that has the highest similarity to the motion codes of the character models;

[0055] Among them, the character models uploaded by the upload unit are stored synchronously in the storage module. During the operation of the marking unit, the similarities between the motion codes of each character model are synchronously recognized, and further based on the recognition result of the similarity, a marking operation is performed;

[0056] The similarity recognition logic for the motion codes of the character models in the marking unit is:

[0057] Receive the motion codes of the character models stored in the storage module, select any two motion codes of the character models to perform similarity recognition, so that all combinations of the motion codes of the character models in pairs perform similarity recognition operations:

[0058]

[0059] In the formula: SIMM(A,B) is the similarity between the motion code A of the character model and the motion code B of the character model; n is the set of vectors in the motion code of the character model; A i is the value of the i-th vector in the motion code A of the character model; B iIt is the value of the i-th vector in the motion encoding B of the character model;

[0060] Among them, the larger the value of SIMM(A,B), the higher the similarity between the two sets of motion encodings of the character models;

[0061] Through the setting of the above logical formula, a specified marking logic is provided for the marking of the motion encoding of the character model.

[0062] The capture module is used to select the motion encoding of the character model stored in the storage module, drive the motion of the character model based on the motion encoding of the character model, so as to capture the 3D animation of the character model;

[0063] When the capture module runs and selects the motion encoding of the character model, it obeys:

[0064] S1: Set the number of selected motion encodings of the character model, and select an equal number of motion encodings of the character model from the motion encodings of the character model stored in the storage module;

[0065] S2: Traverse the marked motion encodings of each character model motion encoding, and identify whether the marked motion encodings of the character model motion encoding exist in the previously selected motion encodings of the character model;

[0066] S3: Discard the determined existing motion encodings of the character model, and select the number of discarded motion encodings from the storage module again;

[0067] S4: Jump to S2;

[0068] S5: Repeat S2 - S4 until the number of selected motion encodings of the character model is equal to the set number of selected motion encodings of the character model and then end;

[0069] The disassembly module is used to receive the 3D animation of the character model captured by the capture module, disassemble the character model in the 3D animation of the character model, so that the limbs of the character model in the 3D animation of the character model are separated at each joint position;

[0070] The hiding module is used to select a limb model on the disassembled character model as the non-hidden target, and use the remaining unselected limb models as the hidden targets to hide them in the 3D animation of the character model;

[0071] The acquisition module is used to obtain the 3D animation of the character model with only the non-hidden target limb model, play the 3D animation of the character model, and acquire the motion space limited by the motion trajectory of the non-hidden target limb model based on the played 3D animation of the character model;

[0072] An analysis module, configured to receive the motion spaces collected by the collection module, place all the received motion spaces in the three-dimensional space where the three-dimensional animation of the character model is located, and analyze the tendency of modeling defects of the character model based on the placement results of the motion spaces in the three-dimensional space;

[0073] The analysis logic for the tendency of modeling defects of the character model in the analysis module is expressed as:

[0074]

[0075] In the formula: F is the value of the tendency of modeling defects; j and j + 1 are two adjacent motion spaces;

[0076] Wherein, when the intersection of two adjacent motion spaces j and j + 1 is an empty set, the value of the tendency of modeling defects F is zero; in other cases, the value of the tendency of modeling defects F is 1. Based on the above formula, all adjacent two motion spaces are calculated;

[0077] Through the setting of the above logical formula, the system in this embodiment analyzes the tendency of modeling defects of the character model, providing necessary operation data support for the further operation of the determination module in the system.

[0078] A determination module, configured to receive the analysis result of the tendency of modeling defects of the character model in the analysis module, and determine whether there are modeling defects in the character model by applying the analysis result;

[0079] There are sub-modules set under the determination module, including:

[0080] A sniffing unit, configured to sniff the defective limb model in the character model;

[0081] Wherein, during the operation stage of the sniffing unit, all motion spaces with the value of the tendency of modeling defects F being 1 are obtained, and further the limb model to which the intersection source motion space of the two motion spaces belongs is determined as the defective limb model, that is, the defective limb model in the character model sniffed;

[0082] A feedback module, configured to receive the determination result in the determination module and feedback to the system-end user;

[0083] There is an upload unit and a marking unit connected to the lower level of the storage module through wireless network interaction. The storage module is connected to a capture module, a decomposition module, a hiding module, an analysis module, and a determination module through wireless network interaction. The lower level of the determination module is connected to a sniffing unit through wireless network interaction. The determination module is connected to a feedback module through wireless network interaction.

[0084] In this embodiment, the storage module runs to edit the motion encoding of the character model, stores the motion encoding of the character model, the uploading unit synchronously uploads the character model, the marking unit marks each motion encoding of the character model in real time, and the marking content is another set of motion encodings of the character model in the storage module that has the highest similarity with the motion encoding of the character model. The capturing module runs later to select the motion encoding of the character model stored in the storage module, drives the motion of the character model based on the motion encoding of the character model to capture the 3D animation of the character model. The disassembling module further receives the 3D animation of the character model captured by the capturing module, disassembles the character model in the 3D animation of the character model, separates the limbs of the character model in the 3D animation of the character model at each joint position, and then the hiding module selects a limb model as the non-hidden target on the disassembled character model, and uses the remaining unselected limb models as hidden targets to be hidden in the 3D animation of the character model. The acquisition module runs to obtain the 3D animation of the character model with only the non-hidden target limb model, plays the 3D animation of the character model, acquires the motion space defined by the motion trajectory of the non-hidden target limb model based on the played 3D animation of the character model, and then the analysis module receives the motion space acquired by the acquisition module, places all the received motion spaces in the 3D space where the 3D animation of the character model is located, analyzes the modeling defect tendency of the character model based on the placement result of the motion space in the 3D space, the determination module runs to receive the analysis result of the modeling defect tendency of the character model in the analysis module, and determines whether the character model has modeling defects by applying the analysis result. The sniffing unit synchronously sniffs the defective limb model in the character model, and finally receives the determination result in the determination module through the feedback module and feeds it back to the system-end user.

[0085] Through the system operation in the above embodiment, for the character model after modeling, in a dynamic detection manner, the positions on the surface of the character model with defects are detected, which more comprehensively ensures that during the later anime production process of the character model, there will be no defect problems such as model penetration in the motion of the character model, and effectively improves the accuracy of character modeling.

[0086] Embodiment 2:

[0087] At the specific implementation level, on the basis of Embodiment 1, this embodiment refers to Figure 1 to further specifically describe an anime character modeling defect detection system in Embodiment 1:

[0088] The motion encoding of the character model stored in the storage module is manually edited by the system-end user. The motion encoding of the character model includes: several groups of continuous timestamps, several groups of position coordinates are bound to each group of timestamps, and a discrimination mark is set for each group of position coordinates;

[0089] Each difference marker corresponds to a limb joint on the character model. The character model is in a three-dimensional space and is placed based on the set of character model motion encodings with the earliest timestamp in the character model motion encoding. Further, based on the timestamp order of the character model motion encoding, continuous movement transformation is performed on the position coordinates of the same marker to capture the three-dimensional animation of the character model;

[0090] Among them, after the capture module captures the three-dimensional animation of the character model, it synchronously stores the three-dimensional animation of the character model. The difference markers set for the position coordinates can be either a number or the literal name of a limb joint.

[0091] Through the above settings, the storage operation performed by the character model in the storage module is further defined, providing more indicative data support for the subsequent module operation in Embodiment 1 and ensuring the stable operation of the system.

[0092] Such as Figure 1 As shown, during the construction phase of the character model, each limb model of the character model is independently constructed, and then the independently constructed limb models are spliced together to obtain the character model;

[0093] The acquisition module runs synchronously with the hiding module. The hiding module runs continuously in the system. The number of continuous runs of the hiding module in the system is equal to the number of limb models. During the continuous running of the hiding module, the non-hidden targets selected each time are different;

[0094] The motion space defined by the non-hidden target limb model in the acquisition module is obtained in the three-dimensional animation of the character model. After the acquisition module acquires the motion space of the non-hidden target limb model, it synchronously stores the motion space of the non-hidden target limb model;

[0095] Among them, the motion space of the non-hidden target limb model is a set of three-dimensional space models.

[0096] Through the above settings, further operation logic limitations are provided for the system operation in Embodiment 1, ensuring the stable operation of the system in Embodiment 1.

[0097] Such as Figure 1 As shown, the feedback module is connected to the computer device that plays the three-dimensional animation of the character model through a wireless network, using the computer device as the feedback target;

[0098] Among them, the content fed back by the feedback module to the system-side user also includes the limb model pointed to by the sniffing result of the sniffing unit.

[0099] Through the above settings, the interaction logic of the feedback module in the system in Embodiment 1 is further defined, ensuring that the operation result of the system in Embodiment 1 can be stably output and obtained by the system-side user, so as to optimize and improve the suitability of the character model.

[0100] In summary, during the operation of the system in the above embodiments, by editing the motion codes of the character model, the driving conditions of the constructed character model are provided, enabling the character model to perform defect detection in the motion state. Compared with the existing static defect detection technology for character modeling, its detection accuracy is higher and the detection effect is more comprehensive. It can detect character modeling defects that cannot be detected by static detection, thereby improving the compatibility of character modeling and providing a greater optimization space for the subsequent motions and actions of character modeling. At the same time, the system can also sniff the upper limb model of the character model with modeling defects, thus assisting the modeling user to improve the character model more quickly.

[0101] The above embodiments are only used to illustrate the technical solutions of the present invention, not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements will not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An anime character modeling defect detection system, characterized in that Including: A storage module, which is used to edit the movement coding of the character model and store the movement coding of the character model; The movement coding of the character model stored in the storage module is manually edited by the system-end user. The movement coding of the character model includes: several groups of consecutive timestamps, several groups of position coordinates are bound to each group of timestamps, and each group of position coordinates is set with a difference mark; Each difference mark corresponds to a limb joint on the character model. The character model is in a three-dimensional space. It is placed based on the group of character model movement coding with the earliest timestamp in the character model movement coding. Further, based on the timestamp order of the character model movement coding, continuous movement transformation is performed on the position coordinates with the same mark to capture the three-dimensional animation of the character model; Among them, after the capture module captures the three-dimensional animation of the character model, it synchronously stores the three-dimensional animation of the character model. The difference mark set for the position coordinates is any one of a number or the text name of a limb joint; A capture module, which is used to select the movement coding of the character model stored in the storage module and drive the movement of the character model based on the movement coding of the character model to capture the three-dimensional animation of the character model; When the capture module runs and selects the movement coding of the character model, it obeys: S1: Set the number of selected character model movement codings, and select an equal number of character model movement codings from the character model movement codings stored in the storage module; S2: Traverse the marked character model movement codings of each character model movement coding, and identify whether the marked character model movement coding of the character model movement coding exists in the previously selected character model movement coding; S3: Discard the character model movement coding determined to exist, and select the discarded number of character model movement codings from the storage module again; S4: Jump to S2; S5: Repeat S2 - S4 until the number of selected character model movement codings is equal to the set number of selected character model movement codings and then end; A disassembly module, which is used to receive the three-dimensional animation of the character model captured in the capture module, disassemble the character model in the three-dimensional animation of the character model, and separate the limbs of the character model in the three-dimensional animation of the character model at each joint position; A hiding module, which is used to select a limb model on the disassembled character model as the non-hidden target, and use the remaining unselected limb models as the hidden targets to hide them in the three-dimensional animation of the character model; An acquisition module, which is used to obtain the three-dimensional animation of the character model with only the non-hidden target limb model, play the three-dimensional animation of the character model, and acquire the movement space limited by the movement trajectory of the non-hidden target limb model based on the played three-dimensional animation of the character model; An analysis module, which is used to receive the movement space collected in the acquisition module, place all the received movement spaces in the three-dimensional space where the three-dimensional animation of the character model is located, and analyze the modeling defect tendency of the character model based on the placement result of the movement space in the three-dimensional space; A determination module, which is used to receive the analysis result of the modeling defect tendency of the character model in the analysis module and determine whether the character model has modeling defects by applying the analysis result; A feedback module, which is used to receive the determination result in the determination module and feedback it to the system-end user.

2. The anime character modeling defect detection system according to claim 1, characterized in that There is a sub-module set under the storage module, including: An upload unit for uploading a character model; A marking unit for marking the motion coding of each character model, and the marked content is another set of character model motion coding in the storage module that has the highest similarity with the motion coding of the character model; Among them, the character models uploaded by the upload unit are stored synchronously in the storage module. During the operation stage of the marking unit, the similarities between the motion codings of each character model are synchronously identified, and further based on the similarity recognition results, a marking operation is performed.

3. The anime character modeling defect detection system according to claim 2, characterized in that, The similarity recognition logic for the motion coding of the character model in the marking unit is as follows: Receive the motion coding of the character model stored in the storage module, select any two motion codings of the character model to perform similarity recognition, so that all combinations of the motion codings of the character model in pairs perform similarity recognition operations: ； In the formula: is the similarity between the motion code A of the character model and the motion code B of the character model; is the set of vectors in the motion code of the character model; is the value of the i-th vector in the motion code A of the character model; is the value of the i-th vector in the motion code B of the character model; Among them, The larger the value, the higher the similarity of the motion coding of the two groups of character models.

4. The anime character modeling defect detection system according to claim 1, characterized in that During the construction stage of the character model, each limb model of the character model is independently constructed, and then the independently constructed limb models are spliced to obtain the character model; The acquisition module runs synchronously with the hidden module. The hidden module runs continuously in the system. The number of continuous runs of the hidden module in the system is equal to the number of limb models. During the continuous running of the hidden module, the non-hidden targets selected each time are different; The motion space defined by the motion trajectory of the non-hidden target limb model in the acquisition module is obtained in the 3D animation of the character model. After the acquisition module acquires the motion space of the non-hidden target limb model, it synchronously stores the motion space of the non-hidden target limb model; Among them, the motion space of the non-hidden target limb model is a three-dimensional space model.

5. The anime character modeling defect detection system according to claim 1, characterized in that The modeling defect tendency analysis logic of the character model in the analysis module is expressed as: ； In the formula: is the modeling defect tendency value; , are two adjacent motion spaces; Among them, when the intersection of two adjacent motion spaces , is an empty set, the modeling defect tendency value is zero. In other cases, the modeling defect tendency value is 1. Based on the above formula, all adjacent motion spaces are calculated.

6. The defect detection system for anime character modeling according to claim 1, wherein, There is a sub-module set under the determination module, including: A sniffing unit for sniffing defective limb models in the character model; Among them, during the operation stage of the sniffing unit, all modeling defect tendency values are obtained For the motion space with a value of 1, further determine the limb model to which the intersection source motion space of the two motion spaces belongs as a defective limb model, that is, the defective limb model in the sniffed human model.

7. An anime character modeling defect detection system according to claim 1, characterized in that, The feedback module is connected to the computer device that plays the 3D animation of the character model through a wireless network, and uses the computer device as the feedback target; Among them, the content fed back by the feedback module to the system-end user also includes the limb model pointed to by the sniffing result of the sniffing unit during operation.

8. The anime character modeling defect detection system according to claim 1, characterized in that, The storage module is connected to the upload unit and the marking unit through wireless network interaction at the lower level. The storage module is connected to the capture module, the decomposition module, the hidden module, the analysis module and the determination module through wireless network interaction. The determination module is connected to the sniffing unit through wireless network interaction at the lower level. The determination module is connected to the feedback module through wireless network interaction.

Images