Two-dimensional dynamic image production method

Abstract

The present invention provides a two-dimensional dynamic image production method, comprising: determining an object behavior of a target object between a first original drawing and a second original drawing; separately inputting the first original drawing and the second original drawing into a first adversarial network model, and outputting first posture parameters and second posture parameters on the basis of first-type information in a natural motion pattern library; inputting the first posture parameters, the second posture parameters, and the object behavior into a second adversarial network model, and outputting all target body dynamic parameters of the target object at all given moments during completion of the object behavior on the basis of second-type information in the natural motion pattern library; and substituting the target body dynamic parameters into a preset modeling model to obtain a target object model. The beneficial effects are: by using artificial intelligence technology, two-dimensional dynamic images can be generated more efficiently, and more realistic and natural object shape changes and spatial movements can be simulated, thereby greatly improving the production efficiency and final product quality of two-dimensional dynamic images and reducing time consumption and limitations on labor and capital investment.

Description

A method for producing two-dimensional dynamic images Technical Field

[0001] The present invention relates to the technical field of two-dimensional dynamic images, and in particular to a method for producing two-dimensional dynamic images. Background Art

[0002] 2D motion graphics is a technique that creates a dynamic effect by rapidly switching between a series of static images. 2D motion graphics technology is widely used in a variety of fields, including film and television production, game development, and advertising design.

[0003] The traditional process for producing 2D motion graphics requires the artist to first draw, piece by piece, original drawings of the motion of a person or object at different points in time. Based on these original drawings, the artist then cleans up the lines, fills in any overlooked details, and creates in-betweens for the undrawn time points between adjacent time points. Next, all the original drawings and in-betweens are colored. Finally, the colored original drawings and in-betweens are played back sequentially along a timeline to create a dynamic visual effect. This process is time-consuming and requires significant manpower and financial investment, limiting the production capacity and quality of 2D motion graphics.

[0004] Existing animation software (such as Flash and CACAni) uses vector lines to derive the lines or graphics for intermediate animations. However, while this approach can quickly create blank intermediate sections and movement trajectories for well-defined lines or graphics, it still requires extensive manual fine-tuning to handle complex issues such as the absence of preceding and following lines or changes in the number of lines drawn. For example, when a character turns dramatically, the appearance and disappearance of limbs must be manually processed; while objects in motion must also handle splitting and merging. This significantly reduces the efficiency of automated software and makes the entire production process more complex and uncontrollable. Furthermore, the finished products produced using these software programs do not exhibit significant quality improvements and even lack the aesthetic appeal of pure hand-drawn animations. Therefore, these software programs are primarily used for small-scale individual productions or simple, child-oriented animations. For diverse animation types, the entire process, including line drawing, intermediate animation, and coloring, still relies heavily on manual labor.

[0005] Summary of the Invention

[0006] In order to solve the above technical problems, the present invention provides a two-dimensional dynamic image production method.

[0007] The technical problem solved by the present invention can be achieved by adopting the following technical solutions:

[0008] A method for producing two-dimensional dynamic images, pre-establishing a natural motion pattern library, wherein the natural motion pattern library pre-stores parameter information of different categories of objects, the parameter information including first-category information representing all body posture parameters of each of the objects at a fixed moment, and second-category information representing all body dynamic parameters of each of the objects at all given moments in the process of completing a predetermined action;

[0009] The two-dimensional dynamic image production method comprises:

[0010] Step S1, obtaining a first original picture and a second original picture, wherein both the first original picture and the second original picture contain the same target object, and determining an object behavior of the target object between the first original picture and the second original picture;

[0011] Step S2: inputting the first original image and the second original image into a pre-built first adversarial network model, respectively. The first adversarial network model is configured to output a first posture parameter of the target object in the first original image at a fixed time and a second posture parameter of the target object in the second original image at a fixed time based on the first type of information in the natural motion pattern library.

[0012] Step S3: inputting the first body posture parameter, the second body posture parameter, and the object behavior into a pre-built second adversarial network model, wherein the second adversarial network model is configured to output all target body dynamic parameters of the target object at all given moments in the process of completing the object behavior based on the second type of information in the natural motion pattern library;

[0013] Step S4: Substitute all the target body dynamic parameters output by the second adversarial network model into a preset modeling model to obtain a target object model.

[0014] Preferably, the method further comprises: pre-establishing a body posture parameter library, wherein the body posture parameter library pre-stores all callable body posture parameters of the objects of different categories without overall spatial motion;

[0015] Each of the objects includes a plurality of blocks;

[0016] The steps of establishing the natural motion pattern library include:

[0017] Step A1, capturing motion data of each subject in the body parameter library when performing a series of body parameters to complete a predetermined action;

[0018] Step A2, using the correlation ratio of the movement distance of all the body blocks of the object in the motion data to the actual measurement information of the body block features as the predetermined action to generate a series of dynamic parameters;

[0019] Step A3: define the predetermined action, and package the action definition, the body parameters and the dynamic parameters in chronological order to form the body dynamic parameters of each object and include them in the natural movement pattern library.

[0020] Preferably, step A3 also includes: assigning exaggerated variables to the body parameters and / or the dynamic parameters corresponding to the action definition, and packaging the exaggerated variables with the action definition, the body parameters and the dynamic parameters in chronological order to form the body dynamic parameters of each object and including them in the natural motion pattern library.

[0021] Preferably, each of the blocks has a corresponding block code;

[0022] The steps of establishing the body parameter library include:

[0023] Step B1, for each object, determining the connection relationship between all the blocks in the object;

[0024] Step B2, determining a movable spatial range of each of the blocks when fixed to one of the adjacent blocks, wherein the movable spatial range includes a plurality of block regions, each of the block regions having a corresponding region code;

[0025] Step B3: determining the movement information of each body block relative to a preset standard posture when the body block is located in each body block region, summarizing the body block code, the region code, and the movement information and including them in the body posture parameter library.

[0026] Preferably, after step S3 and before step S4, the following steps are further included:

[0027] Repeating steps S2-S3 multiple times, outputting an output result each time after executing steps S2-S3, wherein each output result includes all target body dynamic parameters of the target object at all given moments in the process of completing the object behavior;

[0028] The dynamic parameter range of the target body at each given moment is determined based on the multiple output results, and a set of target body dynamic parameters is selected and output from the dynamic parameter range of the target body at all given moments as all the target body dynamic parameters output by the second adversarial network model.

[0029] Preferably, the step S4 further comprises: adding texture and / or coloring to the target object model.

[0030] Preferably, the step S4 further includes:

[0031] Motion simulation is performed using the target object model, and screenshots are taken during the simulation process, and the screenshots are output as the first original picture, the second original picture, and an intermediate picture between the first original picture and the second original picture.

[0032] Preferably, the step S4 further includes:

[0033] The two intermediate pictures are used as the first original picture and the second original picture respectively, and the process returns to step S2.

[0034] Preferably, the first adversarial network model includes a first generative model and a first discriminative model;

[0035] Then in step S2, the first generation model is used to generate a plurality of first candidate items according to the input first original picture or the second original picture, each of the first candidate items includes the body parameters of the target object at a fixed moment;

[0036] The first discriminant model is connected to the first generative model, and is configured to sequentially evaluate the plurality of first candidate items output by the first generative model using the first type of information in the natural motion pattern library as a real sample and the body feature information expressed by the first original painting or the second original painting as a condition to obtain a first evaluation result; and

[0037] The first evaluation result is input into the first discrimination model again to determine whether the first candidate in the first evaluation result is the real sample, and the first candidate in the first evaluation result is output after it is determined to be the real sample.

[0038] Preferably, the second adversarial network model includes a second generative model and a second discriminative model;

[0039] Then, in step S3, the second generation model is used to generate a plurality of second alternative items according to the input first posture parameter, the second posture parameter, and the object behavior, each of the second alternative items including all target body dynamic parameters of the target object at all given moments in the process of completing the object behavior;

[0040] The second discriminant model is connected to the second generative model, and is configured to sequentially evaluate the plurality of second alternatives output by the second generative model using the second type of information in the natural motion pattern library as a true sample and the output of the first adversarial network model as a condition to obtain a second evaluation result; and

[0041] The second evaluation result is input into the second discrimination model again to determine whether the second candidate item in the second evaluation result is the true sample, and the second candidate item in the second evaluation result is output after it is determined to be the true sample.

[0042] The advantages or beneficial effects of the technical solution of the present invention are:

[0043] By adopting artificial intelligence technology, the present invention can generate two-dimensional dynamic images more efficiently, and can also simulate more realistic and natural object shape changes and spatial movements, greatly improving the production capacity and finished product effects of two-dimensional dynamic images, and reducing the time consumption and manpower and capital investment limitations. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] FIG1 is a flow chart of a method for producing a two-dimensional dynamic image in a preferred embodiment of the present invention;

[0045] FIG2 is a flow chart of steps for establishing a body parameter library in a preferred embodiment of the present invention;

[0046] FIG3 is a flow chart of steps for establishing a natural motion pattern library in a preferred embodiment of the present invention;

[0047] FIG4 is a diagram showing the principle of implementing a method for producing two-dimensional dynamic images in a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0048] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making any creative efforts shall fall within the scope of protection of the present invention.

[0049] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features therein may be combined with each other.

[0050] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0051] Referring to FIG1 , a preferred embodiment of the present invention, addressing the aforementioned problems existing in the prior art, provides a method for producing two-dimensional dynamic images. A natural motion pattern library is pre-established. The natural motion pattern library pre-stores parameter information of objects of different categories. The parameter information includes first-category information representing all body posture parameters of each object at a fixed moment, and second-category information representing all body dynamic parameters of each object at all given moments while completing a predetermined action.

[0052] The two-dimensional dynamic image production method includes:

[0053] Step S1, obtaining a first original picture and a second original picture, wherein both the first original picture and the second original picture contain the same target object, and determining an object behavior of the target object between the first original picture and the second original picture;

[0054] Step S2: Inputting the first original image and the second original image into a pre-built first adversarial network model, respectively. The first adversarial network model is configured to output a first posture parameter of the target object in the first original image at a fixed time and a second posture parameter of the target object in the second original image at a fixed time based on the first type of information in the natural motion pattern library.

[0055] Step S3: inputting the first body posture parameter, the second body posture parameter, and the object behavior into a pre-built second adversarial network model, wherein the second adversarial network model is configured to output all target body dynamic parameters of the target object at all given moments in the process of completing the object behavior based on the second type of information in the natural motion pattern library;

[0056] Step S4: Substitute all target body dynamic parameters output by the second adversarial network model into a preset modeling model to obtain a target object model.

[0057] Specifically, before creating a 2D dynamic image, parameter information for different object categories is first collected, including first-category information and second-category information. A natural motion pattern library is pre-established based on this parameter information. Objects may include, but are not limited to, people and objects. The first-category information represents all body parameters of each object at a fixed moment. Body parameters describe static characteristics of an object at that moment, such as posture, form, and position. They may include, but are not limited to, the object's position coordinates, rotation angle, size, and shape. Body parameters accurately represent the static state of the object at a given moment. The second-category information represents all dynamic body parameters of each object at all given moments while it completes a predetermined motion. Dynamic body parameters describe dynamic characteristics of an object, such as posture, form, and position, while completing a predetermined motion. Dynamic body parameters accurately represent the dynamic state of the object at different moments. By combining these parameters, the motion and changes of the object in the 2D dynamic image can be fully described, accurately simulating and presenting the object's motion trajectory and changes.

[0058] In the process of producing two-dimensional dynamic images, the first step is to obtain the original drawings drawn by the original artist. The original drawing refers to the first and most basic stage in the production process of two-dimensional dynamic images. In the original drawing stage, the original artist will create sketches of key frames and important actions based on the character setting drawings, storyboard design drafts (layouts) and storyboards. However, these sketches are usually rough, using simple lines and curves to outline the outline and main features of the target object, but they can express the character's movements. Original drawings play a vital role in the production of two-dimensional dynamic images, providing the basis for subsequent production, including the drawing of in-between pictures, color filling, background design, etc.

[0059] In this example, no more than one object is drawn in each original drawing. By subsequently setting up layers within the original drawing, each object can be easily edited and adjusted without affecting other objects. Furthermore, using layers allows for easy combination, separation, and rearrangement of different objects to achieve more complex drawing effects.

[0060] After obtaining the original drawings, the target object's behavior between the two original drawings can be determined. Object behavior refers to the change in the target object's action or state between the two original drawings. By observing and analyzing object behavior, we can better understand the changes and development of the target object between the two original drawings, and thus infer the target object's dynamics and intentions. Object behavior can include but is not limited to several aspects: movement, deformation, action, etc. Movement refers to the change in the position of the target object between the two original drawings, such as the target object moving, walking, flying, or changing position. Deformation refers to the change in the shape, size, or appearance of the target object, such as the target object undergoing a change, transformation, transformation, or morphological change. Action refers to the target object performing a specific action or behavior, such as the target object attacking, jumping, running, flying, or other specific actions.

[0061] Object behavior can be identified from the storyboard, the storyboard design, and the two original drawings; or it can be determined based on the annotations on the storyboard (primary) and the storyboard design (secondary). The storyboard and storyboard design are important documents in the animation production process. They show the layout and action instructions of each shot. Annotations are added to the storyboard and storyboard design to explain the object's behavior and action details.

[0062] Next, for each original image, the original image is input into the first adversarial network model. The first adversarial network model uses the first type of information in the natural motion pattern library to output the body posture parameters of the target object in the original image at a fixed time. Preferably, two original images are input into the first adversarial network model separately to obtain the first body posture parameters and the second body posture parameters.

[0063] Then, the first posture parameter, the second posture parameter and the object behavior are input into the second adversarial network model. The second adversarial network model uses the second type of information in the natural motion pattern library to output the target body dynamic parameters; the target body dynamic parameters refer to the set of all body dynamic parameters of the target object at all given moments in the process of completing the object behavior, and the set includes the posture parameters and dynamic parameters at each given moment.

[0064] All target body dynamic parameters output by the second adversarial network model are substituted into a preset modeling model. The modeling model can be an accurate model for different categories of objects, such as a swingable human body structure model. The modeling model includes various blocks of the target object and can be dynamically changed. By extracting the target body dynamic parameters output by the second adversarial network model and substituting these parameters into the modeling model, that is, applying the target body dynamic parameters to the corresponding blocks of the modeling model so that they meet the requirements of the modeling model, the generated target object model can be used to produce a two-dimensional dynamic image, for example, by presenting the target object model through rendering, animation and other technologies.

[0065] By adopting artificial intelligence technology, the above solution can generate two-dimensional dynamic images more efficiently, and can also simulate more realistic and natural changes in object shape and spatial movement, greatly improving the production capacity and finished product effects of two-dimensional dynamic images, and reducing the time consumption and labor and capital investment limitations.

[0066] As a preferred embodiment, the method further includes: pre-establishing a body posture parameter library, wherein the body posture parameter library pre-stores all callable body posture parameters of different categories of objects without overall spatial motion;

[0067] Each object includes a plurality of blocks, and each block has a corresponding block code;

[0068] As a preferred embodiment, as shown in FIG2 , the steps of establishing the body parameter library include:

[0069] Step B1, for each object, determining the connection relationship between all blocks in the object;

[0070] Step B2, determining a movable spatial range of each block when it is fixed to one of its adjacent connected blocks, the movable spatial range including a plurality of block regions, each of which has a corresponding region code;

[0071] Step B3: determining the movement information of each body block relative to a preset standard posture when it is located in each body block region, summarizing the body block code, region code, and movement information and including them in a body posture parameter library.

[0072] Specifically, in this embodiment, each object is composed of different blocks, each block is numbered, and the connection relationship between the blocks is clarified. The connection relationship can be a relationship of upper and lower connection, left and right connection, and front and back connection. Taking the human body as an example, the human body can be divided into body blocks such as the head, torso and limbs. The head includes parts such as the skull, face and neck, the torso includes parts such as the chest and abdomen, and the limbs include parts such as the upper limbs and lower limbs. In the human body, the head is connected to the neck, the neck is connected to the torso, and the torso is connected to the limbs. In addition, there is also a connection relationship between the limbs of the human body. For example, the upper limbs include parts such as the shoulders, arms, forearms and hands, and the lower limbs include parts such as the hips, thighs, calves and feet; in the upper limbs, the shoulders are connected to the arms, the arms are connected to the forearms, and the forearms are connected to the hands; in the lower limbs, the hips are connected to the thighs, the thighs are connected to the calves, and the calves are connected to the feet.

[0073] Next, the movable range of each block when attached to the block at one end is precisely divided, and each region is labeled with a number. For example, the movable range of the arm is roughly a hemisphere around the shoulder. This hemisphere is divided from the center into several arm regions. Medical anatomical knowledge can be used to further narrow the range of the arm regions. In this way, the movable range of all blocks in each object can be determined, as well as the region codes of all block regions within the movable range of each block.

[0074] Finally, the movement of each body block relative to a given standard posture within the movable space is compared across different body block regions. This movement information is then assigned to a number, resulting in a comprehensive set of callable body parameters for the subject without overall spatial motion. The standard posture can be the subject's standard standing position. For example, if the subject raises their arm sideways, the movement could be the arm's position relative to the standard standing position, with the arm rotated 90 degrees upwards and to the side of the subject.

[0075] By collecting all the body parameters of all categories of objects, the body parameter library can be established. The library is formed by formulating a prescribed expression method for the body parameters of the objects and can be used as the basic movable data of the stored objects.

[0076] As shown in Figure 3, the steps for establishing a natural motion pattern library include:

[0077] Step A1, capturing motion data of each subject in the body parameter library when performing a series of body parameters to complete a predetermined action;

[0078] Step A2, using the correlation ratios of the moving distances of all the blocks of the object in the motion data and the actual measured information of the block features as predetermined actions to generate a series of dynamic parameters;

[0079] Step A3: define the predetermined action, and package the action definition, body parameters and dynamic parameters in chronological order to form the body dynamic parameters of each object and include them in the natural movement pattern library.

[0080] Specifically, after formulating a prescribed expression method for the object's body parameters, the motion data of the object in the body parameter library is captured. The full dynamic (body + dynamic) motion data of the object when executing a series of body parameters in three-dimensional space to complete a predetermined action can be observed and captured using motion capture technology or game modeling and simulation technology;

[0081] Then, the correlation ratio between the movement distance captured when the object completes a predetermined action (i.e., a series of changes in body parameters) and the actual measurement information of the body block characteristics is used as a series of dynamic parameters generated by the predetermined action. The actual measurement information of the body block characteristics is recorded as the body block involved in the action, and the body block shape feature information that can be actually measured. As an example, assuming that the leg steps forward 30 centimeters as the movement distance of the body block, during the action, considering that at the same lifting angle, the longer the leg, the larger the step. The embodiment of the present invention uses the correlation ratio of the movement distance and the body characteristics as a dynamic parameter to extract the unchanging motion logic, thereby eliminating the influence of variable factors.

[0082] Next, a verbal definition is made for the current scheduled action. For example, walking can be defined as random, purposeless, unmarked, slow walking, fast running, etc., and the body parameters and dynamic parameters involved in the above action definition are sequenced by time and packaged to form the body dynamic parameters of each object. The natural motion pattern library contains all the body dynamic parameters of all objects when performing different scheduled actions.

[0083] As a preferred embodiment, step A3 also includes: assigning exaggerated variables to the body parameters and / or dynamic parameters corresponding to the action definition, and packaging the exaggerated variables with the action definition, body parameters and dynamic parameters in chronological order to form the body dynamic parameters of each object and including them in the natural motion pattern library.

[0084] Specifically, in this embodiment, for each predetermined action, in addition to being verbally defined, an exaggeration variable can also be assigned. The exaggeration variable can be constant, accelerated, or otherwise varied. For example, when the exaggeration variable is constant, the corresponding predetermined action can be performed directly according to the body and dynamic parameters involved in the action definition. When the exaggeration variable is accelerated, the body and dynamic parameters need to be adjusted based on the degree of acceleration to change the speed of the action. The exaggeration variable can also be a change in size, such as increasing or decreasing. Depending on the exaggeration variable, the body and dynamic parameters can be adjusted to achieve an exaggerated effect on the action.

[0085] The embodiment of the present invention uses a generative adversarial network (GAN) model to generate the body dynamic parameters required for the sample according to the corresponding needs, and repeatedly uses the body posture parameters in the body posture parameter library to repeatedly accurately determine the natural movement pattern of the object, so that it conforms to the spatial position and body shape characteristics to be expressed.

[0086] The GAN model iteratively trains a generative model (G) and a discriminator (D), and then lets the discriminator D try to determine whether the generated samples output by the generator model G are real samples. Based on the judgment results, the generator model G and the discriminator D are adjusted and upgraded respectively until the generated samples output by the generator model G can completely deceive the discriminator D.

[0087] In an embodiment of the present invention, the first adversarial network model can adopt a sequential adversarial network (SeqGAN) model, and the second adversarial network model can also adopt a sequential adversarial network model. In a single operation, the sequential adversarial network model can call the generative model G and the discriminative model D multiple times. As shown in Figure 4, the state State represents the generated token, and the action action represents the next token to be generated. The state behavior value is estimated by Monte Carlo search, and gradient training is performed using policy gradient. The discriminative model D is a CNN convolutional neural network used to evaluate the generated sequence to guide the learning of the generative model. Taking the generation of human body blocks as an example, each red circle in Figure 4 represents the operation of generating a body block, and State represents the generated body block string. In Figure 4, when generating the next body block Nextaction, the generative model G is first called to generate multiple options, and then the discriminative model D is used to score (reward) each option, select the best policy (Policy) based on the score, and adjust the policy model (Policy Gradient).

[0088] In the embodiments of the present invention, the first adversarial network model and the second adversarial network model can be set up separately or stacked to form a stacked adversarial network model (referred to as stacked GAN). Stacked GAN uses multiple GANs placed in a row, each of which can solve a simplified problem module, avoiding the possibility that a single GAN may not be able to effectively handle certain tasks, thereby enhancing the performance of the entire model.

[0089] The first adversarial network model serves as the first layer (stage 1) of the stacked GAN. The first layer is a local adversarial network. As a preferred embodiment, the first adversarial network model includes a first generative model and a first discriminative model.

[0090] Then in step S2, the first generation model is used to generate a plurality of first candidate items according to the input first original picture or second original picture, each first candidate item includes the body parameters of the target object at a fixed moment;

[0091] The first discriminant model is connected to the first generative model, and is configured to sequentially evaluate the plurality of first candidate items output by the first generative model using the first type of information in the natural motion pattern library as a real sample and the body feature information expressed by the first original painting or the second original painting as a condition, to obtain a first evaluation result; and

[0092] The first evaluation result is input into the first discrimination model again to determine whether the first candidate in the first evaluation result is a true sample, and the first candidate in the first evaluation result is output after it is determined to be a true sample.

[0093] Specifically, the first adversarial network model uses all the body parameters of the object at a fixed moment captured in the natural motion pattern library as real samples. At the same time, the body feature information expressed in the original painting is input into the discriminant model D as a condition. The discriminant model D sequentially evaluates the alternative results of multiple blocks generated by the generation model G and selects all the body parameters of the optimal complete block string; finally, the discriminant model D determines whether the complete block string is a real sample, and continuously conducts confrontation, adjustment and upgrading until the generated complete block string is difficult to distinguish between true and false and is output.

[0094] The second adversarial network model serves as the second layer (stage 2) of the stacked GAN. The second layer is global adversarial. As a preferred embodiment, the second adversarial network model includes a second generative model and a second discriminative model.

[0095] Then in step S3, the second generation model is used to generate a plurality of second alternative options according to the input first posture parameter, second posture parameter and object behavior, each second alternative option including all target body dynamic parameters at all given moments in the process of the target object completing the object behavior;

[0096] The second discriminant model is connected to the second generative model, and is configured to sequentially evaluate the plurality of second alternatives output by the second generative model using the second type of information in the natural motion pattern library as a true sample and the output of the first adversarial network model as a condition, to obtain a second evaluation result; and

[0097] The second evaluation result is input into the second discrimination model again to determine whether the second candidate in the second evaluation result is a true sample, and the second candidate in the second evaluation result is output after it is determined to be a true sample.

[0098] Specifically, the second adversarial network model selects the posture and dynamic parameters corresponding to the corresponding predetermined action according to the object's behavior as the input of the generative model G to generate multiple second alternatives. All posture parameters and dynamic parameters at all given moments in the process of executing the complete predetermined action in the natural motion pattern library are used as real samples; at the same time, all posture parameters of the complete body block string of the object in the output alternatives of the first adversarial network model are input as conditions into the discriminant model D. The discriminant model D sequentially evaluates the multiple second alternatives generated by the generative model G. Each alternative contains all posture parameters and dynamic parameters at all given moments in the intermediate transition from the first posture parameter to the second posture parameter during the execution of the given object behavior. The discriminant model D determines the optimal posture parameters and dynamic parameters at all given moments on the object's motion trajectory.

[0099] Finally, the optimal body posture and dynamic parameters are fed back into the discriminant model D to determine whether these parameters are true samples. Through continuous adversarial training, adjustments, and upgrades, the generated results become indistinguishable from true samples. Ultimately, all body dynamic parameters that conform to natural physical movement are obtained. A two-layer adversarial network model is stacked to enhance its processing capabilities.

[0100] The physical feature information expressed in the above original painting can be identified from the original painting, or can be obtained based on the character setting diagram or other specific instructions of the two-dimensional dynamic image, or can be obtained by manually inputting precise values ​​or introducing exaggerated variables.

[0101] Discriminative model D uses parameter information corresponding to the object category in the natural motion pattern library as real samples. The object category can be determined based on the instructions in the character design diagram. The object category and physical feature information are substituted into the model together, and the first and second category information corresponding to the object category in the natural motion pattern library are used as real samples.

[0102] Specifically, shape feature information, in addition to conveying information such as the length, width, and outline of the object's volume, may also include the object's position. In storyboard designs and scenes for 2D dynamic images, indicator lines and image information can be used to deduce the target object's 3D environment and determine shots that meet the original design requirements. Indicator lines can display the target object's position and orientation in space. By analyzing the length and / or angle of the indicator lines, the target object's positional relationship relative to the overall spatial environment and other objects can be inferred. Indicator lines can directly define a clear perspective space. Given the object's actual size, the angle of sight between the vanishing point and the lens azimuth can be used, utilizing the principle of "similar triangles," to determine the object's size and orientation in the shot. This directly derives the range of the object's 3D coordinates. Image information can also provide clues to construct a 3D environment. For example, perspective effects, shadows, and lighting can help determine an object's position and shape. By observing the perspective relationships and shadow effects within the image, the target object's spatial position can be inferred. According to the perspective phenomenon, the size of objects in the same direction becomes smaller as the distance increases. The position of the moving object in each original painting can be roughly estimated based on the size of the object drawn in the original painting.

[0103] As a preferred embodiment, after step S3 and before step S4, the method further includes:

[0104] Repeat steps S2-S3 multiple times, outputting an output result each time after executing steps S2-S3, and each output result includes all target body dynamic parameters at all given moments in the process of completing the object behavior;

[0105] The dynamic parameter range of the target body at each given moment is determined based on multiple output results, and a set of target body dynamic parameters is selected and output from the dynamic parameter range of the target body at all given moments as all target body dynamic parameters output by the second adversarial network model.

[0106] Specifically, embodiments of the present invention can directly substitute the parameters output from steps S2-S3 into the model. Alternatively, steps S2-S3 can be repeated multiple times. Repeating steps S2-S3 multiple times will yield multiple output results, from which the range of posture parameters with the highest probability of occurring at each given moment can be determined. This decision-making step can be implemented based on statistics or other existing decision-making methods.

[0107] Furthermore, a random forest decision can be used to determine the range of posture parameters with the highest probability of occurring at each given moment. A posture parameter is then selected from this range so that the dynamic parameters generated during the overall motion can still be identified as true samples by the discriminant model D. This can be viewed as stacking another GAN, where the true samples for the discriminant model D are the same as the samples for the second adversarial network model. In some embodiments, generation can also be performed in reverse chronological order. This reverse chronological generation refers to the process of gradually reversing the output of the generative model to generate true samples. For example, given original images 1, 3, 5, 7, and 9, intermediate image 2 is derived from original images 1 and 3, intermediate image 4 is derived from original images 3 and 5, and intermediate images 6 and 8 are derived in this order. The relationship between the original images and the intermediate images is arranged in order. Reverse chronological generation means that intermediate image 8 is derived from original images 9 and 7, and intermediate images 6, 4, and 2 are derived in this order.

[0108] As a preferred embodiment, step S4 further includes: adding texture and / or coloring to the target object model.

[0109] Specifically, the model directly generated using the parameters output by the adversarial network model can be called a "naked model" or "untextured model" of the object. Textures need to be added, and the surface material of the model, such as the object's exterior, including clothing, hairstyle, hair color, and accessories, needs to be selected for physical simulation.

[0110] In addition to textures, you can also add coloring effects to change the appearance of your model. This is done by selecting appropriate colors or material properties and applying them to your model.

[0111] The global information of the model is extracted according to the needs, and the production of two-dimensional dynamic images can be assisted based on the extracted global information, reducing the drawing of the front object framework, and the artist will subsequently improve the details.

[0112] Furthermore, the extracted global information can be used to perform image style transfer. Image style transfer refers to applying the style of one image to another, so that the target image has similar stylistic features to the source image. In the production of 2D dynamic images, this can be used to transfer the extracted global information to create a dynamic image with a similar style to the source image, achieving better results.

[0113] As a preferred embodiment, step S4 further includes:

[0114] A motion simulation is performed using the target object model, and screenshots are taken during the simulation process, and the screenshots are output as the first original picture, the second original picture, and the intermediate picture between the first original picture and the second original picture.

[0115] Specifically, in this embodiment, the global information of the model can be directly screenshoted to obtain the in-between paintings between the original paintings, thereby realizing the automatic generation of in-between paintings, greatly reducing the time for drawing the in-between paintings and the workload of repeated drawing, improving production efficiency, while maintaining the overall style and details of the original paintings, and ensuring the dynamic smoothness of the in-between paintings.

[0116] Furthermore, since the parameters in the natural motion model library are captured and align with human motion, screenshots can be used directly as original artwork. Since some original artwork is extremely rudimentary (such as stick figures), detailed skeletal and muscular models can add greater detail and realism to these crude artworks, optimizing them to a certain extent.

[0117] As a preferred embodiment, the method further includes after step S4:

[0118] The two intermediate pictures are respectively used as the first original picture and the second original picture, and the process returns to step S2.

[0119] Specifically, in this embodiment, the intermediate picture obtained according to the parameters generated by the adversarial network model can be used as the original picture, re-input into the adversarial network model for iteration, and a new intermediate picture can be continuously generated. The original picture between the two intermediate pictures can be optimized through the new intermediate picture.

[0120] For example, a concept artist creates original paintings 1, 3, and 5. Original paintings 1 and 3 are fed into the adversarial network model to generate parameters, which then generate intermediate painting 2. Original paintings 3 and 5 are fed into the adversarial network model to generate parameters, which then generate intermediate painting 4. Following the order of original painting 1, intermediate painting 2, original painting 3, intermediate painting 4, and original painting 5, a dynamic image can be created. In this embodiment of the present invention, intermediate paintings 2 and 4 can be re-input into the model to generate new intermediate painting 31. This intermediate painting 31 can be used to optimize original painting 3. By continuously optimizing the original paintings, the desired effect can be achieved.

[0121] In the embodiment of the present invention, after the original picture is input into the adversarial network model, the intermediate pictures generated using the model output parameters can be more than one, or multiple consecutive pictures.

[0122] The 2D dynamic image production method of the present invention can be applied to animation, film, and game production. By centrally capturing and generating motion, the required motion can be directly called, simplifying the production process and saving costs. Furthermore, a series of interpolation operations can be implemented, and motion exaggeration can be requested without the need for subsequent modification, further simplifying the process.

[0123] This method can also be applied to the creation of the metaverse. The metaverse is a virtual reality world that contains various virtual scenes and characters. The two-dimensional dynamic image production method of the present invention can easily generate character movements in the metaverse, making them more vivid and realistic.

[0124] This method can also be applied to the design of robotic motions. Based on the robot's starting and final states, existing data can be used to automatically generate the necessary processes. The generated results can then be continuously evaluated to select the optimal strategy. For example, if a robot is required to walk from point A to point B, it can automatically perform a series of leg-lifting and arm-swinging movements based on existing data.

[0125] In addition to the above application fields, the method of the present invention is applicable to other fields involving the production of two-dimensional dynamic images, which will not be listed here one by one.

[0126] The advantages or beneficial effects of adopting the above technical solution are: the present invention can generate two-dimensional dynamic images more efficiently by adopting artificial intelligence technology, and can also simulate more realistic and natural object shape changes and spatial movements, greatly improving the production capacity and finished product effects of two-dimensional dynamic images, and reducing the time consumption and manpower and capital investment limitations.

[0127] The above are only preferred embodiments of the present invention and do not limit the implementation mode and protection scope of the present invention. For those skilled in the art, it should be aware that all solutions obtained by equivalent substitutions and obvious changes made using the contents of this specification and illustrations should be included in the protection scope of the present invention.

Claims

1. A method for producing a two-dimensional dynamic image, characterized in that: Pre-establishing a natural motion pattern library, wherein the natural motion pattern library pre-stores parameter information of different categories of objects, the parameter information including first-category information representing all body posture parameters of each of the objects at a fixed moment and second-category information representing all body dynamic parameters of each of the objects at all given moments in the process of completing a predetermined movement; The two-dimensional dynamic image production method comprises: Step S1, obtaining a first original picture and a second original picture, wherein both the first original picture and the second original picture contain the same target object, and determining an object behavior of the target object between the first original picture and the second original picture; Step S2: inputting the first original image and the second original image into a pre-built first adversarial network model, respectively. The first adversarial network model is configured to output a first posture parameter of the target object in the first original image at a fixed time and a second posture parameter of the target object in the second original image at a fixed time based on the first type of information in the natural motion pattern library. Step S3: inputting the first body posture parameter, the second body posture parameter, and the object behavior into a pre-built second adversarial network model, wherein the second adversarial network model is configured to output all target body dynamic parameters of the target object at all given moments in the process of completing the object behavior based on the second type of information in the natural motion pattern library; Step S4: Substitute all the target body dynamic parameters output by the second adversarial network model into a preset modeling model to obtain a target object model.

2. The two-dimensional dynamic image production method according to claim 1, characterized in that: Also includes: Pre-establishing a body posture parameter library, wherein the body posture parameter library pre-stores all callable body posture parameters of the objects of different categories without overall spatial motion; Each of the objects includes a plurality of blocks; The steps of establishing the natural motion pattern library include: Step A1, capturing motion data of each subject in the body parameter library when performing a series of body parameters to complete a predetermined action; Step A2, using the correlation ratio of the movement distance of all the body blocks of the object in the motion data to the actual measurement information of the body block features as the predetermined action to generate a series of dynamic parameters; Step A3: define the predetermined action, and package the action definition, the body parameters and the dynamic parameters in chronological order to form the body dynamic parameters of each object and include them in the natural movement pattern library.

3. The two-dimensional dynamic image production method according to claim 2, characterized in that: The step A3 also includes: assigning exaggerated variables to the body parameters and / or the dynamic parameters corresponding to the action definition, and packaging the exaggerated variables with the action definition, the body parameters and the dynamic parameters in chronological order to form the body dynamic parameters of each object and including them in the natural motion pattern library.

4. The method for producing a two-dimensional dynamic image according to claim 2, wherein: Each of the blocks has a corresponding block code; The steps of establishing the body parameter library include: Step B1, for each object, determining the connection relationship between all the blocks in the object; Step B2, determining a movable spatial range of each of the blocks when fixed to one of the adjacent blocks, wherein the movable spatial range includes a plurality of block regions, each of the block regions having a corresponding region code; Step B3, determining when each of the blocks is located in each of the block areas relative to the pre-set The movement information of the standard posture is set, and the body block code, the region code, and the movement information are summarized and included in the body posture parameter library.

5. The two-dimensional dynamic image production method according to claim 1, characterized in that: After step S3 and before step S4, the following steps are further included: Repeating steps S2-S3 multiple times, outputting an output result each time after executing steps S2-S3, wherein each output result includes all target body dynamic parameters of the target object at all given moments in the process of completing the object behavior; The dynamic parameter range of the target body at each given moment is determined based on the multiple output results, and a set of target body dynamic parameters is selected and output from the dynamic parameter range of the target body at all given moments as all the target body dynamic parameters output by the second adversarial network model.

6. The method for producing a two-dimensional dynamic image according to claim 1, wherein: The step S4 further includes: adding texture and / or coloring to the target object model.

7. The two-dimensional dynamic image production method according to claim 1 or 6, characterized in that: The step S4 further includes: Motion simulation is performed using the target object model, and screenshots are taken during the simulation process, and the screenshots are output as the first original picture, the second original picture, and an intermediate picture between the first original picture and the second original picture.

8. The method for producing a two-dimensional dynamic image according to claim 7, wherein: After step S4, the following steps are also included: The two intermediate pictures are used as the first original picture and the second original picture respectively, and the process returns to step S2.

9. The two-dimensional dynamic image production method according to claim 1, characterized in that: The first adversarial network model includes a first generative model and a first discriminative model; Then in step S2, the first generation model is used to generate a plurality of first candidate items according to the input first original picture or the second original picture, each of the first candidate items includes the body parameters of the target object at a fixed moment; The first discriminant model is connected to the first generative model, and is configured to sequentially evaluate the plurality of first candidate items output by the first generative model using the first type of information in the natural motion pattern library as a real sample and the body feature information expressed by the first original painting or the second original painting as a condition, to obtain a first evaluation result; as well as The first evaluation result is input into the first discrimination model again to determine whether the first candidate in the first evaluation result is the real sample, and the first candidate in the first evaluation result is output after it is determined to be the real sample.

10. The two-dimensional dynamic image production method according to claim 1, characterized in that: The second adversarial network model includes a second generative model and a second discriminative model; Then, in step S3, the second generation model is used to generate a plurality of second alternative items according to the input first posture parameter, the second posture parameter, and the object behavior, each of the second alternative items including all target body dynamic parameters of the target object at all given moments in the process of completing the object behavior; The second discriminant model is connected to the second generative model, and is configured to sequentially evaluate the plurality of second alternatives output by the second generative model using the second type of information in the natural motion pattern library as a true sample and the output of the first adversarial network model as a condition to obtain a second evaluation result; as well as The second evaluation result is input into the second discrimination model again to determine whether the second candidate item in the second evaluation result is the true sample, and the second candidate item in the second evaluation result is output after it is determined to be the true sample.

Images