Emulation of Hand-Drawn Lines in CG Animation

Abstract

To provide a method, a system, and a device that emulate artistic hand-drawn lines in CG animation and generate a drawing.SOLUTION: The method includes: a first attaching step of attaching ink lines drawn directly onto characters of CG animation; a first enabling step of modifying the ink lines using nudges to generate a drawing; a first moving step of moving the drawing into a UV space and generating first pass training data; a second attaching step of attaching the ink lines onto the characters of the CG animation; a second enabling step of modifying the ink lines; a second moving step of moving the drawing into the UV space and generating second pass training data; a step of combining the first pass training data and the second pass training data to generate combined data; and a step of creating and outputting second generation machine learning using the combined data.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] [Cross - Reference to Related Applications]

[0001] This application claims the benefit of priority under 35 U.S.C. § 119(e) of U.S. Provisional Patent Application No. 62 / 775,843, filed on Dec. 5, 2018, entitled "Ink Lines and Machine Learning for Emulating Hand Drawn Lines", which is co - pending. The disclosure of the above - mentioned related application is incorporated herein by reference.

[0002]

[0002] This disclosure relates to computer - generated (CG) animation, and more specifically, to emulating artistic hand - drawn lines in CG animation.

Background Art

[0003]

[0003] In computer - graphics (CG) animation, the goal may include creating a rendered character that gives the feel and look of a hand - drawn character. Techniques consisting of "line work / ink lines" based on procedural "rules" (e.g., using a Toon Shader) have been determined to be ineffective in achieving satisfactory results. The main problem was that the artist did not draw based on a limited set of rules. A more robust solution to the problem of creating artistically - looking hand - drawn line work was needed. Further, the created line work needed to be adjusted (if necessary) and incorporated into existing CG animations.

Summary of the Invention

Problems to be Solved by the Invention

[0004]

[0004] This disclosure provides for emulating artistic hand - drawn lines in CG animation.

Means for Solving the Problems

[0005]

[0005] In one implementation form, a method for emulating artistic hand-drawn lines in CG animation is disclosed. The method includes a first pasting step of pasting ink lines directly drawn on a character of the CG animation, a first realizing step of enabling an artist to use a nudge to modify the ink lines to generate a drawing, a first moving step of moving the drawing into a UV space to generate first pass-training data, a second pasting step of pasting the ink lines on the character of the CG animation, a second realizing step of enabling the artist to modify the ink lines, a second moving step of moving the drawing into the UV space to generate second pass-training data, a step of combining the first pass-training data and the second pass-training data to generate combined data, and a step of creating and outputting second-generation machine learning using the combined data.

[0006]

[0006] In one implementation form, the character exists within a frame of the CG animation. In one implementation form, the ink lines directly drawn on the character provide appropriate ink lines for a specific pose of the character. In one implementation form, the nudge is created using a nudging brush. In one implementation form, the method further includes a step of keying and interpolating the nudge by directly projecting the ink lines onto the character. In one implementation form, the method further includes a step of generating key frames and interpolating between the key frames. In one implementation form, the character is formed in 3-D geometry. In one implementation form, the first moving step of moving the drawing into the UV space enables the artist to work on the 3-D geometry of the character in a 2-D screen space. In one implementation form, the method further includes a step of inverse-transforming the work performed in the 2-D screen space into the 3-D geometry.

[0007]

[0007] In another implementation, in CG animation, a system for generating a drawing by emulating lines hand-drawn by an artist is disclosed. The system includes a prediction module configured to receive shot data including a camera angle and an expression and a second generative machine learning model to generate a prediction of the drawing, an adhesion module configured to receive the prediction and attach ink lines onto the 3-D geometry of an animated character, and a nudging module configured to enable the artist to generate a nudge using a nudging brush to modify the ink lines and move the drawing in the UV space.

[0008]

[0008] In one implementation, the training data for the prediction module uses animated curves created for pairs of angles having pairs of corresponding expression changes.

[0009]

[0009] In another implementation, an apparatus for generating a drawing by emulating an artistic hand-drawn line in a CG animation is disclosed. The apparatus includes a first realization means for enabling an artist to directly draw an ink line on an animated character in the drawing, an attachment means for directly attaching the drawn ink line onto the animated character, a second realization means for enabling the artist to modify the ink line using a nudge, and a movement means for moving the drawing into the UV space so that the artist can work within a 2-D screen space, wherein the movement means generates first path training data when the path is a first path, while the movement means generates second path training data when the path is not the first path, a first creation means for creating a first machine learning model and sending it to the attachment means for the second path, a combination means for combining the first path training data and the second path training data to generate combined data, and a second creation means for creating a second machine learning model using the combined data.

[0010]

[0010] In one implementation, the first realization means includes a tablet or pad-type device and an electronic pen. In one implementation, the tablet or pad-type device includes at least one application capable of displaying the CG animation in frame units. In one implementation, the attachment means includes at least one application capable of processing an input from the artist to the attachment means and integrating the input onto the CG animation. In one implementation, the second realization means enables the artist to create a nudge using a nudging brush. In one implementation, the movement means creates keyframes using the nudge and interpolates between the keyframes. In one implementation, the movement means interpolates between the keyframes by automatically compositing all frames between the keyframes.

[0011]

[0011] Other features and advantages should become apparent from this specification, which shows an example of an aspect of the present disclosure.

[0012]

[0012] By considering the accompanying drawings in which like parts are designated by like reference numerals, the details of the present disclosure can be partially gathered with respect to both its structure and operation.

Brief Description of the Drawings

[0013]

Figure 1

Figure 2

Figure 3

Figure 4A

Figure 4B

Figure 5

Figure 6

Figure 7

Figure 8

Modes for Carrying Out the Invention

[0014]

[0019] As described above, a more robust solution to the problem of creating artistically-appearing hand-drawn line work was needed. Further, the created line work needed to be adhered to and (if necessary) adjusted and incorporated into existing CG animations. Some implementations of the present disclosure provide for emulating artistic hand-drawn lines in CG animation. After reading these descriptions, the implementation methods of the present disclosure in various implementations and applications will become apparent. Although various implementations of the present disclosure are described herein, it should be understood that these implementations are presented by way of example only and not by way of limitation. Accordingly, the detailed descriptions of the various implementations should not be construed as limiting the scope or extent of the present disclosure.

[0015]

[0020] In one implementation, a enhanced ink line application having two modes was implemented to generate a set of realistic hand-drawn ink lines. The first mode was the "adhesion mode" and the second mode was the "adjustment mode". The adhesion mode of the enhanced ink line application was aimed at adhering or pasting ink lines (e.g., curved geometries) onto the 3-D geometry of a CG animation. Thus, in the adhesion mode, the application enables the artist to adjust the drawing with easy control to maintain a proper 2-D conceptual design. For example, in one implementation, the application enables the artist to draw on top of the frame of a CG animation using an electronic pen. Thereby, the drawing of the frame can appear like dynamic drawing rather than a standard texture map. The adjustment mode of the enhanced ink line application is aimed at capturing and cataloging the pasting performed on the ink lines by the artist and improving the speed of the workflow using machine learning.

[0016]

[0021] In one implementation, the adhesion mode of the enhanced ink line application enables the artist to draw directly on the character (and / or object) in the drawing, providing lines appropriate for a particular pose of the character. Once the ink lines are drawn, the application enables the artist to quickly and easily adjust the drawing using a nudging "brush". In one implementation, the adhesion mode is used for the most complex shots when the adjustment of machine learning in the adjustment mode is slower than obtaining a shot from scratch.

[0017]

[0022] In one implementation, the nudges created by the nudging brush are automatically keyframed and interpolated by the application. In one implementation, keyframing and interpolation include projecting the ink lines directly onto the geometry and interpolating the nudges within the screen space or any UV space. The enhanced ink line application enables the artist to work within the 2-D screen space, providing the most intuitive space for the artist to draw and nudge to produce the desired results. The inverse transformation to the 3-D space is automatically performed without the artist's intervention.

[0018]

[0023] In one implementation, keyframing and interpolation use nudging to create keyframes and interpolate between the keyframes. For example, the artist nudges the ink lines of the drawing in frame 1 and then performs nudging on the ink lines of the drawing in frame 10 to create two keyframes 1 and 10. Next, the adhesion mode of the enhanced ink line application interpolates between the keyframes by automatically compositing all the frames between the two keyframes.

[0019]

[0024] In one implementation, the curve position in UV space controls where the character will ultimately be located in 3-D space. When the drawing is nudged, these edits can be shifted to the UV space. Therefore, there is a complex relationship between the position of the curve in UV space relative to the viewing angle and the character's representation. Using sufficient examples, machine learning can find patterns between the shape of the curve in UV space and the character's viewing angle and representation. Therefore, machine learning maps this relationship.

[0020]

[0025] In one implementation, the adjustment mode of the enhanced ink line application includes: (1) a first pass prediction that uses machine learning to create ink lines on the character to generate an initial prediction result; (2) the artist uses a "nudging brush" to make corrections; and (3) after the artist uses the nudge brush to correct the initial prediction, shifting the adjustment to the UV space. Next, the data is exported and combined with the initial data. Then, the combined data is used to "retrain" the machine learning model. Therefore, "retraining" improves the accuracy of the model.

[0021]

[0026] In one implementation, the training data for the first pass prediction uses a "turn table" animation created from different angles with expression changes. Next, the enhanced ink line application's adhesion mode is used to create animated curves (i.e., drawings) for these angles and expressions. Then, the animated curves are shifted into the UV space to enable machine learning training. This becomes the "target" for the first set of training data generation (about 600 frames per character). The "features" are the relative camera angle and expression changes. Therefore, once the training data is generated, the machine learning algorithm is trained to find the relationship between the drawing position and the "features".

[0022]

[0027] In one implementation, after training a machine learning algorithm to find relationships, predictions are made for another set of animation data. Next, an artist uses a nudge brush to adjust (or "correct") these initial predictions. Next, these corrected adjustments are moved into UV space and exported as additional training data. Next, the machine learning model is retrained using this combined training data of these corrections and the initial training data.

[0023]

[0028] FIG. 1 is a flowchart of a process 100 for emulating artistic hand-drawn lines in a CG animation according to one implementation of the present disclosure. In one implementation, process 100 enables an artist to adjust the drawing with easy control to maintain a proper 2-D conceptual design. For example, in one implementation, process 100 enables an artist to draw on top of a frame of a CG animation using an electronic pen. Thereby, the drawing of the frame can appear like dynamic drawing rather than a standard texture map.

[0024]

[0029] In the implementation shown in FIG. 1, in block 110, in process 100, the artist can directly draw ink lines on the character in the drawing so as to provide appropriate lines for a specific pose of the character. In one implementation, the character exists within the frame of a CG animation. In one implementation, the ink lines directly drawn on the character provide appropriate ink lines for a specific pose of the character. Next, in block 112, the ink lines are pasted onto the animated character (i.e., on the 3-D geometry). Once the ink lines are drawn and pasted, in block 114, process 100 enables the artist to adjust the drawing using a nudging “brush”. Next, in block 116, the drawing is moved to the UV space to generate training data. Process 100 enables the artist to work within the 2-D screen space and provides the most intuitive space, enabling the artist to draw and nudge to generate the desired result. The inverse transformation to the 3-D space is automatically performed without the intervention of the artist.

[0025]

[0030] In one implementation, nudging creates keyframes and interpolates between the keyframes. For example, the artist nudges the ink lines of the drawing in frame 1 and then performs nudging on the ink lines of the drawing in frame 10 to create two keyframes 1 and 10. Next, process 100 interpolates between the keyframes by automatically synthesizing all the frames between the two keyframes.

[0026]

[0031] In one implementation, when the drawing is moved to the UV space to generate training data, at block 118, process 100 creates a first generative machine learning model, and at block 120, calls the first generative machine learning model. At block 122, again, ink lines are pasted onto the animated character, and at block 124, an artist can use a "nudging brush" to correct the prediction of the ink lines. Next, at block 126, the drawing is moved to the UV space to generate training data again. At block 128, the first pass training data (generated at block 116) and the second pass training data (generated at block 126) are combined. Thus, when the training data is combined, at block 130, process 100 creates a second generative machine learning model.

[0027]

[0032] In one implementation, training data for the first pass prediction uses a "turn table" animation created from different angles with expression changes. Next, process 100 creates animated curves for these angles and expressions. Next, the animated curves are converted to the UV space to enable machine learning training. This becomes the "target" for the first set of training data generation (about 600 frames per character). The "features" are the relative camera angles and expression changes. Thus, when the training data is generated, a machine learning algorithm is trained to find the relationship between the drawing position and the "features".

[0028]

[0033] Figure 2 is a block diagram of a system 200 for emulating artistic hand-drawn lines in a CG animation according to one implementation of the present disclosure. In the implementation shown in Figure 2, system 200 includes a prediction module 230, an adhesion module 240, and a nudging module 250.

[0029]

[0034] In one implementation, the prediction module 230 receives the shot data 210 and the second generative machine learning "model" 220 created by block 130 of FIG. 1 and creates a prediction. The shot data 210 includes relative camera angles and relative representations.

[0030]

[0035] In one implementation, the adhesion module 240 is configured to implement an adhesion mode that receives a prediction from the prediction module 230 and is aimed at adhering or pasting ink lines (e.g., curve geometry) onto the animated character. Thus, the adhesion module 240 enables the artist to easily control and adjust the drawing to maintain a proper 2-D conceptual design. For example, in one implementation, the adhesion module 240 enables the artist to draw on top of the frames of a CG animation using an electronic pen.

[0031]

[0037] In one implementation, when the ink lines are drawn and pasted, the nudging module 250 enables the artist to use a nudging "brush" to adjust the drawing and move it in the UV space. Thus, the nudging module 250 enables the artist to work within the 2-D screen space and provides the most intuitive space, enabling the artist to draw and nudge to generate the desired results. The nudging module 250 performs the inverse transformation to the 3-D space without the artist's intervention. For lighting purposes, the drawing 260 output by the nudging module 250 is exported.

[0032]

[0038] Figure 3 is a block diagram of an apparatus 300 for emulating artistic hand-drawn lines in CG animation according to an implementation of the present disclosure. In the implementation shown in Figure 3, the apparatus 300 includes means 310 for enabling an artist to draw ink lines, means 320 for pasting the ink lines, means 330 for enabling the artist to modify the ink lines, means 340 for moving the drawing into the UV space, means 360 for creating a first generative machine learning model, means 380 for combining training data, and means 390 for creating a second generative machine learning model.

[0033]

[0039] The means 310 for enabling an artist to draw ink lines enables the artist to directly draw ink lines on a character in a drawing to provide appropriate lines for a specific pose of the character. In one implementation, the means 310 includes a tablet or pad-type device and an electronic pen. In one implementation, the tablet device includes at least one application capable of displaying CG animation on a frame-by-frame basis.

[0034]

[0040] The means 320 for pasting the ink lines pastes the ink lines directly drawn on the character of the CG animation. In one implementation, the means 320 includes at least one application capable of processing an input from the artist to the means 310 and integrating the input onto the CG animation.

[0035]

[0041] Means 330 for enabling an artist to modify ink lines allows the artist to adjust the drawing using a nudging "brush". In one implementation, means 330 includes a tablet or pad-type device and an electronic pen used as the nudging brush. In one implementation, the tablet device includes at least one application capable of displaying CG animation on a frame-by-frame basis. Means 330 for enabling an artist to modify ink lines also enables the artist to modify the prediction of the ink lines.

[0036]

[0042] Means 340 for moving the drawing into UV space moves the drawing into UV space to provide the most intuitive space for the artist to draw and nudge to generate the desired result, enabling the artist to work within the 2-D screen space. In one implementation, means 340 for moving the drawing into UV space uses nudging to create keyframes and interpolate between the keyframes. For example, the artist can nudge the ink lines of the drawing in frame 1 and then perform nudging on the ink lines of the drawing in frame 10 to create two keyframes 1 and 10. Next, means 340 interpolates between the keyframes by automatically synthesizing all the frames between the two keyframes. Means 340 for moving the drawing into UV space generates first path training data 350 when the path is the first path, while means 340 for moving the drawing into UV space generates second path training data 370 when the path is not the first path (i.e., the second path).

[0037]

[0043] Means 360 for creating a first generative machine learning model creates a first generative machine learning model and sends it to means 320 for pasting ink lines for the second path.

[0038]

[0044] In the second path, means 380 for combining training data combines first-path training data 350 and second-path training data 370. Means 390 for creating a second generative machine learning model creates the second generative machine learning model using the combined training data.

[0039]

[0045] FIG. 4A is a diagram of a computer system 400 and a user 402 according to one implementation of the present disclosure. User 402 implements an emulation application 490 for emulating artistic hand-drawn lines in a CG animation as illustrated and described with respect to process 100 for emulating artistic hand-drawn lines in the CG animation shown in FIG. 1 and emulator 200 of the block diagram shown in FIG. 2 using computer system 400.

[0040]

[0046] Computer system 400 stores and executes emulation application 490 of FIG. 4B. Further, computer system 400 can communicate with software program 404. Software program 404 can include software code for a screen tone look application. Software program 404 can be loaded onto an external medium such as a CD, DVD, or storage drive as further described below.

[0041]

[0047] Furthermore, computer system 400 can be connected to network 480. Network 480 can be connected in a variety of different architectures, such as a client-server architecture, a peer-to-peer network architecture, or other types of architectures. For example, network 480 can communicate with server 485 that coordinates the engines and data used within emulation application 490. Also, the network can be of different types. For example, network 480 can be the Internet, a local area network or any variation of a local area network, a wide area network, a metropolitan area network, an intranet or an extranet, or a wireless network.

[0042]

[0048] FIG. 4B is a functional block diagram showing computer system 400 that hosts emulation application 490 according to one implementation of the present disclosure. Controller 410 is a programmable processor that controls the operation of computer system 400 and its components. Controller 410 loads instructions (e.g., in the form of a computer program) from memory 420 or a built-in controller memory (not shown) and executes these instructions to control the system. In its execution, controller 410 provides a software system to emulation application 490, for example, enabling the creation and configuration of engines and data extraction units within emulation application 490. Alternatively, this service can be implemented as a separate hardware component in controller 410 or computer system 400.

[0043]

[0049] Memory 420 temporarily stores data for use by other components of computer system 400. In one implementation, memory 420 is implemented as RAM. In one implementation, memory 420 also includes long-term or permanent memory such as flash memory and / or ROM.

[0044]

[0050] Storage 430 stores data temporarily or over a long period of time for use by other components of computer system 400. For example, storage 430 stores data used by emulation application 490. In one implementation, storage 430 is a hard disk drive.

[0045]

[0051] Media device 440 accepts removable media and reads and / or writes data to the inserted media. In one implementation, for example, media device 440 is an optical disk drive.

[0046]

[0052] User interface 450 includes components for receiving user input from a user of computer system 400 and presenting information to user 402. In one implementation, user interface 450 includes a keyboard, a mouse, an audio speaker, and a display. Controller 410 uses input from user 402 to adjust the operation of computer system 400.

[0047]

[0053] I / O interface 460 includes one or more I / O ports and connects to corresponding I / O devices such as external storage or auxiliary devices (e.g., a printer or a PDA). In one implementation, the ports of I / O interface 460 include ports such as USB ports, PCMCIA ports, serial ports, and / or parallel ports. In another implementation, I / O interface 460 includes a wireless interface for communicating wirelessly with external devices.

[0048]

[0054] Network interface 470 includes wired and / or wireless network connections such as an RJ-45 or “Wi-Fi” interface (including but not limited to 802.11) that supports an Ethernet connection.

[0049]

[0055] Computer system 400 includes additional hardware and software typical of a computer system (e.g., power, cooling, operating system), but these components are not specifically shown in FIG. 4B for the sake of simplicity. In other implementations, different configurations of the computer system can be used (e.g., different bus or storage configurations or multiprocessor configurations).

[0050]

[0056] FIGS. 5, 6, 7, and 8 show examples of intermediate results of an emulation system, process, and apparatus.

[0051]

[0057] In one implementation, emulation system 200 is a system entirely composed of hardware including one or more digital signal processors (DSPs), general-purpose microprocessors, application-specific integrated circuits (ASICs), field programmable gate / logic arrays (FPGAs), or other equivalent integrated or discrete logic circuits. In another implementation, emulation system 200 is composed of a combination of hardware and software.

[0052]

[0058] The description of the present disclosure for the disclosed implementations has been made so that those skilled in the art can implement or utilize the present disclosure. It will be readily apparent to those skilled in the art that numerous modifications to these implementations are possible, and the principles defined herein can also be applied to other implementations without departing from the spirit or scope of the present disclosure. Accordingly, the present disclosure is not intended to be limited to the implementations shown herein, but the broadest scope consistent with the principles and novel features disclosed herein should be given.

[0053]

[0059] Various implementations of the present disclosure are realized in the form of electronic hardware, computer software, or a combination of these technologies. Some implementations include one or more computer programs executed by one or more computer devices. Generally, a computer device includes one or more processors, one or more data storage components (e.g., volatile or non-volatile memory modules and persistent optical and magnetic storage devices such as hard disk drives, floppy disk drives, CD-ROM drives, and magnetic tape drives), one or more input devices (e.g., game controllers, mice, and keyboards), and one or more output devices (e.g., display devices).

[0054]

[0060] A computer program typically includes executable code stored in a persistent storage medium and copied into memory at runtime. At least one processor executes the code by fetching program instructions from memory in a predetermined order. During the execution of the program code, the computer receives data from input and / or storage devices, performs processing on the data, and supplies the resulting data to output and / or storage devices.

[0055]

[0061] One of ordinary skill in the art will appreciate that the various illustrative modules and method steps described herein can be implemented as electronic hardware, software, firmware, or combinations thereof. To clearly demonstrate this interchangeability of hardware and software, various illustrative modules and method steps are generally described herein in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. One of ordinary skill in the art may implement the described functionality in varying ways for each particular application, but such implementations should not be construed as causing a departure from the scope of the present disclosure. Also, the grouping of functions within a module or step is for ease of explanation. Without departing from the present disclosure, specific functions may be transferred from one module or step to another.

[0056]

[0062] Not all features of the above-described embodiments are necessarily required in a particular implementation of the present disclosure. Further, it should be understood that the descriptions and drawings presented herein represent the broad subject matter contemplated by the present disclosure. Further, the scope of the present disclosure fully encompasses other implementations that may become apparent to those of ordinary skill in the art, and thus, it should be understood that the scope of the present disclosure is not limited by anything other than the appended claims.

Description of the Reference Numerals

[0057] 100 A process for emulating artistic hand-drawn lines in CG animation 110 Enable an artist to directly draw ink lines on a character during drawing 112 Paste ink lines on the 3-D geometry of an animated character 114 Enable an artist to modify ink lines using a nudge brush 116 Move the drawing into UV space to generate training data 118 Create a first generative machine learning model 120 Invoke the first generative machine learning model to generate predictions 122 Attach ink lines onto the 3-D geometry of the animated character 124 Enable the artist to modify the ink lines using a nudging brush 126 Move the drawing into the UV space to generate training data 128 Combine the data 130 Create the second generative machine learning model 200 System / emulator for emulating artistic hand-drawn lines in CG animation 210 Shot data 220 Second generative machine learning model 230 Prediction module 240 Adhesion module 250 Nudging module 260 Drawing 300 Device for emulating artistic hand-drawn lines in CG animation 310 Means for enabling the artist to draw ink lines 320 Means for attaching ink lines 330 Means for enabling the artist to modify ink lines 340 Means for moving the drawing into the UV space 350 First post-training data 360 Means for creating the first generative machine learning model 370 Second post-training data 380 Means for combining the training data 390 Means for creating the second generative machine learning model 400 Computer system 402 User 404 Software program 410 Controller 420 Memory 430 Storage 440 Media Device 450 User Interface 460 I / O Interface 470 Network Interface 480 Network 485 Server 490 Emulation Application

Claims

1. A method for emulating artistic hand-drawn lines and generating drawings in CG animation, the method comprising: a step of an artist directly drawing ink lines on a character of a CG animation in a drawing; a first pasting step of pasting the ink lines drawn by the artist onto a character of a CG animation formed in 3-D geometry; a first realization step of enabling the artist to modify the ink lines using a nudge to generate a drawing and to adjust the drawing, wherein the modification of the ink lines is made by the artist drawing on top of a frame of the CG animation using an electronic pen, and the nudge (1) generates key frames and (2) interpolates between the key frames by automatically synthesizing all frames between the key frames; a first movement step of moving the drawing into UV space to enable the artist to work on the 3-D geometry of the character in 2-D screen space, where the curve position in UV space controls where the character will ultimately be located in 3-D space by associating the curve position in UV space with the camera angle and the expression; a step of inverse-transforming a character in 2-D screen space into 3-D geometry to generate first pass-training data; a step of creating a first machine learning model using the first pass-training data; a second pasting step of pasting the ink lines predicted by the first machine learning model onto a character of a CG animation formed in 3-D geometry; a second realization step of enabling the artist to modify the ink lines to adjust the drawing; a second movement step of moving the drawing into the UV space to enable the artist to work on the 3-D geometry of the character in 2-D screen space; a step of inverse-transforming a character in 2-D screen space into 3-D geometry to generate second pass-training data; a step of combining the first pass-training data and the second pass-training data to generate combined data; a step of creating and outputting a second machine learning model using the combined data; A method characterized by including the above steps.

2. The method according to claim 1, wherein the character exists within a frame of the CG animation.

3. The method according to claim 1, wherein the nudge is created using a nudging brush.

4. Furthermore, a step of key-operating and interpolating the nudge by directly projecting the ink line onto the character. The method according to claim 1, characterized by including this.

5. The method according to claim 1, further comprising a step of generating key frames and interpolating between the key frames.

6. A system for generating a drawing by emulating a line hand-drawn by an artist in a CG animation, the system comprising: a prediction unit that predicts an ink line from shot data including a camera angle and an expression and a machine learning model; an adhesion unit that attaches the predicted ink line onto a character of a CG animation formed in 3-D geometry; a nudge unit that enables an artist to correct the ink line using a nudging brush to generate a nudge and move and adjust the drawing within the UV space; a first pasting unit that directly pastes an ink line drawn by an artist onto a character of a CG animation to provide an appropriate line for a specific pose of the character; a first realization unit that enables an artist to correct the ink line using a nudge to generate a drawing and adjust the drawing; comprising the correction of the ink line is made by an artist drawing on top of a frame of a CG animation using an electronic pen, and the nudge is (1) generating key frames and (2) interpolating between the key frames by automatically synthesizing all frames between the key frames. Furthermore, a first moving unit that moves the drawing within the UV space so that the artist can work on the 3-D geometry of the character in a 2-D screen space, where the curve position in the UV space is associated with the camera angle and the expression, thereby controlling where the character will ultimately be located in 3-D space, a first conversion unit that inverse-converts a character in a 2-D screen space into 3-D geometry and generates first past-training data. A second pasting unit that pastes the ink lines predicted by the first machine learning model created using the first past training data onto the character of the CG animation formed in the 3-D geometry; A second realization unit that enables an artist to correct the ink lines pasted by the second pasting unit in order to adjust the drawing; A second moving unit that moves the drawing within the UV space so that an artist can work on the 3-D geometry of the character in the 2-D screen space; A second conversion unit that inverse-converts the character in the 2-D screen space into 3-D geometry and generates second past training data; A combining unit that combines the first past training data and the second past training data to generate combined data; A creating unit that creates and outputs a second machine learning model using the combined data; A system characterized by including the above. **Claim 7** The system according to claim 6, characterized in that the training data for the prediction unit uses animated curves created for pairs of angles having pairs of corresponding expression changes. **Claim 8** An apparatus for generating a drawing by emulating artistic hand-drawn lines in a CG animation, the apparatus comprising: Means for predicting ink lines from shot data including camera angles and expressions and a machine learning model; First realization means for enabling an artist to directly draw ink lines on the animated character in the drawing; Pasting means for pasting, in a first pass, the ink lines drawn by the artist onto the character of the CG animation formed in the 3-D geometry, and for pasting, in a second pass, the predicted ink lines onto the character of the CG animation formed in the 3-D geometry; Second realization means for enabling an artist to correct the ink lines using a nudge to generate a drawing and to adjust the drawing, wherein the correction of the ink lines is made by the artist drawing on top of the frame of the CG animation using an electronic pen, and the nudge (1) generates key frames and (2) interpolates between the key frames by automatically synthesizing all frames between the key frames. Further, move the drawing into UV space so that the artist can work within the 2-D screen space, generate first pass training data when the path is the first path, and create second pass training data when the path is not the first path. Here, by associating the curve position in the UV space with the camera angle and expression, it controls where the character will ultimately be located in the 3-D space based on the curve position in the UV space. Means for inverse-transforming the animated character in the 2-D screen space into 3-D geometry. First generation generating means for generating a first generation machine learning model using the first pass training data and sending it to means for predicting ink lines. Combining means for combining the first pass training data and the second pass training data to generate combined data. Second generation generating means for generating a second generation machine learning model using the combined data. An apparatus characterized by including the above.

9. The apparatus according to claim 8, wherein the first realizing means includes a tablet or pad-type device and an electronic pen.

10. The apparatus according to claim 9, wherein the tablet or pad-type device includes at least one application capable of displaying the CG animation in frame units.

11. The apparatus according to claim 8, wherein the pasting means includes at least one application capable of processing an input by the artist to the pasting means and integrating the input onto the CG animation.

12. The apparatus according to claim 8, wherein the second realizing means enables the artist to create a nudge using a nudging brush.

13. The apparatus according to claim 8, wherein the moving means uses the nudge to create keyframes and interpolate between the keyframes.

14. The apparatus according to claim 13, wherein the moving means interpolates between the keyframes by automatically synthesizing all frames between the keyframes.

Images