Live2D Cubism Editor 5.1 alpha1 User Manual |
03/28/2024 r1
Setting information for alpha version of the Editor
About the English and Chinese editors
External Application Integration
[AI function] Auto Generation of Deformer
[AI function] Auto Generation of Sway Motion
Version | Revision date | Revision details |
r1 | 03/28/2024 | First edition |
This alpha version is intended to evaluate the usability and performance of the new features and improvements planned for Cubism Editor 5.1. Therefore, you will be able to try out new features before they are available. However, some functions may not yet be implemented, or the operation may be unstable and behave unexpectedly.
The UI is based on specifications currently under development and may change in the future. In addition, there is a possibility that data such as model files (.cmo3) and animation files (.can3) that you have created may no longer be opened (incompatible) due to bugs or specification changes.
While it is possible to create works using the alpha version, we strongly recommend that you use it to “evaluate” the usability and performance of the new features.
Please ensure you fully understand the purpose of the alpha version before using it.
Please send your requests for new features and bug reports below!
We also accept requests and bug reports on X (formerly Twitter)!
Post with the hashtag #cubism51_alpha!
We also accept requests and bug reports on X (formerly Twitter)!
Post with the hashtag #cubism51_alpha!
• bilibili
Provide your feedback in the comments section!
If you are already using the full version of Cubism on your PC, Editor settings such as keyboard shortcuts will be inherited when you start Cubism 5.1 alpha. At this time, the settings for the alpha version will be saved in a different location than the full version and will not be shared thereafter. This is to preserve the operating environment of the full version.
Please note that due to this fact, setting items changed in Cubism 5.1 alpha will not be carried over to Cubism 5.1 beta and the official version.
Note: Setting items changed in 3.x and 4.x will be carried over to the 5.1 beta and the full version.
The settings files for Cubism 5.1 alpha are saved in the following folders. Note that these settings files are not used in beta and later versions, so there is no problem with deleting them.
Windows | C:\Users\[USER_NAME]\AppData\Roaming\Live2D\Cubism5.1alpha_Editor C:\Users\[USER_NAME]\AppData\Roaming\Live2D\Cubism5.1alpha_Viewer |
Mac | /Users/[USER_NAME]/Library/Live2D/Cubism5.1alpha_Editor /Users/[USER_NAME]/Library/Live2D/Cubism5.1alpha_Viewer |
Note: Please be careful when handling these folders as they are hidden folders.
Note: If Cubism 5.1 alpha does not work properly, it may help to delete the above folders. Once the folders are deleted, if the settings files for the full version Editor (3.x /4.x) are already on your PC, they will be transferred over again.
The display language can be switched in [Environment settings] starting from Cubism 5.1 beta.
In Cubism 5.1 alpha, you cannot switch from [Environment settings], so please download and use the editor for each language.
Please note that some wording newly added in Cubism 5.1 alpha may be of lower translation quality. We will provide high-quality translations from Cubism 5.1 beta.
Enhanced model template functionality.
The previous model template function could be used only once at the start of model creation, but the function has been greatly expanded so that it can be applied to models during creation.
This document describes enhancements from the previous model template functionality.
For new ways to use the enhanced model templates, see the “User Tutorial.”
In the alpha version, in the Model template selection dialog box described below, the model template is empty by default.
To try the model template function, the following files must be unzipped and placed in the model template folder as described below.
ZIP: https://cubism.live2d.com/editor-alpha/doc/template-model/Template_Model_en.zip
See the [User Tutorial.] for details.
Re-application is now possible
Previously, application of templates was limited to models with only an ArtMesh on the canvas, such as PSD imports.
With this enhancement, templates can now be applied to models that have been edited, such as after deformers or parameters have been added.
Partial application is now possible
Templates can now be applied only to selected ArtMeshes.
Specifically, when executing “Apply Template,” only the selected ArtMesh is targeted.
ArtMeshes that are not targeted will not appear in the Template dialog box and the template will not be applied to those ArtMeshes.
If you apply the template without selecting anything, the entire model will be applied.
Adding elements to apply templates
Other improvements
In the modeling process, select [File] menu -> [Apply Template...] to display the [Model template] dialog box.
In this dialog box, you can select the template to be applied.
There are two ways to select a template: [Select from model template folder] or [Select from .cmo3 file].
[Select from model template folder] displays a list of model templates. You can select a template by clicking on the thumbnail.
In [Select from .cmo3 file], you can directly specify a model file as a template using the [...] button on the right side.
Model Template Folder
Select [Edit] menu -> [Open model template folder] to open the model template folder.
Model templates placed in this folder will appear in the [Select from model template folder] dialog box.
Search Filter
The [Search filter] allows you to narrow down the templates that appear in the [Select from model template folder].
Specifically, only templates that include the string you entered in either [Name] or [Details] will be displayed.
After clicking OK in [Preview], the [Apply Settings] dialog box appears.
Applying a template adds new parameters, deformers, parts, and other objects to the model to which it is applied, and changes the selected ArtMesh.
In this dialog box, you can adjust the properties of the object being added and the ArtMesh being modified when the template is applied.
Same Parameters
[Same parameters] should be used when the template model and the model to which it will be applied contain the same parameters.
When [Add] is selected, if the template and the destination have parameters with the same ID, the parameter on the template side will be added as a parameter with a different ID.
Selecting [Overwrite] will not add the parameter as a separate new parameter, even if the template and the destination have parameters with the same ID.
An exception is if one of the same parameters is a blend shape and the other is a regular parameter, the process is the same as if you had chosen [Add].
Label Color
[Label color] is the label color setting for the object to which the template will be applied.
The label color is set for each of the following four items.
You can choose one label color setting for parameter groups, parts, and deformers from [Use template color] and [Use specified color].
Use template color
The label color of the template is used as is.
Change to specified color
Changes the label color to the specified color.
You can select one of three settings for the label color of the ArtMesh: [Do not change], [Change to template color], or [Change to specified color].
Do not change
The label color is not changed from the destination label color.
Change to template color
Change the color to the same as the label color of the template.
Change to specified color
Changes the label color to the specified color.
In modeling mode, execute [File] menu -> [Export Template...] to display the [Export Template] dialog box.
This dialog box allows you to export the model you are editing as a model template.
[Name] is the name that appears in the Model Template Selection dialog box.
Note: Do not enter characters that cannot be used for file names, as they will also be used as folder names and file names.
[Details] is text that appears as a tooltip for the thumbnail of the Model Template Selection dialog box. It is also a target of the [Search filter] in the Model Template Selection dialog box.
Model template information (.template.json)
Model template information is represented in JSON format.
Precautions Model template information created in the alpha editor cannot be used in the editors other than the alpha version. |
Model template information for alpha1 consists of the following seven properties.
Number | Property | Description |
1 | FormatVersion | The version number related to the format of the model template information. Only referenced internally in the system and not displayed. |
2 | ModelVersion | The version number related to the model of the model template. Referenced internally in the system and not displayed. |
3 | Name | The name of the model template. It appears in the [Model template selection] dialog box. |
4 | BaseModel | Model files (.cmo3) of the model template. This can be loaded from the [Model template selection] dialog box and displayed as a template model. |
5 | Thumbnail | Image files (.png/.gif) identifying the model template. They are displayed as thumbnails in the [Model template selection] dialog box. When executing “Export Template,” a still image of the entire model to be registered as [BaseModel] is registered. If you want a preview of how the model template will work, for example, replace it with a GIF. |
6 | Description | Explanation of the model template. It appears as a tooltip in the [Model template selection] dialog box. |
7 | ReferenceFiles | Reference files for the model template. You can register model template description files (.txt), source image files (.psd), etc. The files appear items on the [Reference] button menu in the [Model template selection] dialog box, and when selected, a duplicate of that file can be created. |
The external application integration function allows for communication with external applications to interact with the editor. You can set and retrieve model parameters.
For information on implementing external applications, see the “Developer’s Manual.”
External application samples are published on GitHub.
https://github.com/Live2D-Garage/CubismExternalAppPluginSamples
You can configure settings for linking with external applications from the menu [File] -> [External Application Integration Settings].
Number | Name | Details |
1 | Port number | Specifies the port number to connect to. The default value is “22033.” If another application is using the same port number, it cannot be used, so please set it so that there are no duplicates. |
2 | Start/Stop connection button | Start: Enables connection to external apps. Stop: Disconnects and disables all connected external apps. |
3 | State of connectivity | Allows you to check the status of connections with all external applications. |
4 | Allow remote connections | When this checkbox is checked, remote connections with external applications are allowed. |
5 | External applications | Displays information on currently connected external applications. Check the [Allow] checkbox to enable the integration. |
6 | Disconnect | You can disconnect a connected application from the right-click menu. |
Set the port number.
Next, click the Start connection button.
When it is shown as enabled, connections with external applications will be possible.
Once you have connected the external application you wish to use, check the [Allow] checkbox.
Permission is required to use the external application integration function.
If an unexpected external application appears, do not allow it.
If you want to make a connection from a different machine, check the [Allow remote connections] checkbox.
Checking it may result in malicious attacks from third parties via the internet.
Please be sure to check your router and local network settings and use it with caution.
Click the Stop connection button.
All external applications that are connected will be disconnected.
Apply 3D Expression is a function that assists in the generation of 3D-like motion shapes.
It targets drawable objects (ArtMesh, ArtPath) and deformers (warp deformer, rotational deformer) and estimates their shapes in 3D space to generate three-dimensional expressions.
Apply 3D Expression is available in the Modeling View.
Select [Modeling] menu -> [Apply 3D Expression] -> [3D Rotation Expression] to open the [Apply 3D Expression] dialog box.
This dialog box is not modal, so other modeling operations can be performed while it is open.
Number | Item | Description |
1 | List | Objects registered under [Register objects] will appear in this list. Right-clicking in the list will bring up a submenu. |
2 | Register Objects | Creates a setting for Apply 3D Expression to the object selected in the Modeling View. |
3 | Plane | Selecting this tab displays the “3D rotation in a plane” setting items. |
4 | Depth Estimation | Selecting this tab displays the “3D rotation with depth estimation” setting items. |
5 | Ties to parameters | Selects the parameters that tie the rotation in the X and Y directions. You can also leave it blank to have nothing tied to it. For example, if you want to create a rotation expression only in the X direction, leave the Y direction column blank. |
6 | Rotation settings | Sets the rotation settings. |
7 | Automatically update keyforms when settings are updated | If checked, the keyform is automatically created whenever the contents of [Rotation settings] are changed. |
8 | Apply plane form to keyform | Creates a keyform based on the [Rotation settings]. This button creates a 3D rotated keyform with the frontal shape as a plane, without Depth Estimation. |
Number | Item | Description |
9 | Depth Estimation | Check a box to set the depth estimation. None of them are symmetrical in shape itself. Bilaterally symmetrical (warp deformers only): The Z position of each point by depth estimation is symmetrical on the left and right. Vertically symmetrical (warp deformers only): The Z position of each point by depth estimation is symmetrical from top to bottom. |
10 | Apply the estimated shape from the current keyform | Creates a keyform based on the [Rotation settings]. This button first performs depth estimation on the displayed keyform and then adds depth information to the shape. Then creates a keyform with a 3D rotation of the frontal shape. |
Number | Item | Description |
1 | Camera position | Specifies the position of the camera that captures the object in 3D space. Specifies XY coordinates in canvas coordinates (Marker will appear in the Modeling View depending on the settings). Z is automatically changed according to the strength of the parse. |
2 | Center of rotation | Specifies the center position of the object when it is rotated in 3D space. Specifies XY coordinates in canvas coordinates (Marker will appear in the Modeling View depending on the settings). Z is the same as the canvas. |
3 | Offset in Z direction | Specifies how much the depth is shifted in the Z direction. Specifies the scale to be the same as the canvas coordinates, with the front of the screen as the positive side. The procedure for shifting in the Z direction is performed before rotation. If depth estimation is used, it is performed on the estimated position. |
4 | Strength of parse | Specifies the perspective after the image is taken. If set to 0, parallel projection is used. If set to a value greater than 0, perspective projection is used. |
5 | Angle range | Specifies the limit angle of rotation. The default value of the parameter is the front, and the minimum and maximum values, whichever is farther from the default value, is the limit of rotation possible. |
6 | Rotation method | Specifies the rotation calculation method. This setting mainly affects the appearance of the four corners. |
Below is a submenu that appears when right-clicking on the list.
Item | Description |
Synchronize selections with views | If checked, the same object will be selected in the View and Settings dialog box. |
Display | Changes the display type of the list. By checking the checkbox, you can set whether the object is displayed by Name or ID. |
Copy settings | Copies the settings in [Ties to parameters] and [Rotation settings] of the selected object. |
Paste settings | Copied settings with [Copy settings] will be applied to the selected object. |
Delete settings | Deletes settings related to the Apply 3D Expression of the selected object. Objects whose settings have been deleted disappear from the list. |
Create an expression of 3D rotation by completing following three phases.
Each phase is intended to be done in sequence, but may be skipped or redone one step earlier if necessary.
At this stage, the objects that will be the subject of subsequent operations are registered.
By registering the objects, a configuration is created, and “Apply 3D Expression” can be used.
The operating procedure for this phase is as follows.
At this stage, the keyform is created as a foundation for creating expressions with depth.
Assign a 3D rotation expression to the parameter and set it up so that it can be oriented to add depth.
The operating procedure for this phase is as follows.
If the parameter has not yet been keyed:
A message box appears asking if you want to add the three key points.
Clicking the OK button will key in the default, minimum, and maximum values for the parameter.
If the parameter has already been keyed:
The keyform of that key’s position is overwritten by the keyform of the 3D rotation expression.
If there is no key in the default value, the creation of the keyform will result in an error and the process will be aborted because the shape that will be the front cannot be determined.
At this stage, depth is added to the keyform created in “3D rotation in a plane” to create a three-dimensional rotation.
The operating procedure for this phase is as follows.
Since the default value is considered the front, depth estimation is not possible with the default value keyform.
OK: Non-default value | Not OK: Default value (Angle X 0.0, Angle Y 0.0) |
The next step is to select the orientation for adding depth. It is best to select areas where it is easy to get a sense of three-dimensionality.
A keyform of 3D rotation expression is then created, similar to “3D rotation in a plane.”
To achieve a better look, it is effective to arrange the keyforms in multiple orientations.
This section provides supplemental explanation of each setting item and how changing it will affect the results.
The [Camera position] setting determines how the image will look when rotated in 3D space. It is affected by the strength of the parse described elsewhere. The greater the strength of the parse, the greater the change in appearance due to changes in camera position.
The figure below shows the change in the rightward keyform when the camera position is changed (the X symbol is the camera position). Moving the camera position to the left gives the appearance of going around to the side of the subject.
Change in keyform facing right when camera position (X) is changed |
The [Center of rotation] setting is used to set the center of rotation of the object in 3D space. Specify X and Y in canvas coordinates. Z is on the canvas (fixed at 0).
The figure below shows the movement when the center of rotation is placed on the right and left sides of the canvas. The further away a point is from the center of rotation, the greater the
Center of rotation on right side of canvas | Center of rotation on left side of canvas |
The [Offset in Z direction] setting specifies how much the depth of the target should be shifted in the Z direction. The shifting procedure is performed before rotation.
If depth estimation is used, it is performed on the estimated depth. By changing this setting, the rotation can be expressed as floating forward or retracting backward from the center.
The figure below shows the motion when the center of rotation is set to the center of the object and the Offset in Z direction is positive and negative.
A positive Offset in the Z direction gives the object the appearance of floating in the foreground. Conversely, a negative Offset in the Z direction gives the object the appearance of being pulled back.
Center of rotation set to center of object and Offset in Z direction is positive | Center of rotation set to center of object and Offset in Z direction is negative |
The [Strength of parse] setting specifies the perspective after the image is taken. Increasing the parse value increases the distortion after rotation and gives the appearance of a closer camera. A parse value of 0 gives the appearance of a parallel projection.
The figure below shows the change in the rightward keyform as the strength of the parse is increased from 0. Increasing the strength of the parse will emphasize the difference in appearance between the foreground and background.
Change in rightward keyforms as strength of parse is increased from 0 |
The figure below shows the difference in appearance of the rotation with different strengths of parse. When Strength of parse is 0, the rotational movement is modest, but when Strength of parse is 100, the movement is more dynamic.
Strength of parse is 0 | Strength of parse is 100 |
The [Angle range] setting specifies the limit range of rotation.
Larger values give the appearance of greater rotation. The value specified here is the angle between the minimum and maximum values, whichever is farther from the default value, with the default value of the parameter as the front.
The figure below shows the change in the rightward keyform as the angle range is changed. Larger values give the appearance of greater rotation.
Change in rightward keyform when changing the Angle range |
The [Rotation method] setting specifies the rotation calculation method. This setting mainly affects the appearance of the four corners.
The figure below shows the difference in appearance of rotation for each value of [Rotate in X-Y order], [Rotate in Y-X order], and [Rotate on any axis].
Rotate in X-Y order | Rotate in Y-X order | Rotate on any axis |
The “Bilaterally symmetrical” and “Vertically symmetrical” settings for Depth Estimation can be used to automatically make the estimated depth of warp deformers symmetrical. Simply shape one side with keyform editing to create a bilaterally symmetrical or vertically symmetrical 3D curved surface.
The figure below shows the result of Depth Estimation with [Bilaterally symmetrical] enabled. If you make a keyform edit on the right (left) side of the warp deformer, the other side will have the same depth.
Result of Depth Estimation with [Bilaterally symmetrical] enabled |
Auto Generation of Deformer is a function that automatically generates deformer structures for the entire body of a humanoid model.
It creates a deformer structure of a typical human figure and automatically estimates where it should be placed in the deformer structure based on the placement of the ArtMesh.
It is possible to quickly create the initial deformer structure from an ArtMesh only, such as immediately after PSD import.
This function is intended for the following models.
The following models can be used, but the accuracy of auto generation will be reduced.
This function is not intended for use with the following models.
The generated deformer has the following structure.
This structure is fixed regardless of the model on which the function is executed.
However, deformers with no child objects are not generated.
When the function is executed, the ArtMesh in the model is automatically assigned to the children of the deformer in this structure.
The position and size of the generated deformer are determined by the position and size of the child ArtMesh.
This function is used in the Modeling View. The operating procedure is as follows.
When the function is executed from the menu, the following dialog box appears.
This dialog box allows you to review and partially modify the deformer structure to be created.
Number | Name | Details |
1 | Deformer structure to be generated | The deformer structure that will be created is displayed here. When the dialog box is launched, the ArtMesh in the model is automatically assigned to a child of a deformer here, and you can see which deformer the ArtMesh will be a child of when the function is executed. You can change the parent deformer by dragging and dropping the ArtMesh. Deformers cannot be moved. Selected ArtMeshes and deformers are also selected in the View, and their position and size can be confirmed. |
2 | ArtMesh not used | A list of ArtMeshes not included in the deformer structure can be found here. ArtMeshes shown here will not change their parent deformer after deformer creation. |
3 | Remove ArtMesh | This button moves an ArtMesh selected in the Deformer structure to be generated (1) to the ArtMesh not used (2) list. Deformers cannot be moved. |
4 | Add to deformer structure | This button moves an ArtMesh selected in the ArtMesh not used (2) to the Deformer structure to be generated (1). It is moved to the location selected in Deformer structure to be generated (1). |
5 | OK | Performs deformer generation as per the edited result. A deformer is created and placed directly under the root of the [Deformer] palette with the same structure as shown in the dialog box, and the ArtMesh is placed in the same manner as shown in the dialog box. The existing deformer structure remains intact. |
6 | Cancel | Cancels the editing result and closes the dialog box. No deformer will be added, and the existing ArtMesh and deformer structure will remain as it was before the dialog box was launched. |
This section describes the linkage between a similar function, Auto generation of facial deformer, and its companion function, Auto generation of facial motion.
This function creates a deformer for the entire body, which includes the facial deformer structure that can be generated by “Auto generation of facial deformer.”
If “Auto generation of facial deformer” is performed after deformers are generated by this function, or if this function is performed after “Auto generation of facial deformer” is performed, unintended generation results may be obtained.
However, it is possible to use this function and Auto generation of facial motion in combination.
When executing this function on a model that has already undergone “Auto generation of facial deformer,” delete the facial deformers generated by “Auto generation of facial deformer” before executing the function.
This function automatically generates the movement of swaying objects such as small objects and hair.
The degree of swaying can be adjusted from the dialog box while viewing the generated motion, streamlining the process of adding motion to swaying objects.
The following two types of motion attachments are supported.
Generate swaying:
This generates movement such as hair swaying from side to side in the wind.
Generate droop by angle Z:
This generates a downward motion such that when the head is tilted, the hair hangs downward due to gravity.
Creating complex sway movements that do not fall into these categories may not result in the intended movement.
This function also applies deformation to the warp deformer containing the ArtMesh of the area to be subjected to motion.
Motion attachment of sway using the rotation deformer is not covered by this function.
Number | Name | Details |
1 | Target deformer | Displays the name of the warp deformer that will be the target of sway deformation. |
2 | Target parameter | Displays the name of the parameter that will be the target of sway deformation. |
3 | Adjustment of sway deformation | Adjusts the setting values that define the degree of sway deformation. Descriptions of each setting value are given below. |
4 | Synchronize left and right settings | When checked, the setting values of the keyform (left side) and keyform (right side) of Adjustment of sway deformation are synchronized. It is checked by default. Uncheck this checkbox if you want to create asymmetrical sway deformations. |
5 | Apply | Clicking the [Apply] button calculates the shape of the sway deformation from the current settings and actually reflects it in the keyform. |
6 | Automatically reflect when adjustment values are changed | If checked, the sway deformation is automatically reflected each time the setting values are changed. It is unchecked by default. |
Note: The above screenshot shows the dialog box that appears when [Generate swaying] is executed. When [Generate droop by angle Z] is executed, the title of the dialog box and the initial value of “Adjustment of sway deformation” will change, but the rest of the functions are the same.
Number | Name | Details |
1 | Sway strength (horizontal) | Sets the strength of the horizontal oscillation. |
2 | Sway strength (vertical) | Sets the strength of the vertical oscillation. If the value is set lower than 0, movement can be applied as if it is being pulled downward. |
3 | Attenuation level | Adjusts swaying of the upper and lower parts of the deformer. If the value is increased, the upper part will almost stop moving and only the lower part will sway. |
4 | Zoom In/Zoom Out | Deformations can be made to increase or decrease the width of the deformer. Larger than 0 means expansion, and smaller than 0 means contraction. |
Note: For sway strength (horizontal) and sway strength (vertical), reasonable values are automatically estimated based on the target deformer information (size, etc.) and used as initial values.
This function is used in the Modeling View. The operating procedure is as follows.
From [Modeling] menu -> [Parameter] -> [Auto generation of sway motion], select one of the following two functions depending on the type of motion you wish to apply.
Change the setting values as needed and press [Apply] to reflect the sway deformation in the keyform.
Alternatively, check the [Automatically reflect when adjustment values are changed] checkbox to automatically reflect the sway deformation when the settings are changed.
This is a non-modal dialog box, so you can adjust the sway deformation while manipulating parameters and views in the main window to check the motion.
When the sway deformation is reflected in step 2, a process is executed to check if there are three keys for the target parameter.
Selecting [OK] deletes the existing keys and creates three new keys.
If [Cancel] is selected, the existing keys will remain in place and the sway deformation will not be reflected.
This alpha version includes some unimplemented features, known defects, and other items of concern that should be specifically noted. Normally, these limitations will be dealt with when upgrading from the alpha version, but this is not guaranteed.
Model template information (.template.json) created with the alpha editor cannot be used with the beta editor.
In the alpha version, the model template will initially be empty, but in the beta version, multiple model templates will be available by default.
If you delete all the keyforms of the edited object in Apply 3D Expression, the 3D rotation expression will be lost and the object will be fixed in the shape it was at the time it was displayed.
From this point on, if the parameters are rekeyed and the keyform is created again with depth estimation, the results will be inconsistent with the previous depth estimation results and correct estimation will not be possible.
If you delete the all keyforms, you will need to start over from 3D rotation in a plane (preferably after restoring the original shape, if possible).
In alpha1, there is a problem in the way the strength of the parse is calculated. The larger the target size, the more extreme the change.
This issue will be fixed in alpha2 or later.
When an ArtMesh is outside of the canvas and Auto Generation of Deformer is executed, the Auto Generation of Deformer dialog box does not appear.
Make sure all ArtMeshes are within the canvas before performing this function.
This issue will be fixed in alpha2 or later.
The setting values for the sway deformation adjustment are not saved. Closing and redisplaying the dialog box returns the settings to their default values.
Target warp deformer and parameters cannot be changed while the dialog box is displayed. To make changes, close the dialog box once.
The following functions use AI technology.
Auto Generation of Deformer:
AI technology is used to estimate which part of the model is which from the image information of each object.
Auto Generation of Sway Motion:
AI technology is used to estimate a reasonable amount of deformation based on information such as the position and size of the target deformer.
Among the works submitted to nizima, those that have given us permission for use in AI research are used as learning data. For more information, see the Live2D AI Research Policy and Request for Cooperation.