Persona renders Cubism 3, 4, and 5 models, up to and including moc3 format v6 (supporting Cubism 5.3 offscreen rendering and blend modes), powered by an in-house rendering engine.
Live2D models exist as flat canvases in screen space: drag to pan, scroll to zoom, and hold Shift while scrolling to rotate. Multiple models can share a single stage, stacked alongside images and videos in the layer list, and one model can also be attached to another as an item.
For the features that differ between Live2D and VRM, see Live2D vs. VRM.
Rendering Engine
Persona features a dedicated Cubism renderer built specifically for the app. It loads .model3.json models along with their motions, expressions, physics, poses, and hit areas, providing full support for Cubism clipping masks and blend modes. For Cubism 5.3 blend models, the renderer renders into private targets with high-precision masks, preserving Atop clipping and Out cutouts in paint order while honoring runtime multiply and screen color overrides. Like the rest of the 3D stage, it runs natively on WebGPU and automatically falls back to WebGL on systems without WebGPU support.
Because Live2D models composite directly into the 3D stage, the complete effects chain applies to them, including tone mapping, color grading, bloom, chromatic aberration, and stylize effects. Live2D-only scenes can also include effect layers and retain full access to the Look adjustments section.
Bundled Models
Two sample Live2D models are available on demand in the Downloads group of the Inventory: LAPLACE's own Persona-chan (nasutouri and LAPLACE Live!) and Sony's official sample character, RAYNOS-chan.
Persona-chan and RAYNOS-chan are demonstration assets, free for evaluation and testing but not permitted for any commercial use, including commercial streaming and video production. Persona flags both models in the info panel with a link to their licensing terms. Please import and use your own model for broadcasts.
Motions and Expressions
The Motions and Expressions panels display all items provided by the current model, with item counts noted beside their respective headings.
Motions
The Motions panel displays the model's motion groups, lets you record new motions, and provides a dropdown menu on each motion button to edit, duplicate, rename, or delete motions and set shortcuts; see Live2D Motions and Recording Motions.
Expressions
Expressions operate on a toggle basis: click a button to activate and highlight it, and click it again to turn it off.
Multiple expressions can run simultaneously and blend additively. For example, if a model separates blushing and pout shapes into distinct expressions, both can be enabled together. Expressions apply in the order they are clicked, and removing one does not affect the others.
The top of the panel includes New Expression to create a new expression file alongside the model. The arrow dropdown beside each expression button allows you to edit, duplicate, set a shortcut, rename, or delete it; see Expression Editor.
Remembering Expressions
Remember Expressions, located inside the Expression Settings menu at the top of the Expressions section, is enabled by default. When enabled, your active expression combination is automatically saved and restored when switching back to the model or relaunching the application. Restored expressions appear immediately in their settled states rather than fading in, preventing visible expression shifts during startup.
Saved expressions are tied to the model file rather than global Persona settings. If the model directory contains a .vtube.json file, the state is written directly to it (using the same field as VTube Studio for bidirectional compatibility); otherwise, it is saved in a <model>.persona.json sidecar file beside the .model3.json. Disabling this toggle clears any saved expressions.
Expression Groups
Applies toLive2DVRMExpression Groups make a set of expressions mutually exclusive (for example, showing only one of several hairstyles or hand poses at a time), following the same rules as VTube Studio's Expression Groups. Click New Group at the top of the section to add a group; the switch beside its name turns the group on or off, and its menu offers Add Expressions…, When Switching, Rename…, and Delete Group…. Deleting a group leaves its expressions as they are.
In the Expression Settings menu at the top of the section, Sort By switches between two views:
- Group (default): Groups appear directly inside the expression list, with expressions that do not belong to any group listed under Ungrouped at the bottom.
- Name: Groups appear in a header bar at the top, and all expressions are listed below in their natural order with colored dots indicating which groups each expression belongs to.
You can manage group members in three ways: drag an expression button onto a group (or onto Ungrouped to remove it from the group it was dragged from), choose Add Expressions… in a group's menu to check expressions from a filterable list, or open the dropdown menu beside any expression button to use Add to Group and Remove from Group. For each group an expression belongs to, its dropdown menu also offers:
| Option | Effect |
|---|---|
| Inverted | Counts as shown while off, like a "Hat Off" expression |
| Fade Time | How long the member takes to switch (0 to 2 seconds, 0.2 by default); in an enabled group it overrides the fade a shortcut or caller asks for |
In a group's menu, When Switching controls how members transition when one replaces another:
| Option | Effect |
|---|---|
| Crossfade | Default; the old expression fades out while the new one fades in, matching VTube Studio |
| Hide, Then Show | When a member becomes shown, its fade-in waits until every other member hidden by that group has completely faded out (including inverted members) |
Expression groups follow these rules:
- In an enabled group, when one member becomes visible, any other visible members in the group hide over their own Fade Time (normal members turn off; inverted members turn on). Hiding a member never activates another.
- Clicking Clear All Expressions, enabling a group, or adding, removing, or inverting a member resolves each group's state: the first visible member in group order stays active, and the rest hide. Renaming a group, changing a fade time, changing When Switching, or deleting a group does not change active expressions.
- Remembered expressions are restored through the same rules, so the last remembered member of a group wins.
- The rules apply regardless of how an expression is switched: panel buttons, shortcuts, automations, the plugin API, Stream Deck, and MCP.
Until <model>.persona.json has its own group list, Persona imports the ExpressionGroups field from the model's .vtube.json read-only. Once you create or edit groups in Persona, the full list is saved to <model>.persona.json, which takes precedence over .vtube.json from then on without modifying .vtube.json. Members match by expression file name (by expression name on VRM), case-insensitively; a member the model lacks is kept and skipped. Renaming or deleting an expression file in the expression editor updates matching entries in both .vtube.json and <model>.persona.json.
VTube Studio Models
Many models created for VTube Studio leave their .motion3.json and .exp3.json files loose in the folder without declaring them in the model settings file, relying on VTube Studio's .vtube.json hotkey configuration instead. Standard Cubism loaders cannot discover these loose files and display empty lists.
Persona automatically scans and discovers these loose files and displays them normally, extracting display names from the model's .vtube.json when present. The scan never modifies files on disk; additional entries are generated in memory when the model loads.
Even if a model explicitly declares its own motion groups or expressions, loose files are seamlessly appended: each loose motion becomes an independent group named after its file, and loose expressions follow those declared in the model settings. As a result, files generated by recording motions or through the expression editor are always recognized on any model.
Idle and Breathing
Configure natural motion loops when no input is active, fallback transitions when face tracking is lost, and breathing cycle speed and amplitude; see Live2D Idle.
Whole-Model Movement
Configure how face or body tracking translates and rotates the entire model within the stage, as well as body sway driven by head rotation; see Avatar Tracking Settings.
Parameter Bindings
Expand the Parameters section to configure tracking inputs, weights, input and output ranges, and custom response curves for each parameter. See Parameter Bindings.
Versions
For a model you imported, the Versions section saves, compares, and restores earlier versions of its files; see Version History. Bundled models keep no versions.
Model Info
Expand the Info section to review technical specifications and loaded assets for the current model:
| Row | Meaning |
|---|---|
| Source | Whether the model is a bundled sample or imported from disk |
| File | The loaded .model3.json path; click the folder icon to reveal it in your system file manager |
| Config | Configuration sidecars detected alongside the model, such as .vtube.json and .persona.json |
| Canvas | The model's original canvas resolution |
| moc3 | The model's moc3 format version, along with the maximum version supported by the bundled Cubism Core |
| Core | The Cubism Core version included in this build |
| Params | Total number of Live2D parameters available for tracking and animation |
| Parts | Total number of model parts |
| Textures | Texture atlas dimensions and estimated uncompressed VRAM usage |
| Motions | Total number of motion groups and motion files |
| Express. | Total number of recognized expression files |
| Blink | Parameter IDs driven by automatic eye blinking |
| Lip Sync | Parameter IDs in the model's LipSync group; flagged when empty, in which case microphone lip sync relies on mouth parameter bindings |
| Perfect Sync | Number of raw ARKit channels bound in parameter bindings (out of 52); reflects the model's configuration, not active tracking inputs |
| Extras | Loaded optional subsystems, such as physics, pose, and blink |
| Author | Creator credit from .persona.json meta; for custom models, edit via Edit Info… in the Inventory row menu |
| Version · Copyright · Contact · Links | Shown when set in .persona.json meta |
| License | License URL, with attribution requirements flagged |
| Load Time | Total wall-clock time spent loading and initializing the model on stage |
If the moc3 row is flagged as too new, the model was created in a newer Cubism version than this build can parse. The model may appear in the list but will fail to render on stage.
Last updated on November 7, 2026