Persona

VRM is an open 3D avatar format rendered within Persona's 3D stage, sharing an underlying WebGPU pipeline with Live2D.

The 3D stage executes on WebGPU, with automatic graceful fallback to WebGL on unsupported hardware. Rendering takes place in a dedicated background worker thread to minimize the impact of complex scenes and high polygon counts on UI responsiveness.

Persona provides native support for both VRM 0.x and VRM 1.0 specifications, respecting the distinct look-at and blendshape conventions of each version. Files with .vrm or .glb extensions load through the standard Loading Custom Models workflow.

For the features that differ between Live2D and VRM, see Live2D vs. VRM.

Bundled AvatarsCopy Anchor Link

Three sample VRM avatars are available on demand in the Downloads section of the Inventory:

Milky Green (LAPLACE Live! and Milky Green aka. 明前奶绿) and Wenjing VRM (LAPLACE Live! × 彩茫) require explicit attribution, which Persona includes alongside other third-party notices in the About Dialog. Specific license terms appear in the License row of the model's info panel, with attribution requirements highlighted.

Camera ControlsCopy Anchor Link

Unlike Live2D models, which operate on a 2D planar canvas, VRM models reside within a 3D environment managed by a full orbit camera:

Camera movement incorporates smooth damping and inertia matching the Live2D canvas. Elevation is clamped before reaching the polar axes to prevent disorienting camera flips. Reset Camera returns to a default framing derived from the avatar's bounding box, and Field of View adjusts the lens between 10° and 90°. Camera coordinates can also be adjusted numerically in the Stage details panel (see Camera).

Camera properties belong to the Scene rather than individual models, meaning all VRM avatars within a scene share the same camera vantage point.

PlacementCopy Anchor Link

Avatar transforms are defined in world-space meters independently of the camera, accompanied by on-stage transform gizmos for translation, rotation, and scale. In the Position section, numeric coordinate fields are paired with a 3D planar pad and a rotation trackball (see Visual Controls).

Scale ranges from 0.05× to 10×. Spring bone physics scale proportionally with the model geometry, ensuring secondary hair and cloth dynamics remain natural regardless of avatar size.

Placement can also be controlled remotely via game controllers; see Stage Movement and Walking While Moving.

Animation, Expressions, and MaterialsCopy Anchor Link

A VRM model's motion dynamics and material appearance are managed through four dedicated sub-panels:

• VRM Idle: Natural breathing, subtle idle sway, automatic blinking, and smooth fallback transitions when face tracking drops.
• VRM Motions: Import and play .vrma and Mixamo animations with custom blend transitions, loop modes, and bone masks.
• VRM Expressions and Outfits: Morph target (blendshape) activation, additive blending, expression groups, and outfit/mesh component visibility toggles.
• MToon Materials: Real-time shader parameter overrides for cel-shading, shadow colors, outlines, and lighting responses.

Pose TrackingCopy Anchor Link

Because VRMs feature a standardized humanoid skeleton, they can receive full-body motion capture data over VMC or mocopi protocols from senders such as VSeeFace, Virtual Motion Capture, or mocopi sensors. Live2D models lack 3D skeletal rigs, so this section appears only for VRM models.

A standard webcam can also drive body posture without specialized hardware: Body Tracking drives torso and leg orientation, while Hand Tracking animates arms and fingers.

VRM ShortcutsCopy Anchor Link

Avatar Shortcuts apply seamlessly to VRM models: expressions bind by shape name, and animation clips bind by file path. Swap Model actions operate cross-format, enabling a single hotkey to swap a layer between Live2D and VRM avatars.

How playback options apply to VRM:

• Hold to Activate, Fade Time, and Turn Off After apply fully to VRM expressions.
• Stop on Last Frame and Stop After apply only to Live2D motions. VRM animation clips follow their authored timelines.

Automations can target specific VRM layers with actions such as Toggle Expression, Play Motion, Clear All Expressions, and Swap Model, as well as Model Position to move avatars to saved coordinates.

Head orientation, body posture, gaze, and expressions can also be bound directly to game controllers (see VRM Targets).

VersionsCopy Anchor Link

For a VRM you imported, the Versions section saves, compares, and restores earlier versions; each version holds the .vrm file, its .persona.json sidecar, and its thumbnail. See Version History. Bundled and catalog-downloaded avatars keep no versions.

Model InfoCopy Anchor Link

Expand the Info section to review technical specifications and asset status for the active VRM avatar:

Scene TransitionsCopy Anchor Link

When switching layers or scenes, VRM avatars transition using smooth fade blending unless the scene's Show and Hide policy is set to Pop or Cut. Fade composites the avatar as a single layer at its alpha while preserving depth sorting, preventing interior geometry from showing through during transitions; resting placement opacity below 1 continues to use screen-space dither dissolve to prevent self-transparency.

Avatars remain hidden during background initialization until fully posed and oriented toward the camera, ensuring incomplete T-poses never appear on stage.