Open Stage → Camera to frame the 3D scene. These settings move the camera without changing avatar placement.
Camera
The 3D scene is viewed through an orbit camera. You can orbit by dragging, pan by right-dragging or holding Alt while dragging, and zoom by scrolling. For a complete list of navigation gestures, see VRM Avatar Camera Gestures.
| Control | Range |
|---|---|
| Position | X · Y · Z in meters, with an interactive 3D transform control |
| Orbit Angle | −180° – 180° |
| Tilt | −80° – 80° |
| Distance | 0.35 – 20 (widens dynamically based on environment model bounds) |
| Look-at Point | Target look-at and rotation pivot X · Y · Z coordinates |
| Field of View | 10° – 90° (default 30°) |
| Reset Camera | Restores default framing computed from the avatar bounding box |
| Camera Motion | Loads a .vmd camera move that drives the viewport in step with avatar motions |
Position moves the camera itself, with the same 3D position control used for avatars and 3D objects. The look-at point moves by the same offset, so the viewing angle and distance stay the same. To adjust the composition separately, use Orbit Angle, Tilt, Distance, and Look-at Point.
In scenes where the camera hasn't been customized yet, a Set Position button appears. While a camera motion track is assigned, these controls are locked; clear the track under Camera Motion below to edit them again.
A narrow field of view (FOV) compresses perspective like a telephoto portrait lens, while a wide FOV accentuates spatial depth like a wide-angle lens. Adjusting the field of view automatically compensates Distance so that the subject maintains its apparent size on stage. This compensation applies only when no camera motion is playing.
Follow
Makes the camera follow and frame a selected layer (a VRM avatar or 3D object) or the Primary Model. Selecting None keeps the camera stationary.
| Control | Description |
|---|---|
| Layer | Selects a specific VRM model, 3D object, or the Primary Model (automatically transfers when the primary avatar changes) |
| Anchor | Selects the tracking point on the model: Root or a Humanoid bone (such as Head, Chest, or Spine); 3D objects always track from their own origin |
| Rotation | Dictates how the target's rotation influences camera orientation: Ignore (follows position only, ignoring rotation), Yaw (follows horizontal turning), Yaw and Pitch, Yaw, Pitch, and Roll (fully tracks all three axes of rotation), or Lazy Follow (trails behind, swinging around only when the target moves across the field of view) |
| Damping | Sets the smoothing time in seconds to settle behind target movement (0 to 10 seconds, 0 locks immediately without lag): includes translation X · Y · Z and rotational damping for yaw, pitch, and roll |
When Follow is active, camera orbit parameters frame relative to the target anchor. Avatar locomotion, procedural motions, and tracking data move the camera view seamlessly without requiring manual camera coordinate updates.
Aim
Rotates the camera to keep a layer or the primary model at a designated position on screen without moving the camera's spatial coordinates (inspired by Warudo's Composer). Selecting None faces the standard look-at point.
| Control | Description |
|---|---|
| Layer | Selects the target layer or the Primary Model; can be configured independently of Follow |
| Anchor | Selects the aim point on the model (a humanoid bone or the root) |
| Screen Position | Normalized coordinates where the anchor sits on screen: Screen X and Screen Y (0 to 1, 0.5 centers) |
| Dead Zone | Dead Zone Width and Dead Zone Height (fraction of screen): the camera remains motionless while the target stays within this zone |
| Soft Zone | Soft Zone Width and Soft Zone Height (never smaller than the dead zone): the camera turns smoothly to re-center the target when it enters the soft zone, and turns immediately if it crosses the soft zone boundary |
| Bias | Bias X and Bias Y (−0.5 to 0.5): offsets the soft zone relative to the target anchor |
| Damping | Horizontal Damping and Vertical Damping (0 to 10 seconds, 0 turns immediately): seconds required to rotate the target back into the dead zone |
| Lookahead | Time (0 to 1 second, 0 disables) and Smoothing (0 to 30): aims ahead of a moving target based on velocity; includes an Ignore Vertical Motion switch |
Focus
Keeps the focal plane anchored directly to a chosen layer or the primary model when the Depth of Field effect is enabled. Selecting None focuses at the standard look-at point.
| Control | Description |
|---|---|
| Layer | Selects the target layer or the Primary Model |
| Anchor | Selects the bone or root anchor kept in focus |
| Focus Damping | Seconds for the focus plane to catch up as the target moves along the camera axis (0 to 10 seconds, default 0.2 seconds; 0 refocuses immediately) |
When the target moves behind the camera's view frustum, focus automatically falls back to the standard look-at distance.
Handheld Movement
Adds subtle, procedural noise-driven sway to the camera view, recreating the natural breathing motion of a physical handheld shot (inspired by Warudo's Handheld Movement).
| Control | Description |
|---|---|
| Handheld Movement | Toggles handheld sway on or off |
| Intensity | Magnitude of camera sway (0 to 1, default 0.5) |
| Speed | Frequency and cadence of sway motion (0 to 2, default 1) |
Handheld sway operates purely as a real-time viewport offset and does not modify saved camera transform coordinates. Sway pauses automatically during interactive 3D gizmo drags and camera motion track playback.
Automations
Automations can also drive the camera:
- Camera Position — switch between preset camera angles.
- Play Camera Motion — trigger a motion clip during a broadcast.
- Camera Follow, Camera Aim, Camera Focus, and Handheld Movement — set or clear the layer the camera follows, aims at, or focuses on, and turn sway on or off.
Last updated on October 21, 2026