mocopi is Sony's wearable motion capture kit. Persona provides native protocol support without requiring third-party middleware — the mocopi mobile app streams motion data directly to your computer, and Persona decodes the skeleton hierarchy and frame poses to drive VRM avatars.
Like VMC, mocopi operates on the Pose Tracking channel. Because Live2D models lack humanoid skeletal rigs, they do not appear in mocopi source assignment lists.
Setup
- Under Pose Tracking, select Add Pose Source… to create a mocopi source, then check the target VRM models under Assign Avatars.
- In the mocopi mobile app under external device settings, set the transfer format to native mocopi (UDP), enter the local IPv4 address of the computer running Persona, and configure the same port number as the source (default:
12351). Ensure both devices are connected to the same local network. - Complete calibration in the mocopi app, switch on Enable Pose Tracking in Persona, and start transmitting motion data from the mobile device.
Ports
Identical to VMC, mocopi sources match incoming streams by port number rather than sender IP address. The default port is 12351. Adding new sources automatically assigns the next available unused port number.
Skeleton Definitions
Persona waits for an initial skeleton definition before applying any motion data. The skeleton definition establishes the performer's resting T-pose; all subsequent frame packets represent angular offsets relative to that rest pose. If you connect to an ongoing stream and the avatar does not move, stop and restart transmission in the mocopi app to resend the skeleton definition.
Each port binds to a single transmitting device at any given moment. To switch devices, wait for the previous session lock to expire; the incoming device must likewise transmit its own rest skeleton definition upon connecting.
mocopi Protocol vs. VMC
The mocopi mobile app supports two external streaming protocols. Configure your Persona source to match the active transfer format in the app:
| App Transfer Format | Persona Source | Default Port |
|---|---|---|
| mocopi (UDP) | mocopi | 12351 |
| VMC | VMC Protocol | 39539 |
The native protocol avoids running intermediate bridge applications. If you already route mocopi data through VSeeFace or similar tools converted to VMC, you can continue using that workflow and configure separate sources for each protocol.
How the Pose Reaches the Rig
A mocopi stream delivers rotation data for 27 joints. Persona calculates the orientation difference between each joint's current pose and its rest skeleton pose, converting that offset into local rotations on the VRM's normalized humanoid bones. This mechanism eliminates the performer's initial rest stance differences and extracts purely the intentional movement. All 27 joints participate in retargeting, including nodes that lack direct humanoid bone counterparts in the VRM specification (such as intermediate thoracic and cervical spine joints) — ensuring subtle torso and neck nuances are preserved. When a VRM model omits optional humanoid bones, their rotations are absorbed by the nearest existing descendant bone.
Translational position is sampled exclusively from the hips. Positional offsets from other joints reflect the performer's physical body proportions and are not mapped directly to avatar models. Persona treats the first received hips coordinate as the world origin, recalibrating the anchor whenever the skeleton definition updates or after connection dropouts.
This pipeline performs rotational retargeting rather than inverse kinematics (IK) solving. Leg raises and crouching are reproduced strictly from joint angles; Persona does not apply automatic foot-to-ground placement locking.
When face tracking is simultaneously driving the head and neck, face tracking rotations take precedence; eye gaze and jaw motion are always governed by face and expression tracking. Motion smoothing, dropout pose hold, and smooth interpolation back to idle when disabled share the same pose driver architecture used by VMC.
mocopi sensors do not capture finger movement. Assigning an avatar's hands to Webcam Tracking overlays wrist and finger tracking atop mocopi shoulder and elbow rotations. When set to Fingers Only mode, wrist rotation remains driven by mocopi.
Last updated on October 11, 2026