CV VTuber Example

iOS & Android & Windows10 UWP support

Win & Mac & Linux Standalone support

WebGL support

ChromeOS support

visionOS support(beta)

Support for preview in the Editor

This asset requires “OpenCV for Unity” and “Dlib FaceLandmark Detector”.

Official Site | ExampleCode | Android Demo WebGL Demo

Features:

- CVVTuberExample(Computer Vision Virtual YouTuber Example) is an example project of controlling 3D humanoid model (Mecanim Humanoid, "Unity-chan!" Model, VRM Model, Live2DCubism5 Model) using WebCamTexture. You can control the head orientation and the facial expression of the 3D humanoid model using WebCamTexture only.

- The head orientation and face expression are controlled by the following procedure.

1. WebCamTextureMatSourceGetter - Convert WebCamTexture to OpenCV's Mat class.

2. DlibFaceLandmarkGetter - Detect a face landmark points from OpenCV's Mat class.

3. DlibHeadRotationGetter - Estimate head orientation from face landmark points.

4. HeadRotationController - Control the head orientation of the 3D model using the estimated head orientation.

5. HeadLookAtIKController - Set Animator.SetLookAtPosition() method using the estimated head orientation.

6. DlibFaceBlendShapeController - Control the face BlendShape of the 3D model using the face landmark point.

Basic Examples:

- MultiSource CV VTuber Example - Control 3D models using video files and webcam feed

Advanced Examples: (require add-ons setup)

- UnityChan CV VTuber Example - Unity-chan! model integration

- VRM10 CV VTuber Example - VRM 1.0 format support

- Live2DCubism5 CV VTuber Example - Live2D Cubism 5 integration

The SDK and model data are not included in the asset package.

System Requirements

Build Win Standalone & Preview Editor : Windows 8 or later

Build Mac Standalone & Preview Editor : OSX 10.13 or later

Build Linux Standalone & Preview Editor : Ubuntu18.04 or later

Build Android : API level 24 or later

Build iOS : iOS Version 12.0 or later

Build VisionOS : visionOS 1 or later (beta)