Purpose

Face Track follows a face and estimates its pose and expression. Its motion output connects directly to Portrait Rig. A smaller pose output exposes head movement, eye ratios and lip ratio as controls.

Use

Place Face Track under Analyze in the Image view and connect a camera or video to in. Start with the MediaPipe detector and inspect preview to check the face crop.

Connect motion to Portrait Rig’s motion input. Adjust the crop before adding smoothing or increasing the rig’s movement scales.

Inputs and outputs

• in supplies the image.
• motion carries the 74-value pose and expression input used by Portrait Rig.
• pose carries valid, pitch, yaw, roll, translation, scale, left and right eye ratios, and lip ratio.
• preview shows the tracked crop.

Controls

• Detector selects MediaPipe or InsightFace. MediaPipe is the default. InsightFace requires user-supplied models marked for non-commercial use.
• Detection Threshold controls face acquisition.
• Crop Scale and Crop V Offset adjust the tracked crop.
• Motion Smoothing blends new motion with earlier values.
• Redetect Face acquires the face again. Reset also clears stored motion and smoothing history.
• Model selects the motion model.

Notes

When tracking loses the face, the motion output retains the last pose and sets valid to zero. Check validity when using pose channels to drive your own controls. Portrait Rig uses that flag to hold the last accepted motion.

Ports

Inputs

• in image required
  RGBA

Outputs

• motion image
  74 x 1
• pose image
  11
• preview image
  RGBA

Parameters

Track

Detector (Detector) Menu op('face_track').par.Detector

mediapipe: works out of box, Apache-2.0. insightface: best quality; NON-COMMERCIAL user-supplied models.

Default:

mediapipe

Options:

mediapipe, insightface

Detection Threshold (Detthreshold) Float op('face_track').par.Detthreshold

Default:

0.5

Range:

0.05 to 0.95

Crop Scale (Cropscale) Float op('face_track').par.Cropscale

Default:

2.3

Range:

1.5 to 4

Crop V Offset (Cropvy) Float op('face_track').par.Cropvy

Default:

-0.125

Range:

-0.5 to 0.5

Motion Smoothing (Smoothing) Float op('face_track').par.Smoothing

Default:

0

Range:

0 to 0.99

Redetect Face (Redetect) Pulse op('face_track').par.Redetect

Default:

None

Reset (Reset) Pulse op('face_track').par.Reset

Default:

None

Model

Model (Modelpath) Str op('face_track').par.Modelpath

Default:

motion_extractor.safetensors