Lens Studio Face Tracking — Reference Guide

Face tracking is the foundation of most Snapchat phone lenses. Lens Studio tracks faces in real time using the front camera, providing mesh, landmarks, attachment points, and expression weights.

---

FaceTrackingComponent

Setting up face tracking

Add a Head asset from the Add menu (Scene Hierarchy panel → + → Face → Head). This creates a face tracking hierarchy automatically.

To access the face tracking component in script:

// Get the FaceTracking component from the head object
const faceTracking = headSceneObject.getComponent('FaceTracking')

Multi-face tracking

Lens Studio can track up to 3 faces simultaneously. Use faceIndex to distinguish them:

@input faceIndex: number = 0  // 0 = first face, 1 = second, 2 = third

const faceTracking = headSceneObject.getComponent('FaceTracking')
faceTracking.faceIndex = this.faceIndex

Each tracked face has its own head scene object. Duplicate the head hierarchy for each additional face.

---

2D & 3D Face Attachments

Objects parented to specific named landmarks follow the face in 3D.

Common attachment anchor names

Anchor	Position
hat	Top of head
forehead	Centre forehead
left_eye	Left eye centre
right_eye	Right eye centre
nose_tip	Tip of nose
mouth	Centre of mouth
chin	Bottom of chin
left_ear	Left ear
right_ear	Right ear

Adding a 3D attachment

1. Add your 3D object to the scene.
2. Parent it to the corresponding anchor object in the face tracking hierarchy (e.g., drag it under hat in the Scene Hierarchy panel).
3. Use Adjust mode on the face mesh to position the object correctly.

Scripting 2D attachment position

// Get the screen-space position of a face landmark
const faceTracking = headObject.getComponent('FaceTracking')

const updateEvent = this.createEvent('UpdateEvent')
updateEvent.bind(() => {
  // Screen-space position of the nose tip (normalised UV)
  const nosePos: vec2 = faceTracking.getLandmark(FaceTracking.LandmarkIndex.NoseTip)
  print(`Nose at: ${nosePos.x}, ${nosePos.y}`)
})

---

Face Mesh

The Face Mesh conforms a textured mesh to the detected face geometry.

Applying a texture to the face mesh

1. Add a Face Mesh to the scene (+ → Face → Face Mesh).
2. Create or import a texture that maps to the UV layout of the face mesh.
3. Assign the texture to the Face Mesh's material.

Scripting face mesh properties

const meshVisual = faceMeshObject.getComponent('Component.BaseMeshVisual')
const mat = meshVisual.material.clone()
meshVisual.material = mat

// Tint the face mask
mat.mainPass.baseColor = new vec4(0.2, 0.8, 0.4, 0.6)  // RGBA
mat.mainPass.blendMode = BlendMode.Normal

---

Face Landmarks (68 Keypoints)

Lens Studio provides 68 facial landmark points that track with the face.

const faceTracking = headObject.getComponent('FaceTracking')

// Get all 68 landmark positions (in local face space)
const landmarks: vec3[] = faceTracking.getFacePositions()

// Common landmark indices (0-indexed)
// 0–16: Jaw line
// 17–21: Left eyebrow
// 22–26: Right eyebrow
// 27–35: Nose bridge and base
// 36–41: Left eye
// 42–47: Right eye
// 48–67: Mouth

const leftEyeCenter: vec3 = landmarks[37]   // mid-left-eye
const rightEyeCenter: vec3 = landmarks[43]  // mid-right-eye
const mouthCenter: vec3 = landmarks[51]     // upper lip centre

---

Face Expressions (Expression Weights)

Expression weights give you normalised [0,1] values for various facial actions.

const faceTracking = headObject.getComponent('FaceTracking')

const updateEvent = this.createEvent('UpdateEvent')
updateEvent.bind(() => {
  const expressions = faceTracking.getFaceExpressionWeights()

  // Common expression keys:
  const mouthOpen: number   = expressions[FaceTracking.Expressions.MouthOpen]
  const leftBlink: number   = expressions[FaceTracking.Expressions.EyeBlinkLeft]
  const rightBlink: number  = expressions[FaceTracking.Expressions.EyeBlinkRight]
  const smileLeft: number   = expressions[FaceTracking.Expressions.SmileLeft]
  const smileRight: number  = expressions[FaceTracking.Expressions.SmileRight]
  const browRaiseLeft: number = expressions[FaceTracking.Expressions.BrowRaiseLeft]

  if (mouthOpen > 0.5) {
    print('Mouth open!')
    triggerEffect()
  }

  if (leftBlink > 0.8 && rightBlink < 0.2) {
    print('Left wink detected')
  }
})

---

Eye Tracking

// Add an Eye Tracking component to your scene object
const eyeTracking = this.sceneObject.getComponent('Component.EyeTrackingComponent')

const updateEvent = this.createEvent('UpdateEvent')
updateEvent.bind(() => {
  // Eye gaze direction in world space
  const leftGazeDir: vec3  = eyeTracking.leftEye.gaze
  const rightGazeDir: vec3 = eyeTracking.rightEye.gaze

  // Eye openness (0 = closed, 1 = fully open)
  const leftOpenness: number  = eyeTracking.leftEye.openness
  const rightOpenness: number = eyeTracking.rightEye.openness
})

---

Face Effects

These are inspector-based effects added via the Add Component menu. Most require no scripting.

Effect	What it does	Scripting access
Face Retouch	Skin smoothing, teeth whitening	Inspector sliders
Eye Color	Recolours the iris	eyeColorComponent.color
Face Liquify	Warp face geometry (e.g., bigger eyes, slimmer face)	Inspector sliders
Face Stretch	Distort facial proportions for comedy effects	Inspector sliders
Face Inset	Embeds a live camera view in the face area	Position/scale in inspector
Face Mask	Overlays a 2D texture mapped to face UV	Material on Face Mesh
Face Texture	Replaces the face with a static or animated image	Texture asset

Changing eye color at runtime

const eyeColorComponent = this.sceneObject.getComponent('Component.EyeColorVisual')
eyeColorComponent.color = new vec4(0.0, 0.4, 1.0, 1.0)  // blue eyes

---

Upper Body Tracking (Spectacles + Front Camera)

Upper Body Tracking 3D

Provides a subset of humanoid attachment points: hips, spine bones, neck, head, shoulders, and upper arms.

1. Add Object Tracking → Upper Body Tracking from the Scene Hierarchy + menu.
2. The component tracks the skeleton and exposes attachment points via the ObjectTracking3D component.

const bodyTracking = upperBodyObject.getComponent('Component.ObjectTracking3D')

bodyTracking.onObjectFound.add(() => print('Upper body detected'))
bodyTracking.onObjectLost.add(() => print('Upper body lost'))

Upper Body Mesh

Creates a full, seamless 3D mesh of the upper body — ideal for selfie lenses. Can be used as:

• An occluder for body accessories (virtual clothes, jewellery)
• A physics collider surface
• A texture target for seamless face-to-neck blending

Add via Scene Hierarchy → + → Upper Body Mesh. Automatically includes:

• Head Component (with Face Mesh and skull)
• Object Tracking 3D (configured for Upper Body)
• Upper Body Mesh (the body surface mesh)

Note: Upper Body Mesh does not support external custom meshes — use the built-in mesh and apply custom materials.

---

Common Gotchas

• getFaceExpressionWeights() returns a Dictionary — check available keys using Object.keys() during development to verify available expression names on your target OS.
• Multi-face: each faceIndex needs its own separate Head scene object hierarchy — one FaceTracking component per face.
• Eye tracking is a separate component from face tracking — add EyeTrackingComponent in addition to FaceTracking.
• Upper Body Mesh is not available on all OS versions — check the Spectacles compatibility list.
• Face Retouch and morphing effects run on a GPU pass — avoid enabling more effects than needed for production performance.
• Face Attach objects: always test object placement by recording with a real face; the Lens Studio simulator's face is symmetrical and doesn't reflect real-world drift.