Expo Docs


ARCore is not yet supported. This lib is iOS only right now. Augmented Reality with ARKit for iOS This library is generally used with expo-three to generate a camera, and manage a 3D scene.

This will check the following condition:
  • Device year is greater than 2014
  • Device is not a simulator
  • Device is iOS, and not Android

Get the version of ARKit running on the device. iOS 11 devices come with 1.0, and the newly released iOS 11.3 comes with ARKit 1.5;

All listeners will return an object capable of removing itself as such:
const listener = AR.onFrameDidUpdate(() => {});
Optionally you can also remove all listeners for a single event.


This will update everytime the ARSession has updated it's frame (has new data)

This will be called with the localized description of any Error thrown by ARKit

Invoked when an anchor is found, updated, or removed. This is the primary way to get data from Detection Images, Planes, Faces, and general Anchors.
AR.onAnchorsDidUpdate(({anchors, eventType}) => {
  for (let anchor of anchors) {
    if (anchor.type === AR.AnchorTypes.Anchor) {
      const { identifier, transform } = anchor;

      if (eventType === AR.AnchorEventTypes.Add) {
        // Something added!  
      } else if (eventType === AR.AnchorEventTypes.Remove) {
        // Now it's changed
      } else if (eventType === AR.AnchorEventTypes.Update) {
        // Now it's gone...

Called whenever the camera changes it's movement / tracking state. Useful for telling the user how to better hold and move the camera.

Simply this is called when the app backgerounds.

This is called when the app returns to the foreground.

Maybe the most powerful function, hit testing allows you to get real world info on a particular position in the screen.

The point is a normalized value, meaning it is between 0-1 this can be achieved by dividing the dimension position by size. Ex: { x: x / width, y: y / height }
const normalizedPoint = { x, y };
const hitTestResultTypes =  AR.HitTestResultTypes.HorizontalPlane;
const { hitTest } = AR.performHitTest(normalizedPoint, hitTestResultTypes);
for (let hit of hitTest) {
  const { 
    anchor: { 
    } = hit;

Possible types for specifying a hit-test search, or for the result of a hit-test search.
HitTestResultTypes = {
   * Result type from intersecting the nearest feature point.
  FeaturePoint: 'featurePoint',
   * Result type from intersecting a horizontal plane estimate, determined for the current frame.
  HorizontalPlane: 'horizontalPlane',
   * Result type from intersecting a vertical plane estimate, determined for the current frame.
  VerticalPlane: 'verticalPlane',
   * Result type from intersecting with an existing plane anchor.
  ExistingPlane: 'existingPlane',
   * Result type from intersecting with an existing plane anchor, taking into account the plane’s extent.
  ExistingPlaneUsingExtent: 'existingPlaneUsingExtent',
   * Result type from intersecting with an existing plane anchor, taking into account the plane’s geometry.
  ExistingPlaneUsingGeometry: 'existingPlaneUsingGeometry',

Given an image, ARKit will update you when it finds it in the real world.
Make sure that all reference images are greater than 100 pixels and have a positive physical size in meters.

const asset = Expo.Asset.fromModule(require('./image.png'))
await asset.downloadAsync();

await AR.setDetectionImagesAsync({
  myDopeImage: {
     * The local uri of the image, this can be obtained with Expo.Asset.fromModule()
    uri: asset.localUri,
     * Name used to identify the Image Anchor returned in a `onAnchorsDidUpdate` listener.
    name: 'myDopeImage',
     * Real-world size in meters. 
    width: 0.1, 

AR.onAnchorsDidUpdate(({anchors, eventType}) => {
  for (let anchor of anchors) {
    if (anchor.type === AR.AnchorTypes.Image) {
      const { identifier, image, transform } = anchor;

      if (eventType === AR.AnchorEventTypes.Add) {
        // Add some node
      } else if (eventType === AR.AnchorEventTypes.Remove) {
        // Remove that node
      } else if (eventType === AR.AnchorEventTypes.Update) {
        // Update whatever node

This synchronous function can return anchors, raw feature points, light estimation, and captured depth data.

You need to provide the attributes you want to recieve.
const {
  } = AR.getCurrentFrame({
  anchors: { 
    [AR.AnchorTypes.Face]: {
        blendShapes: true, 
        geometry: true
  rawFeaturePoints: true, 
  lightEstimation: true, 
  capturedDepthData: true

Getting face data
In order to be efficent you can specify which blendShapes you would like to have returned. You can do filter data like this:
    anchors: { 
        [AR.AnchorTypes.Face]: {
            blendShapes: [
This will return just the amount your left and right eyebrows are down.
This can return the following Anchor Types:
  • ARAnchor
  • ARPlaneAnchor
  • ARImageAnchor
  • ARFaceAnchor
type: "ARAnchor", // AR.AnchorTypes.Anchor
transform: anchor.transform,
identifier: anchor.identifier
type: "ARPlaneAnchor", // AR.AnchorTypes.Plane
transform: anchor.transform,
identifier: anchor.identifier,
center: { 
    x: Float, 
    y: Float, 
    z: Float 
extent: {
    width: Float,
    length: Float
type: "ARImageAnchor", // AR.AnchorTypes.Image
transform: anchor.transform,
identifier: anchor.identifier,
image: { 
    name: anchor.referenceImage.name,
    size: { // Physical size in meters
        width: Float,
        height: Float,
type: "ARFaceAnchor", // AR.AnchorTypes.Face
transform: anchor.transform,
identifier: anchor.identifier,
isTracked: Bool,
geometry: {
    vertexCount: Int, // ARFaceAnchor.geometry.vertexCount
    textureCoordinateCount: Int, // ARFaceAnchor.geometry.textureCoordinateCount
    triangleCount: Int, // ARFaceAnchor.geometry.triangleCount
    vertices: [ { x: Float, y: Float, z: Float } ],
    textureCoordinates: [ { u: Float, v: Float } ],
    triangleIndices: [ Int ],
blendShapes: {
    browDown_L: Float, // AR.BlendShapes.BrowDownL
    browDown_R: Float, // AR.BlendShapes.BrowDownR
    browInnerUp: Float, // AR.BlendShapes.BrowInnerUp
    browOuterUp_L: Float, // AR.BlendShapes.BrowOuterUpL
    browOuterUp_R: Float, // AR.BlendShapes.BrowOuterUpR
    cheekPuff: Float, // AR.BlendShapes.CheekPuff
    cheekSquint_L: Float, // AR.BlendShapes.CheekSquintL
    cheekSquint_R: Float, // AR.BlendShapes.CheekSquintR
    eyeBlink_L: Float, // AR.BlendShapes.EyeBlinkL
    eyeBlink_R: Float, // AR.BlendShapes.EyeBlinkR
    eyeLookDown_L: Float, // AR.BlendShapes.EyeLookDownL
    eyeLookDown_R: Float, // AR.BlendShapes.EyeLookDownR
    eyeLookIn_L: Float, // AR.BlendShapes.EyeLookInL
    eyeLookIn_R: Float, // AR.BlendShapes.EyeLookInR
    eyeLookOut_L: Float, // AR.BlendShapes.EyeLookOutL
    eyeLookOut_R: Float, // AR.BlendShapes.EyeLookOutR
    eyeLookUp_L: Float, // AR.BlendShapes.EyeLookUpL
    eyeLookUp_R: Float, // AR.BlendShapes.EyeLookUpR
    eyeSquint_L: Float, // AR.BlendShapes.EyeSquintL
    eyeSquint_R: Float, // AR.BlendShapes.EyeSquintR
    eyeWide_L: Float, // AR.BlendShapes.EyeWideL
    eyeWide_R: Float, // AR.BlendShapes.EyeWideR
    jawForward: Float, // AR.BlendShapes.JawForward
    jawLeft: Float, // AR.BlendShapes.JawLeft
    jawOpen: Float, // AR.BlendShapes.JawOpen
    jawRight: Float, // AR.BlendShapes.JawRight
    mouthClose: Float, // AR.BlendShapes.MouthClose
    mouthDimple_L: Float, // AR.BlendShapes.MouthDimpleL
    mouthDimple_R: Float, // AR.BlendShapes.MouthDimpleR
    mouthFrown_L: Float, // AR.BlendShapes.MouthFrownL
    mouthFrown_R: Float, // AR.BlendShapes.MouthFrownR
    mouthFunnel: Float, // AR.BlendShapes.MouthFunnel
    mouthLeft: Float, // AR.BlendShapes.MouthLeft
    mouthLowerDown_L: Float, // AR.BlendShapes.MouthLowerDownL
    mouthLowerDown_R: Float, // AR.BlendShapes.MouthLowerDownR
    mouthPress_L: Float, // AR.BlendShapes.MouthPressL
    mouthPress_R: Float, // AR.BlendShapes.MouthPressR
    mouthPucker: Float, // AR.BlendShapes.MouthPucker
    mouthRight: Float, // AR.BlendShapes.MouthRight
    mouthRollLower: Float, // AR.BlendShapes.MouthRollLower
    mouthRollUpper: Float, // AR.BlendShapes.MouthRollUpper
    mouthShrugLower: Float, // AR.BlendShapes.MouthShrugLower
    mouthShrugUpper: Float, // AR.BlendShapes.MouthShrugUpper
    mouthSmile_L: Float, // AR.BlendShapes.MouthSmileL
    mouthSmile_R: Float, // AR.BlendShapes.MouthSmileR
    mouthStretch_L: Float, // AR.BlendShapes.MouthStretchL
    mouthStretch_R: Float, // AR.BlendShapes.MouthStretchR
    mouthUpperUp_L: Float, // AR.BlendShapes.MouthUpperUpL
    mouthUpperUp_R: Float, // AR.BlendShapes.MouthUpperUpR
    noseSneer_L: Float, // AR.BlendShapes.NoseSneerL
    noseSneer_R: Float, // AR.BlendShapes.NoseSneerR

Matrix data can be used in three.js (with expo-three) to generate a 3D camera.

getARMatrices(arsession, width, height, near, far) is now getARMatrices(near, far)

Given reference to a GLView and a ARTrackingConfiguration, this will create an ARSession associated with the GLContextstartARSessionAsync(view, trackingConfiguration)
startARSessionAsync is now startAsync
When invoked, this method will tear-down the ARSession, and WebGLTexture used for the camera stream. This is an end-of-lifecycle method.stopAsync()
stopARSessionAsync() is now stopAsync

Used to reset the anchors and current data in the ARSession.reset()
Used to pause the ARSession.pause()
Used to resume the ARSession after pausing it.resume()

Check availability.isConfigurationAvailable(configuration: TrackingConfiguration): Bool
A Configuration defines how ARKit constructs a scene based on real-world device motion.setConfigurationAsync(configuration: TrackingConfiguration)
Alternatively you could also use:
  • isFrontCameraAvailable()
  • isRearCameraAvailable()
TrackingConfigurations = {
   * Provides high-quality AR experiences that use the rear-facing camera precisely track a device's position and orientation and allow plane detection and hit testing.
  world: 'ARWorldTrackingConfiguration',
   * Provides basic AR experiences that use the rear-facing camera and track only a device's orientation.
  orientation: 'AROrientationTrackingConfiguration',
   * Provides AR experiences that use the front-facing camera and track the movement and expressions of the user's face.
  face: 'ARFaceTrackingConfiguration',

Used to enable or disable plane detection (ARPlaneAnchors will be able to return in onAnchorsDidUpdate).
As of iOS 11.3 (ARKit 1.5) you can now enable vertical plane detection.Default: AR.PlaneDetectionTypes.NoneGET: planeDetection(): PlaneDetectionSET: setPlaneDetection(planeDetection: PlaneDetection)
Options for whether and how ARKit detects flat surfaces in captured images.
PlaneDetectionTypes = {
   * No plane detection is run.
  None: 'none',
   * Plane detection determines horizontal planes in the scene.
  Horizontal: 'horizontal',
   * Plane detection determines vertical planes in the scene.
  Vertical: 'vertical',

The center of the 3D space used by ARKitsetWorldOriginAsync(matrix4x4)
A Matrix 4x4 is an array of 16 doubles

Estimated scene lighting information associated with a captured video frame in an AR session.Default: trueSET: setLightEstimationEnabled(value: Boolean)GET: getLightEstimationEnabled(): Boolean
Light estimation can be retrieved through getCurrentFrame with the lightEstimation key added.

A Boolean value that specifies whether to capture audio during the AR session. You cannot currently access the audio data as it is useless.Default: falseSET: setProvidesAudioData(value: Boolean)GET: getProvidesAudioData(): Boolean

iOS 11.3+ A Boolean value that determines whether the device camera uses fixed focus or autofocus behavior.Default: trueSET: setAutoFocusEnabled(value: Boolean)GET: getAutoFocusEnabled(): Boolean

Options for how ARKit constructs a scene coordinate system based on real-world device motion.Default: AR.WorldAlignmentTypes.GravitySET: setWorldAlignment(worldAlignment: WorldAlignment)GET: getWorldAlignment(): WorldAlignment
const WorldAlignmentTypes = {
  Gravity: 'gravity',
  GravityAndHeading: 'gravityAndHeading',
  AlignmentCamera: 'alignmentCamera',

The internal ID used to render the camera texture.GET: getCameraTexture(): Number

iOS 11.3+ The set of video capture formats available on the current device. The video format cannot be set yet.Default: The first element returned is the default valueGET: AR.getSupportedVideoFormats(configuration: TrackingConfiguration): Array
    imageResolution: {
        width: 1920,
        height: 1080
    framesPerSecond: 60