Send ARKit camera position and quaternion on OSC (SwiftOSC) - ios

I'm currently starting to learn Swift to develop an app as a part of my research. I'm ok with python but I can't seem to wrap my head around using Swift and Xcode. I've been able to print accelerometer data on screen, and send values over OSC. However, I need help:
Find ARKit camera position (X,Y,Z),
Send these values over OSC.
I've set up the ARSession and set the ARDelegate etc, and set up a function to get the position values from the transform matrix, but I cannot seem to be able to display any data on screen and send over OSC. The ARKIT code is below. The first code break is just an example of using SwiftOSC, it should be easy to implement once I am able to get the ARKit data.
I know that this may be pretty simple but I can't get it working! If anyone is willing to help me doing that, it would be amazing. And if someone is willing to work with me on Zoom or Meets, my thesis grant will be able to give honorarium compensation for your time.
Thanks in advance!
override func viewDidAppear(_ animated: Bool) {
motionManager.accelerometerUpdateInterval = (1/60)
motionManager.startAccelerometerUpdates(to: OperationQueue.current!) {(data, error)in
if let myData = data{
let x = myData.acceleration.x
let y = myData.acceleration.y
let z = myData.acceleration.z
self.xAccel.text = "x \(Double(x))"
let message = OSCMessage(OSCAddressPattern("/accelx"), Double(x))
client.send(message)
let message_y = OSCMessage(OSCAddressPattern("/accely"), Double(y))
client.send(message_y)
let message_z = OSCMessage(OSCAddressPattern("/accelz"), Double(z))
client.send(message_z)
}
}
}
However, I can't seem to access camera position using ARKit:
import UIKit
import SceneKit
import ARKit
extension ViewController: ARSessionDelegate {
func session(_ session: ARSession, didUpdate frame: ARFrame) {
let transform = frame.camera.transform
var position = transform.columns.3
print(position.x, position.y, position.z) //Get (x,y,z) and print
}
}
class ViewController: UIViewController, ARSCNViewDelegate {
#IBOutlet var sceneView: ARSCNView!
override func viewDidLoad() {
super.viewDidLoad()
// Set the view's delegate
sceneView.delegate = self
// Show statistics such as fps and timing information
sceneView.showsStatistics = true
// Create a new scene
let scene = SCNScene(named: "art.scnassets/ship.scn")!
// Set the scene to the view
sceneView.scene = scene
}
override func viewWillAppear(_ animated: Bool) {
super.viewWillAppear(animated)
// Create a session configuration
let configuration = ARWorldTrackingConfiguration()
// Run the view's session
sceneView.session.run(configuration)
// How to call function and assign position.[x,y,z], to variable to read on screen and send on OSC? So far I can't see any data at all.
}
override func viewWillDisappear(_ animated: Bool) {
super.viewWillDisappear(animated)
// Pause the view's session
sceneView.session.pause()
}
It always returns zero, or nothing at all.

Related

How to place a 3D model on top of a reference image with RealityKit?

I want to place a 3D Model from the local device on top of the reference image when it's recognized. To achieve this I have tried the following:
Adding the reference image to the session configuration:
override func viewDidLoad() {
super.viewDidLoad()
arView.session.delegate = self
// Check if the device supports the AR experience
if (!ARConfiguration.isSupported) {
TLogger.shared.error_objc("Device does not support Augmented Reality")
return
}
guard let qrCodeReferenceImage = UIImage(named: "QRCode") else { return }
let detectionImages: Set<ARReferenceImage> = convertToReferenceImages([qrCodeReferenceImage])
let configuration = ARWorldTrackingConfiguration()
configuration.detectionImages = detectionImages
arView.session.run(configuration, options: [.resetTracking, .removeExistingAnchors])
}
Using the ARSessionDelegate to get notified when the reference image was detected and placing the 3D model at the same position as his ARImageAnchor:
func session(_ session: ARSession, didAdd anchors: [ARAnchor]) {
for anchor in anchors {
guard let imageAnchor = anchor as? ARImageAnchor else { return }
let position = imageAnchor.transform
addEntity(self.localModelPath!, position)
}
}
func addEntity(_ modelPath: URL, _ position: float4x4) {
// Load 3D Object as Entity
let entity = try! Entity.load(contentsOf: modelPath)
// Create the Anchor which gets added to the AR Scene
let anchor = AnchorEntity(world: position)
anchor.addChild(entity)
anchor.transform.matrix = position
arView.scene.addAnchor(anchor)
}
However, whenever I try to place the anchor including my 3D model (the entity) at a specific position, it doesn't appear in the arView. It seems like the model is getting loaded though since a few frames are getting lost when executing the addEntity function. When I don't specifically set the anchors position the model appears in front of the camera.
Can anyone lead me in the right direction here?
Solution I
To make your code work properly, remove this line:
anchor.transform.matrix = position
#TimLangner – "...But I want to make the model appear on top of the reference image and not at any other location..."
As you can see, my test sphere has appeared on top of the reference image. When changing its position, remember that the Y-axis of the image anchor is directed towards the camera if the QR is vertical, and is directed up if the QR code is on a horizontal surface.
Make sure, a pivot point is located where it should.
In my case (you can see that I'm using AppClipCode), to move sphere 30 cm up, I have to move along negative Z direction.
entity.position.z = anchor.position.z - 0.3
Solution II
In my opinion, the most simple and productive solution would be to use the RealityKit's native AnchorEntity(.image(...)), without the need of implementing ARImageAnchor in delegate's method.
AnchorEntity(.image(group: "GroupName", name: "forModel"))
Here's a code:
import UIKit
import RealityKit
class ViewController: UIViewController {
#IBOutlet var arView: ARView!
override func viewDidLoad() {
super.viewDidLoad()
let entity = ModelEntity(mesh: .generateSphere(radius: 0.1))
let anchor = AnchorEntity(.image(group: "AR Resources",
name: "appClipCode"))
anchor.addChild(entity)
entity.position.z = anchor.position.z - 0.3
arView.scene.anchors.append(anchor)
}
}

iOS ARKit: Large size object always appears to move with the change in the position of the device camera

I am creating an iOS ARKit app where I wanted to place a large object in Augmented Reality.
When I am trying to place the object at a particular position it always appears to be moving with the change in camera position and I am not able to view the object from all angles by changing the camera position.
But if I reduce it's scale value to 0.001 (Reducing the size of the object), I am able to view the object from all angles and the position of the placed object also does not change to that extent.
Bounding Box of the Object:-
Width = 3.66
Height = 1.83
Depth = 2.438
Model/Object Url:-
https://drive.google.com/open?id=1uDDlrTIu5iSRJ0cgp70WFo7Dz0hCUz9D
Source Code:-
import UIKit
import ARKit
import SceneKit
class ViewController: UIViewController {
#IBOutlet weak var sceneView: ARSCNView!
private let configuration = ARWorldTrackingConfiguration()
private var node: SCNNode!
//MARK: - Life cycle
override func viewDidLoad() {
super.viewDidLoad()
self.sceneView.showsStatistics = false
self.sceneView.debugOptions = [ARSCNDebugOptions.showFeaturePoints]
self.sceneView.automaticallyUpdatesLighting = false
self.sceneView.delegate = self
self.addTapGesture()
}
override func viewWillAppear(_ animated: Bool) {
super.viewWillAppear(animated)
configuration.planeDetection = .horizontal
self.sceneView.session.run(configuration)
}
override func viewWillDisappear(_ animated: Bool) {
super.viewWillDisappear(animated)
self.sceneView.session.pause()
}
//MARK: - Methods
func addObject(hitTestResult: ARHitTestResult) {
let scene = SCNScene(named: "art.scnassets/Cube.obj")!
let modelNode = scene.rootNode.childNodes.first
modelNode?.position = SCNVector3(hitTestResult.worldTransform.columns.3.x,
hitTestResult.worldTransform.columns.3.y,
hitTestResult.worldTransform.columns.3.z)
let scale = 1
modelNode?.scale = SCNVector3(scale, scale, scale)
self.node = modelNode
self.sceneView.scene.rootNode.addChildNode(modelNode!)
let lightNode = SCNNode()
lightNode.light = SCNLight()
lightNode.light?.type = .omni
lightNode.position = SCNVector3(x: 0, y: 10, z: 20)
self.sceneView.scene.rootNode.addChildNode(lightNode)
let ambientLightNode = SCNNode()
ambientLightNode.light = SCNLight()
ambientLightNode.light?.type = .ambient
ambientLightNode.light?.color = UIColor.darkGray
self.sceneView.scene.rootNode.addChildNode(ambientLightNode)
}
private func addTapGesture() {
let tapGesture = UITapGestureRecognizer(target: self, action: #selector(didTap(_:)))
self.sceneView.addGestureRecognizer(tapGesture)
}
#objc func didTap(_ gesture: UIPanGestureRecognizer) {
let tapLocation = gesture.location(in: self.sceneView)
let results = self.sceneView.hitTest(tapLocation, types: .featurePoint)
guard let result = results.first else {
return
}
let translation = result.worldTransform.translation
guard let node = self.node else {
self.addObject(hitTestResult: result)
return
}
node.position = SCNVector3Make(translation.x, translation.y, translation.z)
self.sceneView.scene.rootNode.addChildNode(self.node)
}
}
extension float4x4 {
var translation: SIMD3<Float> {
let translation = self.columns.3
return SIMD3<Float>(translation.x, translation.y, translation.z)
}
}
GIF of the Problem:-
Video URL of the Problem:-
https://drive.google.com/open?id=1E4euZ0ArEtj2Ffto1pAOfVZocV08EYKN
Approaches Tried:-
Tried to place the object at the origin
modelNode?.position = SCNVector3(0, 0, 0)
Tried to place the object at some distance away from the device camera
modelNode?.position = SCNVector3(0, 0, -800)
Tried with the different combinations of worldTransform/localTransform columns
modelNode?.position = SCNVector3(hitTestResult.worldTransform.columns.3.x, hitTestResult.worldTransform.columns.3.y, hitTestResult.worldTransform.columns.3.z)
modelNode?.position = SCNVector3(hitTestResult.worldTransform.columns.2.x, hitTestResult.worldTransform.columns.2.y, hitTestResult.worldTransform.columns.2.z)
modelNode?.position = SCNVector3(hitTestResult.worldTransform.columns.1.x, hitTestResult.worldTransform.columns.1.y, hitTestResult.worldTransform.columns.1.z)
modelNode?.position = SCNVector3(hitTestResult.worldTransform.columns.1.x, hitTestResult.worldTransform.columns.2.y, hitTestResult.worldTransform.columns.3.z)
modelNode?.position = SCNVector3(hitTestResult.localTransform.columns.3.x, hitTestResult.localTransform.columns.3.y, hitTestResult.localTransform.columns.3.z)
But still of no luck. It still appears to be moving with the device camera and not stuck to a position where it has been placed.
Expected Result:-
Object should be of actual size (Scale should be of 1.0). Their should be no reduction in the scale value.
Once placed at a particular position it should not move with the movement of the device camera.
Object can be seen from all angles with the movement of the device camera without any change in object position.
Unlike stated in the accepted answer, the issue is probably not about the tracking quality or a bug in the model. It looks like the model is not correctly placed on top of the ground, probably due to a mispositioned pivot point, and some part of the model stays under ground. So when you move the camera, since the part under ground is not occluded by the floor, it looks like it is shifting. Have a look at this picture:
The pivot points of the models provided by Apple are positioned correctly so that when it is placed on top of a plane on the ground, its parts stay above ground.
If you correctly position the pivot point of the model, it should work correctly, independent of the model type.
I found out the root cause of the issue. The issue was related to the Model which I was using for AR. When, I replaced the model with the one provided in this link:- https://developer.apple.com/augmented-reality/quick-look/. I was not facing any issues. So, if anyone face such type of issues in future I would recommend to use any of the model provided by Apple to check if the issue persists with it or not.
I experienced the same issue.
Whenever I try to change anything in our Model of .usdz type (which is actually an Encrypted and compressed type) we cannot edit or change anything in it. If I edit or change a little position then it behaves same way as highlighted in the question.
To handle this issue, I just moved the old model .usdz to trash and and copied the original file (again) to XCode and then it worked.

First Person Experience in RealityKit

I am fairly new to Swift and just started playing with RealityKit and ARKit. I am working on a personal project where I would like to have a 3D object stick to the camera in first person. Similar to the AR Angry birds or any FPS game. I have seen a few examples in SceneKit or SpriteKit, I am sure it is just a misunderstanding of how anchoring entities works.
My main question is:
How would I go about sticking a reality object I created in Reality Composer to the camera in first person? I want to create a Reality Scene, in this case an arm cannon and upon tapping it shots.
Below is the code for my ViewController
extension ViewController: ARSessionDelegate
{
func session(_ session: ARSession, didUpdate frame: ARFrame)
{
guard let arCamera = session.currentFrame?.camera else { return }
// Probably where I update the location of my reality experience
}
}
class ViewController: UIViewController
{
#IBOutlet var arView: ARView!
override func viewDidLoad()
{
super.viewDidLoad()
arView.session.delegate = self
// Load the "ArmCannon" scene from the "Experience" Reality File
let armCannonAnim = try! Experience.loadArmcannon()
// Create Anchor to anchor arm cannon to
let anchor = AnchorEntity(.camera)
anchor.transform = arView.cameraTransform
// Add the anchor to the scene
arView.scene.addAnchor(anchor)
// Setup tap gesture on arm cannon
let tapGesture = UITapGestureRecognizer(target: self, action:#selector(onTap))
arView.addGestureRecognizer(tapGesture)
// Add the the cannon animation to arView
arView.scene.anchors.append(armCannonAnim)
}
#IBAction func onTap(_ sender: UITapGestureRecognizer)
{
let tapLocation = sender.location(in: arView)
// Get the entity at the location we've tapped, if one exists
if let cannonFiring = arView.entity(at: tapLocation)
{
print(cannonFiring.name)
print("firing Cannon")
}
}
}
I have looked at and read Track camera position with RealityKit and Where is the .camera AnchorEntity located?
Instead of:
arView.scene.anchors.append(armCannonAnim)
put:
anchor.addChild(armCannonAnim)
You need this armCannonAnim to be a child of the camera, and the anchor object is an anchor at the camera transform. This is equivalent to in SceneKit adding a child to the cameraNode.

How to tap and move scene nodes in ARKit

I'm currently trying to build an AR Chess app and I'm having trouble getting the movement of the pieces working.
I would like to be able to tap on a chess piece, then the legal moves it can make on the chess board will be highlighted and it will move to whichever square the user tapped on.
Pic of the chess board design and nodes:
https://gyazo.com/2a88f9cda3f127301ed9b4a44f8be047
What I would like to implement:
https://imgur.com/a/IGhUDBW
Would greatly appreciate any suggestions on how to get this working.
Thanks!
ViewController Code:
import UIKit
import SceneKit
import ARKit
class ViewController: UIViewController, ARSCNViewDelegate {
#IBOutlet var sceneView: ARSCNView!
override func viewDidLoad() {
super.viewDidLoad()
// Set the view's delegate
sceneView.delegate = self
// Show statistics such as fps and timing information
sceneView.showsStatistics = true
// Add lighting to the scene
sceneView.autoenablesDefaultLighting = true
}
override func viewWillAppear(_ animated: Bool) {
super.viewWillAppear(animated)
// Create a session configuration to track an external image
let configuration = ARImageTrackingConfiguration()
// Image detection
// Reference which group to find the image to detect in the Assets folder e.g. "Detection Card"
if let imageDetect = ARReferenceImage.referenceImages(inGroupNamed: "Detection Card", bundle: Bundle.main) {
// Sets image tracking properties to the image in the referenced group
configuration.trackingImages = imageDetect
// Amount of images to be tracked
configuration.maximumNumberOfTrackedImages = 1
}
// Run the view's session
sceneView.session.run(configuration)
}
// Run when horizontal surface is detected and display 3D object onto image
// ARAnchor - tells a certain point in world space is relevant to your app, makes virtual content appear "attached" to some real-world point of interest
func renderer(_ renderer: SCNSceneRenderer, nodeFor anchor: ARAnchor) -> SCNNode {
// Creates 3D object
let obj = SCNNode()
// Check if image detected through camera is an ARImageAnchor - which contains position and orientation data about the image detected in the session
if let imageAnchor = anchor as? ARImageAnchor {
// Set dimensions of the horizontal plane to be displayed onto the image to be the same as the image uploaded
let plane = SCNPlane(width: imageAnchor.referenceImage.physicalSize.width, height: imageAnchor.referenceImage.physicalSize.height)
// Display mild transparent layer onto detected image
// This is to ensure image detection works by display a faint layer on the image
plane.firstMaterial?.diffuse.contents = UIColor(white: 1.0, alpha: 0.2)
// Set geometry shape of the plane
let planeNode = SCNNode(geometry: plane)
// Flip vertical plane to horizontal plane
planeNode.eulerAngles.x = -Float.pi / 2
obj.addChildNode(planeNode)
// Initialise chess scene
if let chessBoardSCN = SCNScene(named: "art.scnassets/chess.scn") {
// If there is a first in the scene file
if let chessNodes = chessBoardSCN.rootNode.childNodes.first {
// Displays chessboard upright
chessNodes.eulerAngles.x = Float.pi / 2
// Adds chessboard to the overall 3D scene
obj.addChildNode(chessNodes)
}
}
}
return obj
}
override func viewWillDisappear(_ animated: Bool) {
super.viewWillDisappear(animated)
// Pause the view's session
sceneView.session.pause()
}
}
You will need to add gestures on to your view and use the ARSceneViews hitTest method to detect what the gesture is touching in your scene. You can then update the positions based on the movement from the gestures.
Here is a question that deals with roughly the same requirement of dragging nodes around.
Placing, Dragging and Removing SCNNodes in ARKit
First, you need to add a gesture recognizer for tap into your viewDidLoad, like this:
let tapGesture = UITapGestureRecognizer(target: self, action: #selector(handleTap(_:)))
myScnView.addGestureRecognizer(tapGesture)
Then realize the handler function:
#objc
func handleTap(_ gestureRecognize: UIGestureRecognizer) {
// HERE YOU NEED TO DETECT THE TAP
// check what nodes are tapped
let location = gestureRecognize.location(in: myScnView)
let hitResults = myScnView.hitTest(location, options: [:])
// check that we clicked on at least one object
if hitResults.count > 0 {
// retrieved the first clicked object
let tappedPiece = hitResults[0].node
// HERE YOU CAN SHOW POSSIBLE MOVES
//Ex. showPossibleMoves(for: tappedPiece)
}
}
Now, to show the possible moves, you need to identify all quadrants and your node position on the chessboard.
To do this, you can assign a name or a number, or a combination of letter and number, or moreover a combination of numbers. (I suggest combination of number, like row 1 column 1, like a matrix).
let's take my suggestion, so you can name each quadrant 1.1 1.2 ... 2.1 2.2 and so on.
Now, to detect where your piece is, you can check contact with the PhysicsContactDelegate.
Now you have the tappedPiece and the place where it is, so you have to define the rule for the pieces, for example:
let rules = ["tower":"cross"] //add the others
N.B You can choose what you want to define the rules.
Let's take my suggestion for good, now you should create the function to highlight:
func highlight(quadrant: SCNNode){
quadrant.geometry?.firstMaterial?.emission.contents = UIColor.yellow
}
Finally the showPossibleMoves(for: tappedPiece) could be something this:
func showPossibleMoves(for piece: SCNNode){
let pieceType = piece.name //You have to give the name as you did into your rules variable
//ex. if you have rules like ["tower":"cross"] you have to set all towers name to "tower"
let rule = rules[pieceType]
switch rule{
case "cross":
//you have to highlight all nodes on the right, left, above and bottom
// you can achieve this by selecting the start point and increase it
//assuming you named your quadrants like 1.1 1.2 or 11 12 13 ecc...
let startRow = Int(startQuadrant.name.first)
let startColumn = Int(startQuadrant.name.last)
//Now loop the highlight on right
for column in startColumn+1...MAX_COLUMN-1{
let quadrant = myScnView.scene.rootNode.childNode(withName:"\(startRow).\(column)" , recursively: true)
// call highlight function
highlight(quadrant: quadrant)
}
//Now loop for above quadrants
for row in startRow+1...MAX_ROW-1{
let quadrant = myScnView.scene.rootNode.childNode(withName:"\(row).\(startColumn)" , recursively: true)
// call highlight function
highlight(quadrant: quadrant)
}
//DO THE SAME FOR ALL DIRECTIONS
}
// ADD ALL CASES, like bishop movements "diagonals" and so on
}
NOTE: In the handlerTap function you have to check what you're tapping, for example, to check if you're tapping on a quadrant after selecting a piece (you want to move you're piece) you can check a boolean value and the name of the selected node
//assuming you have set the boolean value after selecting a piece
if pieceSelected && node.name != "tower"{
//HERE YOU CAN MOVE YOUR PIECE
}

How can i have multiple select options for a single image?

I am using SceneKit to import a 3d image model of a human body. When i select a particular location point in the image, i want the app to recognize the body part and perform a different function for each part. How do i go about implementing this? What is the best way to do this?
P.s. when the image is rotated it shows a different view. I need the app to be able to recognize the body part even when it is rotated by the user. Any guidance as to how to proceed would be appreciated..
Here's a simple SceneKit picking example.
The scene is setup in the viewDidLoad, for your use case I'd expect a scene would be loaded from file (best done in another method). This file will hopefully have the different components you wish to pick as separate components in a tree-like hierarchy. The author of this 3D body model will have hopefully labelled these components appropriately so that your code can identify what to do when your left-femur is selected (and not comp2345).
For a complex model expect several 'hits' for any xy coordinate as you will be returned all nodes intersected by the hit ray. You may wish to only use the first hit.
import UIKit
import SceneKit
class ViewController: UIViewController {
#IBOutlet var scenekitView: SCNView!
override func viewDidLoad() {
super.viewDidLoad()
let scene = SCNScene()
let boxNode = SCNNode(geometry: SCNBox(width: 1, height: 1, length: 1, chamferRadius: 0))
boxNode.name = "box"
scene.rootNode.addChildNode(boxNode)
let sphereNode = SCNNode(geometry: SCNSphere(radius: 1))
sphereNode.name = "sphere"
sphereNode.position = SCNVector3Make(2, 0, 0)
boxNode.addChildNode(sphereNode)
let torusNode = SCNNode(geometry: SCNTorus(ringRadius: 1, pipeRadius: 0.3))
torusNode.name = "torus"
torusNode.position = SCNVector3Make(2, 0, 0)
sphereNode.addChildNode(torusNode)
scenekitView.scene = scene
scenekitView.autoenablesDefaultLighting = true
scenekitView.allowsCameraControl = true
}
override func touchesBegan(touches: Set<UITouch>, withEvent event: UIEvent?) {
//get the first touch location in screen coordinates
guard let touch = touches.first else {
return
}
//convert the screen coordinates to view coordinates as the SCNView make not take
//up the entire screen.
let pt = touch.locationInView(self.scenekitView)
//pass a ray from the points 2d coordinates into the scene, returning a list
//of objects it hits
let hits = self.scenekitView.hitTest(pt, options: nil)
for hit in hits {
//do something with each hit
print("touched ", hit.node.name!)
}
}
override func didReceiveMemoryWarning() {
super.didReceiveMemoryWarning()
}
}

Resources