Hello, i've recently received the entitlements to access the main camera stream for a project on the Apple Vision Pro. What happens : When executing code from this WWDC tutorial , i'm getting this error when trying to use a Camera Frame Provider : ar_camera_frame_provider_t <0x300d58870>: Failed to start camera stream with error: <ar_error_t: 0x303fcc4c0 Error Domain=com.apple.arkit Code=100 "App not authorized." UserInfo={NSLocalizedFailureReason=Using camera frame provider requires an entitlement., NSLocalizedRecoverySuggestion=, NSLocalizedDescription=App not authorized.} What I've tried : I followed the instructions given by mail, by : adding the .license file at the root of my project, adding the .entitlements file by adding capabilities in the project (Main Camera Access & Passthrough in screen capture are there). I've added NSCameraDescription, NSEnterpriseMCAMUsageDescription and NSWorldSensingUsageDescription (they all have a value assigned). I've also followed those post & post advices. When checking on the Account settings, i do see the capabilities in the "additional capabilities" On first launch, I'm also getting prompted to accept the NSEnterpriseMCAMUsageDescription, so I assume the info.plist file is valid? What did i missed to get the entitlements working ? Here's the code : import ARKit import SwiftUI import Vision import RealityKit class MainCameraAccess { var arKitSession = ARKitSession() var cameraFrameProvider = CameraFrameProvider() var pixelBuffer: CVPixelBuffer? func startCameraSession() async { let formats = CameraVideoFormat.supportedVideoFormats(for: .main, cameraPositions: [.left]) // Request authorization await arKitSession.requestAuthorization(for: [.cameraAccess]) // Start the session do { try await arKitSession.run([cameraFrameProvider]) } catch { print("Failed to start ARKit session: \(error)") return } // Get camera frame updates guard let cameraFrameUpdates = cameraFrameProvider.cameraFrameUpdates(for: formats[0]) else { return } // Process frames for await cameraFrame in cameraFrameUpdates { guard let mainCameraSample = cameraFrame.sample(for: .left) else { continue } self.pixelBuffer = mainCameraSample.pixelBuffer } } func saveLatestImage() { guard let pixelBuffer = self.pixelBuffer else { print("No image available to save.") return } // Convert CVPixelBuffer to UIImage let ciImage = CIImage(cvPixelBuffer: pixelBuffer) let context = CIContext() guard let cgImage = context.createCGImage(ciImage, from: ciImage.extent) else { print("Failed to create CGImage.") return } let uiImage = UIImage(cgImage: cgImage) // Save UIImage to Photos Album UIImageWriteToSavedPhotosAlbum(uiImage, nil, nil, nil) print("Image saved to photo library.") } } Thanks in advance for the help, Jeremy

Hi, I would like to know how predicted hand tracking in ARKit works. Does it use the Kalman filter, or does it have another approach?

Hello, I am currently working on an app that features multiple environments in which I combine Reality Composer Pro scenes with objects managed at runtime as well as make heavy use of RealityView attachments that modify the appearance of certain objects. Is it possible to keep track of an AR anchor when transitioning between immersive spaces? About my app: There are two main contexts/scenes in the app that the user progresses through. The first takes place in AR and is non-interactive and driven by a timeline animation. The second is in VR and allows the user to change materials of select models. Both scenes need to be placed relative to a real-life object that functions as an image anchor. Anchoring is necessary for visual purposes in AR context and it would be nice to use it in the VR context as well in order to provide passive haptics to the user. If the user doesn't have access to the physical object, we make use of plane-based anchoring. Either way, we would like to keep the anchor's position across the scenes.

Hi all, Our app allows a user to scan a room and then save that scan on a separate view, followed by additional scans. We're looking into allowing room combining via CapturedStructure, so we need rooms to be scanned in the same ARWorldMap without necessarily needing to re-localize in the same session. This should fit within the first scenario that Apple described. The only way I have found that allows our requirements is to save RoomCaptureView and to re-use that RoomCaptureView whenever we need to start a session again. This creates a number of other issues, and ideally, we wouldn't need to save a View in something like a singleton. We are using captureSession.stop(pauseARSession: false). Additionally, if we use the same RoomCaptureView and an error occurs during the scanning process, we can't get the instructions overlay to appear again if we reuse this view (specifically, the instructions in the middle of the view that state "Move device to start"). It's as if the instructions are completely removed and scanning is stuck on an error state if an error occurs. These instructions also seem to be separate from the instructions we can grab from RoomCaptureViewDelegate via didProvide instruction: RoomCaptureSession.Instruction), so we can't use that either. There's a couple subviews that seem relevant to this: RoomCaptureCoachingOverlayView and ARGlyphView - but both are not public, so we can't force them to appear. Also attempted a number of other things to try to get these subviews to appear, such as layoutIfNeeded(). Saving the ARSession and using it in let roomCaptureView = RoomCaptureView(frame: viewBounds, arSession: arSession) where we're creating a new view with the same ARSession seems much more ideal as that solves the above issues, but we run into another issue: world tracking seems to be completely lost when a new RoomCaptureView (and thus a new RoomCaptureSession) is started, even with the same already started ARSession, almost as if captureSession.stop(pauseARSession: false) doesn't work as described. Is there any way around needing to use the same RoomCaptureView or RoomCaptureSession for subsequent scans in the same session without needing to re-localize via ARWorldMap loading? Is there a way to force the guiding instructions to appear?

WWDC21 had a cool demo project with fish, with a watery, misty look (Dive into RealityKit). It used post processing in RealityKit, but the ARView class isn’t available in VisionOS. Can CompositorLayer be used instead for post processing in full immersion?

Ever since updating to Xcode 16 my AR app doesn't compile, because Xcode doesn't recognize the .rcproject files used to load the AR experiences in iOS app. The .rcproject files were authored in Reality Composer on iPadOS. The expected behavior is described in this official Apple documentation article: https://developer.apple.com/documentation/realitykit/loading-entities-from-a-file How do I submit a ticket to Apple?

==> Which information will I get from CMSSampleBuffer ? Is there an option to block close up accomodation of the camera ? Is there a way for the object capture module to take a video instead of a series of picture ? It would be fantastic to have an answer on all of these questions to be able to move forward on new implementations.

We are currently using ObjectCapture from ARKit, and we would like to fix exposure time, white balance parameter and ISO. How can we do this ? Additionally, we'd like to obtain the following information from the ARKit : white balance parameters (in case we cannot fix them) and color correction matrices ?

Is it possible to access the raw lidar measurements before the sceneDepth calculation is done to combines the lidar measurements with visual data. In low light environments the lidar scanner should still work and provide depth info but I cannot figure out how to access those pure lidar depth measurements. I am currently using: guard let frame = arView.session.currentFrame, let depthData = frame.sceneDepth?.depthMap else { print("Depth data is unavailable.") return } but this is the depth data after sensor fusion occurs and fails in low light conditions.

In visionOS beta, when using ARKit for image detection, the initially detected AnchorUpdate status is .add, and subsequent detections of the same image are marked as .update. However, after toggling immersiveSpace, the same image is detected with the status .add again. After updating to visionOS 2.1, the first detection status remains `add, and subsequent detections of the same image remain .update, even after toggling immersiveSpace. Could this be due to a change in processing flow?

Hi all, I am having trouble debugging an error where the wireframe object entity representation for the Object Tracking Demo: "Explore object tracking for visionOS" appears incorrect in the right eye of the Vision Pro but correct in the left eye. Would anyone happen to know what is going on? I have attempted to offset the object by changing world coordinates, but this moves the object in both the left and the right eye. Could this be due to the new visionOS beta update (2.0 --> 2.2) ? I am currently using visionOS 2.2. Thanks!

By applying for the enterprise API, we can obtain the data of video frames collected by VisionPro glasses, and then we process the collected video frames to achieve the function of eliminating a certain object. But it was not found how to insert the processed video frames into the data source collected by the system camera. So I would like to ask if there is any API that can insert processed video frames into the original data and present them to the user? This effect is similar to the right side twist of VisionPro glasses, which allows the physical world and digital space to blend perfectly after rotation. So, I would like to ask if there is a related API that can solve this problem? STEPS TO REPRODUCE Obtain video frames, Process the obtained video frames Insert the processed video frames into the VisonOS system camera. System: VisionOS 2.0 API used: Enterprise APIs Main camera access permissions

How to find main (left) camera transform from world anchor? (Enterprise API) From CameraFrameProvider() I can get a frame sample which has an "extrinsics" parameter. How is it defined? Relative to what point/anchor?

Hi! I wanna know that if it's possible that loading Immersive Scene after scanning(recognizing) preregistered images or objects? I tried to load the Immersive scene after scanning image and objects, it didn't work well. Please let me know about the solution if it's possible. Here the ImmersiveView.swift code i tried. // ImmersiveView.swift import SwiftUI import RealityKit import RealityKitContent // Using the RealityKitContent module struct ImmersiveView: View { @ObservedObject var viewModel: TrackingViewModel @State private var immersiveScene: Entity? @State private var isToggleOn: Bool = false // Variable for toggle state var body: some View { ZStack { // Overlay RealityView and UI elements RealityView { content in if let scene = immersiveScene { content.add(scene) print("Immersive scene successfully added.") if let moneyGunsEntity = scene.findEntity(named: "MoneyGuns") { NotificationCenter.default.post( name: Notification.Name("RealityKit.NotificationTrigger"), object: nil, userInfo: [ "RealityKit.NotificationTrigger.Scene": scene, "RealityKit.NotificationTrigger.Identifier": "PlayTimeline" ] ) print("PlayTimeline notification sent.") } else { print("MoneyGuns entity not found.") } } } .onAppear { Task { if let scene = try? await Entity(named: "Immersive", in: realityKitContentBundle) { immersiveScene = scene } else { print("Failed to load immersive scene.") } } } VStack { Spacer() Toggle(isOn: $isToggleOn) { // Add toggle button Text("Toggle Option") .foregroundColor(.white) } .padding() .background(Color.black.opacity(0.7)) .cornerRadius(8) .padding() } } } }

The method of taking screenshots in IOS can be done through the "view. layer. render (in: UIGraphicsGetCurrentContext()!)" method. What should be replaced with "view. layer" in VisionOS to call the ". render (in: UIGraphicsGetCurrentContext()!)" method??

In visionOS, ARKit is to integrate virtual and reality. However, most of the functions RealityKit can be easily implemented (except for Scene reconstruction, Room Tracking and enterprise API), so do I still need to use ARKit? Is there any difference between them?

Hi! I'm making content using Room Tracking for vision pro these days. So I searched information about it. Here the links I visited. But I could not found the info I wanted to know Apple ARKit Create enhanced spatial computing experiences with ARKit RoomTrackingProvider I wanna know that if it's possible remembering room structure that recognized before and adding contents in certain world anchor in the room space when user entered the room again? For example, a developer can save the room structure, room info (with room ID) and world anchor of the room with Room Tracking feature. After this, the developer can add entities via Xcode and Reality Composer Pro in certain position of the room to show contents to users when users enter the room. So users can see the contents whenever they visit the room. Is this possible? If there are example codes or projects about it, please let me know.

I have a VideoMaterial inside a RealityView and want to attach this to a DockingRegion inside an immersive environment. It appears that adding the VideoMaterial entity as a child of the docking region somewhat works, but there are no lighting effects (specular, diffuse) from the playing video. So essentially, how can you add a VideoMaterial to a DockingRegion and achieve the same reflections/behavior as using AVPlayerViewController. The latter is not an option as I need custom controls.

Hi, I'm currently working on an ARKit project where I need to implement object occlusion on devices that do not have a LiDAR sensor (e.g., iPhone XR, iPhone 11). I used CoreML models like DepthAnythingV2 to create depth maps and DETRResnet50SemanticSegmentationF16P8 to to perform real-time segmentation. But these models are too heavy for devices. Much appreciated on any advice or pointers to resources.

I am trying to create reflections and lighting in a fully virtual scene, and the way the documentation is written, the VirtualEnvironmentProbeComponent would be key. I cannot seem to get it to work, however. Has anyone got this figured out?