My VisionOS App (Travel Immersive) has two interface windows: a main 2D interface window and a 3D Earth window. If the user first closes the main interface window and then the Earth window, clicking the app icon again will only launch the Earth window while failing to display the main interface window. However, if the user closes the Earth window first and then the main interface window, the app restarts normally‌. Below is the code of import SwiftUI @main struct Travel_ImmersiveApp: App { @StateObject private var appModel = AppModel() var body: some Scene { WindowGroup(id: "MainWindow") { ContentView() .environmentObject(appModel) .onDisappear { appModel.closeEarthWindow = true } } .windowStyle(.automatic) .defaultSize(width: 1280, height: 825) WindowGroup(id: "Earth") { if !appModel.closeEarthWindow { Globe3DView() .environmentObject(appModel) .onDisappear { appModel.isGlobeWindowOpen = false } } else { EmptyView() // 关闭时渲染空视图 } } .windowStyle(.volumetric) .defaultSize(width: 0.8, height: 0.8, depth: 0.8, in: .meters) ImmersiveSpace(id: "ImmersiveView") { ImmersiveView() .environmentObject(appModel) } } }

In visionOS, there are existing modifiers that can completely conceal the hands. However, I am interested in learning how to achieve the effect of only one hand disappearing while the other hand remains visible. .upperLimbVisibility(.hidden)

I have a MeshResource and I would like to create a collision component from it. let childBounds = child.visualBounds(relativeTo: self) var childShape: ShapeResource do { // Crashed by the following line instead of throwing a Swift Error childShape = try await ShapeResource.generateConvex(from: childModel.mesh); } catch { childShape = ShapeResource.generateBox(size: childBounds.extents) childShape = childShape.offsetBy(translation: childBounds.center) } Based on this document: https://developer.apple.com/documentation/realitykit/shaperesource/generateconvex(from:)-6upj9 Will throw an error if mesh does not define a nonempty convex volume. For example, will fail if all the vertices in mesh are coplanar. But, the method crashes the app instead of throwing a Swift error Incident Identifier: 35CD58F8-FFE3-48EA-85D3-6D241D8B0B4C CrashReporter Key: FE6790CA-6481-BEFD-CB26-F4E27652BEAE Hardware Model: Mac15,11 ... Version: 1.0 (1) Code Type: ARM-64 (Native) Role: Foreground Parent Process: launchd_sim [2057] Coalition: com.apple.CoreSimulator.SimDevice.85A2B8FA-689F-4237-B4E8-DDB93460F7F6 [1496] Responsible Process: SimulatorTrampoline [910] Date/Time: 2025-01-26 16:13:17.5053 +0800 Launch Time: 2025-01-26 16:13:09.5755 +0800 OS Version: macOS 15.2 (24C101) Release Type: User Report Version: 104 Exception Type: EXC_BREAKPOINT (SIGTRAP) Exception Codes: 0x0000000000000001, 0x00000001abf841d0 Termination Reason: SIGNAL 5 Trace/BPT trap: 5 Terminating Process: exc handler [17316] Triggered by Thread: 0 Thread 0 Crashed: 0 CoreRE 0x1abf841d0 REAssetManagerCollisionShapeAssetCreateConvexPolyhedron + 232 1 CoreRE 0x1abf845f0 REAssetManagerCollisionShapeAssetCreateConvexPolyhedronFromMesh + 868 2 RealityFoundation 0x1d25613bc static ShapeResource.generateConvex(from:) + 148 Here is the message on the app console from Xcode /Library/Caches/com.apple.xbs/Sources/REKit_Sim/ThirdParty/PhysX/physx/source/physxcooking/src/convex/QuickHullConvexHullLib.cpp (935) : internal error : QuickHullConvexHullLib::findSimplex: Simplex input points appers to be coplanar. Failed to cook convex mesh (0x3) assertion failure: 'convexPolyhedronShape != nullptr' (REAssetManagerCollisionShapeAssetCreateConvexPolyhedron:line 356) Bad parameters passed for convex mesh creation. Message from debug The above crash happened on a visionOS simulator (visionOS 2.2 (22N840)

I can generate a ShapeResource from a ReakityKit entity's extents. Could I apply some scaling to the generated shape. Is there a way to do that? // model is a ModelResource and bounds is a BoundingBox var shape = ShapeResource.generateConvex(from: model.mesh); shape = shape.offsetBy(translation: bounds.center) // How can I scale the shape to fit within the bounds? The following API only provide the rotation and translation support. and I cannot find the scale support. offsetBy(rotation: simd_quatf = simd_quatf(ix: 0, iy: 0, iz: 0, r: 1), translation: SIMD3<Float> = SIMD3<Float>()) I can put the ShapeResource on an entity and scale the entity. But, I would like to know if it is possible to scale the ShapeResource itself without attaching it to an entity.

I am develop visionOS app. I am now very interested in Metal and Compositor Services, but I have not explored them in depth. I know that Metal has a higher degree of control freedom. I am wondering if using Compositor Services will have fewer functions than RealityKit in AR technology (such as scene reconstruction and understanding, hover effect, etc.).

Hello! I’m excited to see that Look to Scroll has been included in visionOS 26 Beta. I’m aiming to achieve a feature where the user’s gaze at a specific edge automatically scrolls to that position. However, I’ve experimented with ScrollView and haven’t been able to trigger this functionality. Could you advise if additional API modifiers are necessary? Thank you!

Hello again! So, I am trying my best to figure out how to get a simple, fully Immersive scene that is navigable via the PS5 controller to work on Vision Pro. I am even using ChatGPT (both externally AND in Xcode when it makes fix recommendations) however, I seem to be moving around in circles, having to constantly send error messages back to ChatGPT, then try the new "confirmed to work" fixed code, only to have more errors. I also have explicitly let the AI know that I'm using all the current betas, only to learn there has been lots of stuff deprecated (of course!). This is my initial query to ChatGPT: "How can I create a fully immersive VR experience for Apple Vision Pro, using a PS5 game controller to navigate the user via first person view through a 3d scene?" And then down the rabbit hole the past few days. Seriously, I thought AI was supposed to be helpful! LOL Per Xcode logs, I'm seeing thumbstick movement, but simply cannot get the camera to move. Is this even possible?! Is it a camera defined in Reality Composer Pro? Or, does it need to be defined outside. I'm trying to follow along with the AI, but it seems to go back and forth with different solutions. Here is the latest it's spitting out: __ Thanks for this super clear log — and you’re absolutely right: the gamepad input is working, CameraRoot is moving as expected… but your actual camera view is not changing. This strongly suggests the issue is that RealityKit is not using your custom camera (PerspectiveCameraComponent) for rendering the view. ⸻ 🔍 What’s happening Even though CameraRoot is moving (with correct logs), RealityKit will default to its own internal camera unless: 1. You explicitly assign your camera entity to be the one used, or 2. Your camera is the only one and RealityKit picks it up automatically. But in visionOS 2.0 / Xcode 26, there seems to be a bug or behavior change where RealityKit no longer auto-uses your custom PerspectiveCameraComponent in RealityView unless it’s: • properly set with content.camera, or • part of a .usda/Reality Composer camera hierarchy that’s already active. Unfortunately: 🟥 RealityView.camera = .entity(cameraRoot) ❌ Is no longer allowed (we confirmed this breaks with a type error) ⸻ ✅ The Reliable Workaround: Attach your camera to the Immersive scene itself Since we’re loading a .usda scene (Reality Composer Pro), here’s the most reliable path forward: ✔️ Option A: Embed your camera into the Reality Composer Pro scene 1. Open your Reality Composer Pro scene 2. Add a Camera entity directly in your Immersive.usda scene 3. Name it something like "PlayerCamera" 4. Position it at [0, 1.6, 0] 5. Leave it selected as the default camera Then in your Swift code: • Find the camera entity with .findEntity(named:) • Move that entity instead of creating a new one. ⸻ ✔️ Option B: Force RealityKit to use your custom camera (via transform parenting) Since direct camera assignment is unavailable, we can use a trick: Move the whole RealityView scene via a parent entity with a camera. But let’s go with Option A, since you’ve already got a Reality Composer Pro scene. __ Can any developer over there can make a suggestion on how to move forward? I JUST want to create a sandbox template so I can start exploring 3d scenes created in Maya and sent over to headset. I'll deal with animation next, but come on, why is this so difficult to get working? I'm not a programmer, but have been trying to wrap my head around Xcode and SwiftUI. This needs to be much simpler. Or, you need to provide us creatives with better sample templates and non programmer speak on how to set this up properly. Ideally, you HIRE us 3d professionals to work side by side with the programmers to help make these tools useable - especially Reality Composer Pro. Seriously, I am making a concerted effort to use the native tools, even though I would love to be porting Unreal Engine scenes over. If anyone can help point me in the right direction, coming from a 3D Creator/Animator/Modeler perspective, I, and my fellow peers in the XR/AR/VR community would greatly appreciate it. Thank you.

Hi, are we allowed to push the default support in Package.swift up to iOS 18 to allow for the latest APIs? And with the terms of the competition, can we use stock 3D USDZ assets? Thank you!

Hi, I'm developing an app for the Apple Vision Pro. Inside the app the user should be able to load objects from the web into the scene and then be able to move them around (dragging and rotating) via gestures. My question: I'm working with RealityKit and use RealityView.. I have no issues loading in one object and making it interactive by adding gestures to the entire RealityView via the .gestures() function. Also I succeeded in loading multiple objects into the scene. My problem is that I can't figure out how to add my gestures to multiple objects independently. I can't use Reality composer since I'm loading the objects dynamically into the scene. Using .gestures() doesn't work for multiple objects since every gesture needs to be targeted to a specific entity but I have multiple entities. I also tried defining a GestureComponent and adding it to every newly loaded entity but that doesn't seem to work as nothing happens, even though my gesture is targeted to every entity having my GestureComponent. I only found solutions that are not usable on Visionos / RealityView like installGestures. Also I tried following this guide: https://developer.apple.com/documentation/realitykit/transforming-realitykit-entities-with-gestures But I feel like there are things missing and contradictory, like a extension of RealityView is mentioned but not shown and the guide states that we don't need to store the translation values for every entity and therefore creates a EntityGestureState.swift file to store these values but then it's not used and a GestureStateComponent is used instead that was never mentioned and contradicts what was just said about not storing values per entity because a new instance of it is created for every entity. Nevermind, following the guide didn't lead me to a working solution.

I want to display a model image in the windowGroup window. This image is not unique. To display the model image, how should I convert it into an image

Previously I had developed software using SMI eye trackers, both screen mounted and their mobile glasses, for unique therapeutic and physiology applications. Sadly, after SMI was bought by Apple, their hardware and software have been taken off the market and now it is very difficult to get secondhand-market systems. The Apple Vision Pro integrates the SMI hardware. While I can use ARKit to get gaze position, I do not see a way to access information that was previously made accessible on the SMI hardware, particularly: dwell time and pupil diameter information. I am hopeful (or asking) to see that if a user has a properly set up Optic ID and would opt-in if, either on the present or a future version of visionOS, it might be possible to get access to the data streams for dwell times and pupil diameter. Pupil diameter is particularly important as it is a very good physiological measure of how much stress a person is encountering, which is critical to some of the therapeutic applications that formerly we used SMI hardware. Any ideas, or, if this is not possible, proposing this to the visionOS team would be appreciated!

I have a huge sphere where the camera stays inside the sphere and turn on front face culling on my ShaderGraphMaterial applied on that sphere, so that I can place other 3D stuff inside. However when it comes to attachment, the object occlusion never works as I am expecting. Specifically my attachments are occluded by my sphere (some are not so the behavior is not deterministic. Then I suspect it was the issue of depth testing so I started using ModelSortGroup to reorder the rending sequence. However it doesn't work. As I was searching through the internet, this post's comments shows that ModelSortGroup simply doesn't work on attachments. So I wonder how should I tackle this issue now? To let my attachments appear inside my sphere. OS/Sys: VisionOS 2.3/XCode 16.3

As in the title: openImmersiveSpace works as expected. The ImmersiveSpace in the ID opens normally but the function never resolves a Result. Here is a workaround I used to make user it was on the UI thread and scenePhase was active: ` @MainActor func openSpaceWithStateCheck() async { if scenePhase == .active { Task { switch await openImmersiveSpace(id: "RoomCaptureInteraction") { case .opened: isCapturingImagery = true break case .error: print("!! An error occurred when trying to open the immersive space captureRoomImagery") case .userCancelled: print("!! The user declined opening immersive space captureRoomImagery") @unknown default: print("!! unknown default result of opening space") break } } } else { print("Scene not active, deferring immersive space opening") } } I'm on visionOS 2.4 and SDK 2.2. I have tried uninstalling the app and rebuilding. Tried simply opening an empty ImmersiveSpace. The consistency of the ImmersiveSpace opening at least means I can work around it. Even dismissImmersiveSpace works normally and closes the immersive space. But a workaround seems hamfisted.

Hello, I'm working with the new PortalComponent introduced in visionOS 2.0, and I've encountered some issues when transitioning entities between virtual and real-world spaces using crossingMode. Specifically: Lighting inconsistency: When CG content (ModelEntities with PhysicallyBasedMaterial) crosses the portal from virtual space into the real environment, the way light reflects on the objects changes noticeably. This causes a jarring visual effect, as the same material appears differently depending on the space it's in. Unnatural transition visuals: During the transition, the CG models often appear to "emerge from the wall," especially when crossing from virtual to real. This ruins the immersive illusion and feels visually unnatural. IBL adjustment attempts: I’ve tried adding an ImageBasedLightComponent to the world entity, and while it slightly improves the lighting consistency, the issue still remains to a noticeable degree. My goal is to create a seamless visual experience when CG entities cross between spaces, without sudden lighting shifts or immersion-breaking geometry reveals. Has anyone else experienced similar issues? Is there a recommended setup or workaround to better control lighting and visual fidelity when using crossingMode with portals in visionOS 2.0? Any guidance would be greatly appreciated. Thank you!

One of the most common ways to provide a window size in visionOS is to use the defaultSize scene modifier. WindowGroup(id: "someID") { SomeView() } .defaultSize(CGSize(width: 600, height: 600)) Starting in visionOS 26, using this has a side effect. visionOS 26 will restore windows that have been locked in place or snapped to surfaces. If a user has manually adjusted the size of a locked/snapped window, the users size is only restore in some cases. Manual resize respected Leaving a room and returning later Taking the headset off and putting it back on later Manual resize NOT respected Device restart. In this case, the window is reopened where it was locked, but the size is set back to the values passed to defaultSize. The manual resizing adjustments the user has made are lost. This is counter to how all other windows and widgets work. I reported this last month (FB18429638), but haven't heard back if this is a bug or intended behavior. Questions What is the best way to provide a default window size that will only be used when opening new windows–and not used during scene restoration? Should we try to keep track of window size after users adjust them and save that somewhere? If this is intended behavior, can someone please update the docs accordingly?

I am currently developing an app for visionOS and have encountered an issue involving a component and system that moves an entity up and down within a specific Y-axis range. The system works as expected until I introduce sound playback using AVAudioPlayer. Whenever I use AVAudioPlayer to play sound, the entity exhibits unexpected behaviors, such as freezing or becoming unresponsive. The freezing of the entity's movement is particularly noticeable when playing the audio for the first time. After that, it becomes less noticeable, but you can still feel it, especially when the audio is played in quick succession. Also, the issue is more noticable on real device than the simulator // // IssueApp.swift // Issue // // Created by Zhendong Chen on 2/1/25. // import SwiftUI @main struct IssueApp: App { var body: some Scene { WindowGroup { ContentView() } .windowStyle(.volumetric) } } // // ContentView.swift // Issue // // Created by Zhendong Chen on 2/1/25. // import SwiftUI import RealityKit import RealityKitContent struct ContentView: View { @State var enlarge = false var body: some View { RealityView { content, attachments in // Add the initial RealityKit content if let scene = try? await Entity(named: "Scene", in: realityKitContentBundle) { if let sphere = scene.findEntity(named: "Sphere") { sphere.components.set(UpAndDownComponent(speed: 0.03, minY: -0.05, maxY: 0.05)) } if let button = attachments.entity(for: "Button") { button.position.y -= 0.3 scene.addChild(button) } content.add(scene) } } attachments: { Attachment(id: "Button") { VStack { Button { SoundManager.instance.playSound(filePath: "apple_en") } label: { Text("Play audio") } .animation(.none, value: 0) .fontWeight(.semibold) } .padding() .glassBackgroundEffect() } } .onAppear { UpAndDownSystem.registerSystem() } } } // // SoundManager.swift // LinguaBubble // // Created by Zhendong Chen on 1/14/25. // import Foundation import AVFoundation class SoundManager { static let instance = SoundManager() private var audioPlayer: AVAudioPlayer? func playSound(filePath: String) { guard let url = Bundle.main.url(forResource: filePath, withExtension: ".mp3") else { return } do { audioPlayer = try AVAudioPlayer(contentsOf: url) audioPlayer?.play() } catch let error { print("Error playing sound. \(error.localizedDescription)") } } } // // UpAndDownComponent+System.swift // Issue // // Created by Zhendong Chen on 2/1/25. // import RealityKit struct UpAndDownComponent: Component { var speed: Float var axis: SIMD3<Float> var minY: Float var maxY: Float var direction: Float = 1.0 // 1 for up, -1 for down var initialY: Float? init(speed: Float = 1.0, axis: SIMD3<Float> = [0, 1, 0], minY: Float = 0.0, maxY: Float = 1.0) { self.speed = speed self.axis = axis self.minY = minY self.maxY = maxY } } struct UpAndDownSystem: System { static let query = EntityQuery(where: .has(UpAndDownComponent.self)) init(scene: RealityKit.Scene) {} func update(context: SceneUpdateContext) { let deltaTime = Float(context.deltaTime) // Time between frames for entity in context.entities(matching: Self.query, updatingSystemWhen: .rendering) { guard var component: UpAndDownComponent = entity.components[UpAndDownComponent.self] else { continue } // Ensure we have the initial Y value set if component.initialY == nil { component.initialY = entity.transform.translation.y } // Calculate the current position let currentY = entity.transform.translation.y // Move the entity up or down let newY = currentY + (component.speed * component.direction * deltaTime) // If the entity moves out of the allowed range, reverse the direction if newY >= component.initialY! + component.maxY { component.direction = -1.0 // Move down } else if newY <= component.initialY! + component.minY { component.direction = 1.0 // Move up } // Apply the new position entity.transform.translation = SIMD3<Float>(entity.transform.translation.x, newY, entity.transform.translation.z) // Update the component with the new direction entity.components[UpAndDownComponent.self] = component } } } Could someone help me with this?

Hi there. Thanks to amazing help from you guys, I've managed to code a 360 image carousel, where the user can browse 360 images located inside the project package. Is there a way to access the filesystem on AVP outside the app? I know about the FileManager, and I can get access to the .documentsDirectory, but how do I access documents folder from the "Files" app on the AVP? My goal is to read images from a hardcoded folderlocation on the AVP, such that the user never will have to select the images themselves. I know this may not be the "right" way to do this. The app is supposed to be "foolproof" with a minimum of userinteraction. The only way to change the images should be to change the contents of the hardcoded imagefolder. I hope this makes sense =) Thanks in advance! Regards, Kim

I am trying to get the new PresentationComponent working in VisionOS26 as seen in this WWDC video: https://developer.apple.com/videos/play/wwdc2025/274/?time=962 (18:29 minutes into video) Here is some other example code but it doesn't work either: https://stepinto.vision/devlogs/project-graveyard-devlog-002/ My simple Text view (that I am adding as a PresentationComponent) does not appear in my RealityView even though the entity is found. Here is a simple example built from an Xcode immersive view default project: struct ImmersiveView: View { @Environment(AppModel.self) var appModel var body: some View { RealityView { content in // Add the initial RealityKit content if let immersiveContentEntity = try? await Entity(named: "Immersive", in: realityKitContentBundle) { content.add(immersiveContentEntity) if let materializedImmersiveContentEntity = try? await Entity(named: "Test", in: realityKitContentBundle) { content.add(materializedImmersiveContentEntity) var presentation = PresentationComponent( configuration: .popover(arrowEdge: .bottom), content: Text("Hello, World!") .foregroundColor(.red) ) presentation.isPresented = true materializedImmersiveContentEntity.components.set(presentation) } } } } } Here is the Apple reference: https://developer.apple.com/documentation/realitykit/presentationcomponent

HoverEffectComponent on macOS 15 and iOS 18 works fine using RealityView, but seems to be ignored when ARView (even with a SwiftUI UIViewRepresentable) is used. Feedback ID: FB15080805

I am using HelloPhotogrammetry in Xcode I can make one model with something like HelloPhotogrammetry.main([path_to_folder_of images, path_to_output/model.usdz, "-d", "medium", "-o", "unordered", "-f", "high" ]) But how would I request several models simultaneously? I only want to vary the detail. [ ("/Users/you/Desktop/model_medium.usdz", detail: .medium), ("/Users/you/Desktop/model_full.usdz", detail: .full), ("/Users/you/Desktop/model_raw.usdz", detail: .raw ]