Hi all, I'm currently developing a real-time object reconstruction app using ARKit. The goal is to scan large objects using ARKit’s depth and transform data, and generate a point cloud. However, I’m facing a major challenge - Transform Drift / World Alignment Issues The localToWorld transform provided by ARKit frequently seems to drift or become unstable across frames. This results in misaligned point clouds even when the device is moved slowly or kept relatively still. In some cases, a static surface scanned over a few seconds results in clearly misaligned fragments. This makes it difficult to accurately stitch a multi-frame point cloud. I have experimented with various lighting conditions and object textures, but the issue persists in all cases. At times, the relative error between frames reaches up to 20 cm, while in other instances the error is minimal; however, the drift gradually accumulates over time, leading to an overall enlargement of the reconstructed object. I have attached images of both cases here. Questions: Are there specific conditions under which ARKit’s world transform is expected to drift? Is there a way to detect or recover from this drift during runtime? Any best practices for maintaining consistent tracking during scanning or measurement sessions?

Hi! I'm currently experimenting on Apple Vision Pro with hand and head anchors. Is there a way to get an anchor linked to the apple magic keyboard (as the detection is already done to display inputs at the top)? Thanks in advance, Have a good day!

Using the example code posted here: https://developer.apple.com/documentation/visionOS/tracking-images-in-3d-space I can register multiple ReferenceImage s with a ImageTrackingProvider, but only one updates at a time - to have realtime updating, I can only have one ImageAnchor in my field of view at a time. Is it possible to track multiple imageAnchors at the same time in the same field of view? As in having several ImageAnchor's tracked and entities updated to the transforms of the anchor in the same frame/moment from the Apple Vision Pro?

I have a problem with the wall plane detection using visionOS/ARKit: I am using ARKitSession's PlaneDetectionProvider detection.wall in the space of visionOS. I recorded the position and rotation information of the first detected plane, but found that the rotation value will be facing when the user starts the space. There is a deviation in different directions. That is to say, even if the plane is located on the same wall, the rotation quaternion will be different. I hope that no matter from which direction the user enters the scan, the real direction of the wall can be correctly obtained so that the virtual content can be accurately aligned with the wall. I have tried to use anchor.originFromAnchorTransform or Transform.rotation directly, but the rotation value is still affected by the user's initial orientation. In addition, I would like to know whether the user's initial orientation will affect the location information. If so, please provide a solution. Thank you!

I thought the ARCoachingOverlayView was a nice touch, so each apps ARKit coaching was recognizable and I used it in my ARView/ARSCNView based apps. Now with RealityView, is there any replacement planned? Or should we just use UIViewRepresentable and wrap ARCoachingOverlayView?

Recently, questions about ARKit/visionOS seem to be being asked in the Apple forum by internal Apple engineers. Inexperienced and untested makeshift features are being offered, putting average but experienced developers in a difficult position. They are unable to react and get something useful from the posts. Apple needs to review the situation.

In my Vision OS app I am using plane detection and I want to create planes that have physics I want to create an effect that my reality kit entities rest on real world detected planes. I was curious to see that the code below that I found in the Samples is the most efficient way of doing this. func processPlaneDetectionUpdates() async { for await anchorUpdate in planeTracking.anchorUpdates { let anchor = anchorUpdate.anchor if anchorUpdate.event == .removed { planeAnchors.removeValue(forKey: anchor.id) if let entity = planeEntities.removeValue(forKey: anchor.id) { entity.removeFromParent() } return } planeAnchors[anchor.id] = anchor let entity = Entity() entity.name = "Plane \(anchor.id)" entity.setTransformMatrix(anchor.originFromAnchorTransform, relativeTo: nil) // Generate a mesh for the plane (for occlusion). var meshResource: MeshResource? = nil do { let contents = MeshResource.Contents(planeGeometry: anchor.geometry) meshResource = try MeshResource.generate(from: contents) } catch { print("Failed to create a mesh resource for a plane anchor: \(error).") return } var material = UnlitMaterial(color: .red) material.blending = .transparent(opacity: .init(floatLiteral: 0)) if let meshResource { // Make this plane occlude virtual objects behind it. entity.components.set(ModelComponent(mesh: meshResource, materials: [material])) } // Generate a collision shape for the plane (for object placement and physics). var shape: ShapeResource? = nil do { let vertices = anchor.geometry.meshVertices.asSIMD3(ofType: Float.self) shape = try await ShapeResource.generateStaticMesh(positions: vertices, faceIndices: anchor.geometry.meshFaces.asUInt16Array()) } catch { print("Failed to create a static mesh for a plane anchor: \(error).") return } if let shape { entity.components.set(CollisionComponent(shapes: [shape], isStatic: true)) let physics = PhysicsBodyComponent(mode: .static) entity.components.set(physics) } let existingEntity = planeEntities[anchor.id] planeEntities[anchor.id] = entity contentEntity.addChild(entity) existingEntity?.removeFromParent() } } } extension MeshResource.Contents { init(planeGeometry: PlaneAnchor.Geometry) { self.init() self.instances = [MeshResource.Instance(id: "main", model: "model")] var part = MeshResource.Part(id: "part", materialIndex: 0) part.positions = MeshBuffers.Positions(planeGeometry.meshVertices.asSIMD3(ofType: Float.self)) part.triangleIndices = MeshBuffer(planeGeometry.meshFaces.asUInt32Array()) self.models = [MeshResource.Model(id: "model", parts: [part])] } } extension GeometrySource { func asArray<T>(ofType: T.Type) -> [T] { assert(MemoryLayout<T>.stride == stride, "Invalid stride \(MemoryLayout<T>.stride); expected \(stride)") return (0..<count).map { buffer.contents().advanced(by: offset + stride * Int($0)).assumingMemoryBound(to: T.self).pointee } } func asSIMD3<T>(ofType: T.Type) -> [SIMD3<T>] { asArray(ofType: (T, T, T).self).map { .init($0.0, $0.1, $0.2) } } subscript(_ index: Int32) -> (Float, Float, Float) { precondition(format == .float3, "This subscript operator can only be used on GeometrySource instances with format .float3") return buffer.contents().advanced(by: offset + (stride * Int(index))).assumingMemoryBound(to: (Float, Float, Float).self).pointee } } extension GeometryElement { subscript(_ index: Int) -> [Int32] { precondition(bytesPerIndex == MemoryLayout<Int32>.size, """ This subscript operator can only be used on GeometryElement instances with bytesPerIndex == \(MemoryLayout<Int32>.size). This GeometryElement has bytesPerIndex == \(bytesPerIndex) """ ) var data = [Int32]() data.reserveCapacity(primitive.indexCount) for indexOffset in 0 ..< primitive.indexCount { data.append(buffer .contents() .advanced(by: (Int(index) * primitive.indexCount + indexOffset) * MemoryLayout<Int32>.size) .assumingMemoryBound(to: Int32.self).pointee) } return data } func asInt32Array() -> [Int32] { var data = [Int32]() let totalNumberOfInt32 = count * primitive.indexCount data.reserveCapacity(totalNumberOfInt32) for indexOffset in 0 ..< totalNumberOfInt32 { data.append(buffer.contents().advanced(by: indexOffset * MemoryLayout<Int32>.size).assumingMemoryBound(to: Int32.self).pointee) } return data } func asUInt16Array() -> [UInt16] { asInt32Array().map { UInt16($0) } } public func asUInt32Array() -> [UInt32] { asInt32Array().map { UInt32($0) } } } I was also curious to know if I can do this without ARKit using SpatialTrackingSession. My understanding is that using SpatialTrackingSession in RealityKit I can only get the transforms of the AnchorEntities but it won't have geometry information to create the collision shapes.

Now I'm developing a 3D motion capture app by using ARKit. So I tested this sample code, but in iOS18, hand's and leg's orientations seems to be wrong. Forrowing image is sample app's screen captures in iOS17 and iOS18.

Dear all, We are building an XR application demonstrating our research on open-vocabulary 3D instance segmentation for assistive technology. We intend on bringing it to visionOS using the new Enterprise APIs. Our method was trained on datasets resembling ScanNet which contain the following: localized (1) RGB camera frames (2) with Depth (3) and camera intrinsics (4) point cloud (5) I understand, we can query (1), (2), and (4) from the CameraFrameProvider. As for (3) and (4), it is unclear to me if/how we can obtain that data. In handheld ARKit, this example project demos how the depthMap can be used to simulate raw point clouds. However, this property doesn't seem to be available in visionOS. Is there some way for us to obtain depth data associated with camera frames? "Faking" depth data from the SceneReconstructionProvider-generated meshes is too coarse for our method. I hope I'm just missing some detail and there's some way to configure CameraFrameProvider to also deliver depth and/or point clouds. Thanks for any help or pointer in the right direction! ~ Alex

In visionOS, the virtual content is covered by the hand by default, so I want to know that in the hybrid space, if the distance of an entity is behind a real object, how can the object in the room be covered like the virtual content is covered by the hand?

I’m currently working on a project where I need to display a transparent GIF at a specific 3D point within an AR environment. I’ve tried various methods, including using RealityKit with UnlitMaterial and TextureResource, but the GIF still appears with a black background instead of being transparent. Does anyone have experience with displaying animated transparent GIFs on an AR plane or point? Any guidance on how to achieve true transparency for GIFs in ARKit or RealityKit would be appreciated.

I have created a portal and attached it to a wall using the AnchorEntity. However, I am seeking guidance on how to determine the size of the wall so that the portal can fully occupy it. Initially, I attempted to locate relevant information within the demo code, but I encountered difficulties in comprehending certain sections. I would appreciate it if someone could provide a step-by-step explanation or a reference to the appropriate code. Thank you for your assistance.

Hi everyone, I’m working on an app for VisionOS that needs to recognize individual rooms in a hallway based on the person the room belongs to (using the name displayed on each office door). Is there any sample code or resource that can guide me in implementing this feature? Thanks in advance for your help!

Hi everyone, I'm looking for some guidance on how to create an .arobject file. Is there a way to generate it from a 3D model (like a .objcap or .usdz or .fbx/.obj file)? Or can it only be generated by scanning a real-world object using the ARKitScanner project? Any advice or resources on this would be greatly appreciated! I've only found this project for scanning real-world objects. Thanks in advance!

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?

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?

I created an app where the user: first scans a room using RoomCaptureView (RoomPlan) then taps on physical elements (objects, walls...) using an ARView to record some 3d positions I can handle taps in an ARView using a UITapGestureRecognizer and the ARView raycast(from:, allowing:, alignment:) method. This works fine, so I thought I could do the same using the ARView used by RoomCaptureView., so the user can scan a room and record some 3d positions at the same time. Sadly, this approach does not work, as the raycast method always returns nil. What I actually need is mapping a tap on screen to a real-world position during RoomCaptureSession. Does anyone know how to do this?

Hi everyone, I’m having trouble with image anchoring when working on a project in Reality Composer and Reality Composer Pro. Here’s the issue: 1. What I’m Trying to Achieve: I want to create an AR scene where an object anchors to an image I provide. I don't want to create an app for this but just use the USDZ File the Scene creates. The USDZ File then should be viewable via the various integrations of AR Quick Look across the Apple Ecosystem. The image anchoring works perfectly when I preview the scene inside Reality Composer using AR mode. 2. The Problem: When I export the project (tried both USDZ and Reality formats) and open it on my iPhone using the Files app (which uses AR Quick Look), the image anchoring no longer works. The object doesn’t anchor to the provided image as expected. It just anchors to the first plane it recognizes and not the image. 3. What I’ve Tried: Exporting the scene in USDZ format. Exporting the scene in Reality format. Both formats result in the same issue: no image anchoring outside of the Reality Composer environment. Trying different images but all resulting in same manor that the image anchoring is not working Tried different iOS Version but resulting in the same issue 4. Current Setup: Reality Composer Pro version: 2.0 iPhone model: iPhone 13 Pro iOS version: 18.1. 5. What I Need Help With: Is there a way to ensure image anchoring works in exported files when opened via AR Quick Look? Do I need to configure something specific during the export process? Are there limitations in AR Quick Look that prevent image anchoring from functioning correctly? Do i need to create an app to make this work? I’d appreciate any advice or insights from the community. If anyone has experience with similar issues or knows of a workaround, please let me know! Thanks in advance, Mav

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.

Hello All, We're going to do a scene now, kind of like a time travel door. When the user selects the scene, the user passes through the door to show the current scene. The changes in the middle need to be more natural. It's even better if you can walk through an immersive space... There is very little information now. How can I start doing this? Is there any information I can refer to thanks