In the process of using ARKit's image tracking, we found that different images have significant differences in recognizability. How can we judge the quality of this image in ARKit's image tracking for this situation?

Hi Apple, In VisionOS, for real-time streaming of large 3D scenes, I plan to create Metal buffers and textures in multiple threads and then use a compute shader on the main thread to copy the Metal resources into RealityKit, minimizing main thread usage. Given that most of RealityKit's default APIs require execution on the main actor (main thread), it is not ideal for streaming data. Is this approach the best way to handle streaming data and real-time rendering? Thank you very much.

Hi, I wanted to do something quite simple: Put a box on a wall or on the floor. My box: let myBox = ModelEntity( mesh: .generateBox(size: SIMD3<Float>(0.1, 0.1, 0.01)), materials: [SimpleMaterial(color: .systemRed, isMetallic: false)], collisionShape: .generateBox(size: SIMD3<Float>(0.1, 0.1, 0.01)), mass: 0.0) For that I used Plane Detection to identify the walls and floor in the room. Then with SpatialTapGesture I was able to retrieve the position where the user is looking and tap. let position = value.convert(value.location3D, from: .local, to: .scene) And then positioned my box myBox.setPosition(position, relativeTo: nil) When I then tested it I realized that the box was not parallel to the wall but had a slightly inclined angle. I also realized if I tried to put my box on the wall to my left the box was placed perpendicular to this wall and not placed on it. After various searches and several attempts I ended up playing with transform.matrix to identify if the plane is wall or a floor, if it was in front of me or on the side and set up a rotation on the box to "place" it on the wall or a floor. let surfaceTransform = surface.transform.matrix let surfaceNormal = normalize(surfaceTransform.columns.2.xyz) let baseRotation = simd_quatf(angle: .pi, axis: SIMD3<Float>(0, 1, 0)) var finalRotation: simd_quatf if acos(abs(dot(surfaceNormal, SIMD3<Float>(0, 1, 0)))) < 0.3 { logger.info("Surface: ceiling/floor") finalRotation = simd_quatf(angle: surfaceNormal.y > 0 ? 0 : .pi, axis: SIMD3<Float>(1, 0, 0)) } else if abs(surfaceNormal.x) > abs(surfaceNormal.z) { logger.info("Surface: left/right") finalRotation = simd_quatf(angle: surfaceNormal.x > 0 ? .pi/2 : -.pi/2, axis: SIMD3<Float>(0, 1, 0)) } else { logger.info("Surface: front/back") finalRotation = baseRotation } Playing with matrices is not really my thing so I don't know if I'm doing it right. Could you tell me if my tests for the orientation of the walls are correct? During my tests I don't always correctly identify whether the wall is in front or on the side. Is this generally the right way to do it? Is there an easier way to do this? Regards Tof

I have a visionOS app that I’m adding support for IOS and will like to keep using RealityView. I know there are the following modifiers to add some navigation .realityViewCameraControls(.orbit) .realityViewCameraControls(.dolly) .realityViewCameraControls(.pan) But how can I add more than one? For example I would like to orbit with one finger, Pan with 2 fingers and dolly by pinching. Is this possible and if so can someone share some sample code on how to achieve that? Thanks, Guillermo

We are seeing crashes in Xcode organizer. So far we are not able to reproduce them locally. They affect multiple app releases (some older, built with Xcode 15.x and newer built with Xcode 16.0). They only affect iOS 18.5. Is there anything that changed in latest iOS? It's hard to tell what exactly is causing this crash because setting symbolic breakpoint on CA::Render::Image::new_image(unsigned int, unsigned int, unsigned int, unsigned int, CGColorSpace*, void const*, unsigned long const*, void (*)(void const*, void*), void*) triggers this breakpoint all the time, but not necessarily with exactly the previous stack frames matching the crash report. Is it a known issue? crash.crash Thank you.

Hello, I am currently working on a research project under ENINCA Consulting, focused on advanced diagnostic tools for pseudorandom number generators (structural metrics, multi-seed stability, cross-architecture reproducibility, and complementary indicators to TestU01). To validate this diagnostic framework, I prototyped a small non-linear 64-bit PRNG (not as a goal in itself, but simply as a vehicle to test the methodology). During these evaluations, I observed something interesting on Apple Silicon (Mac M1): • bit-exact reproducibility between M1 ARM CPU and M1 Metal GPU, • full BigCrush pass on both CPU and Metal backends, • excellent p-values, • stable behaviour across multiple seeds and runs. This was not the intended objective, the goal was mainly to validate the diagnostic concepts, but these results raised some questions about deterministic compute behaviour in Metal. My question: Is there any official guidance on achieving (or expecting) deterministic RNG or compute behaviour across CPU ↔ Metal GPU on Apple Silicon? More specifically: • Are deterministic compute kernels expected or guaranteed on Metal for scientific workloads? • Are there recommended patterns or best practices to ensure reproducibility across GPU generations (M1 → M2 → M3 → M4)? • Are there known Metal features that can introduce non-determinism? I am not sharing the internal recurrence (this work is proprietary), but I can discuss the high-level diagnostic observations if helpful. Thank you for any insight, very interested in how the Metal engineering team views deterministic compute patterns on Apple Silicon. Pascal ENINCA Consulting

My IOS app generates pdf files. Every time my users open the generated pdf files, the autofill popup jumps out, but my pdf file is NOT for interacting. I'm here to ask if there's a way to mark my pdf files as "not a form", like in metadata or anywhere else?

Hello! I need to "draw" a set of particles into the texture. It would be trivial in render encoder of course. However, I would like to implement the task in compute kernel. Every particle draw operation is expected to set 5 texels - "center" one and left/right/upper/lower. Particles can and will overlap, so concurrent draws are to be expected. I tried using texture atomics - atomic_store() to be more precise. This worked, albeit pretty slowly - too slow for my purpose. Just to test what would happen, I tried using normal texture write(). I was expecting to see some kind of visual artefacts, but to my surprise, it worked very well (and much faster). My question: is it safe? I understand that calling write() doesn't guarantee any ordering of the operations, so if multiple threads write to the same texel, the final value may come from any of those threads. But suppose all the threads were to write the very same color? Can I assume that the texel in question will have said color after the compute kernel finishes? I am using M2 Pro MacBook, but ideally I would love to get the answer for the all Apple Silicon devices. My texture format is R32Int (so as to be able to use atomics), but I could do with any single-channel format, the purpose of the texture is to be binary mask of sorts. Thanks!

RealityKit spatial audio crackles and pops on iOS 26.0 beta 5. It works correctly on iOS 18.6 and visionOS 26.0 beta 5. The APIs used are AudioPlaybackController, Entity.prepareAudio, Entity.play Videos of the expected and observed behavior are attached to the feedback FB19423059. The audio should be a consistent, repeating sound, but it seems oddly abbreviated and the volume varies unexpectedly. Thank you for investigating this issue.

We used below method to resize image while compress the image, Below method is correct or need to do the correction in method or "CGBitmapContextCreate" -(UIImage *)resizeImage:(UIImage *)anImage width:(int)width height:(int)height { CGImageRef imageRef = [anImage CGImage]; CGImageAlphaInfo alphaInfo = CGImageGetAlphaInfo(imageRef); if (alphaInfo == kCGImageAlphaNone) alphaInfo = kCGImageAlphaNoneSkipLast; CGContextRef bitmap = CGBitmapContextCreate(NULL, width, height, CGImageGetBitsPerComponent(imageRef), 4 * width, CGImageGetColorSpace(imageRef), alphaInfo); CGContextDrawImage(bitmap, CGRectMake(0, 0, width, height), imageRef); CGImageRef ref = CGBitmapContextCreateImage(bitmap); UIImage *result = [UIImage imageWithCGImage:ref]; CGContextRelease(bitmap); CGImageRelease(ref); return result; }

Hello, I’m the developer of the “ StepSquad” app. Our app uses the Game Center achievement feature, but we’ve been encountering a problem: the “Global Players” metric always shows 0%, even though there are friends who have already achieved these achievements. Initially, I thought it might be because the app was newly launched. However, it’s now been over two months since release, and it’s still showing 0%. If anyone has any insight into this issue, please leave a comment.

The following minimal snippet SEGFAULTS with SDK 26.0 and 26.1. Won't crash if I remove async from the enclosing function signature - but it's impractical in a real project. import Metal import MetalPerformanceShaders let SEED = UInt64(0x0) typealias T = Float16 /* Why ran in async context? Because global GPU object, and async makeMTLFunction, and async makeMTLComputePipelineState. Nevertheless, can trigger the bug without using global @MainActor let myGPU = MyGPU() */ @main struct CMDLine { static func main() async { let ptr = UnsafeMutablePointer<T>.allocate(capacity: 0) async let future: Void = randomFillOnGPU(ptr, count: 0) print("Main thread is playing around") await future print("Successfully reached the end.") } static func randomFillOnGPU(_ buf: UnsafeMutablePointer<T>, count destbufcount: Int) async { // let (device, queue) = await (myGPU.device, myGPU.commandqueue) let myGPU = MyGPU() let (device, queue) = (myGPU.device, myGPU.commandqueue) // Init MTLBuffer, async let makeFunction, makeComputePipelineState, etc. let tempDataType = MPSDataType.uInt32 let randfiller = MPSMatrixRandomMTGP32(device: device, destinationDataType: tempDataType, seed: Int(bitPattern:UInt(SEED))) print("randomFillOnGPU: successfully created MPSMatrixRandom.") // try await computePipelineState // ^ Crashes before this could return // Or in this minimal case, after randomFillOnGPU() returns // make encoder, set pso, dispatch, commit... } } actor MyGPU { let device : MTLDevice let commandqueue : MTLCommandQueue init() { guard let dev: MTLDevice = MPSGetPreferredDevice(.skipRemovable), let cq = dev.makeCommandQueue(), dev.supportsFamily(.apple6) || dev.supportsFamily(.mac2) else { print("Unable to get Metal Device! Exiting"); exit(EX_UNAVAILABLE) } print("Selected device: \(String(format: "%llX", dev.registryID))") self.device = dev self.commandqueue = cq print("myGPU: initialization complete.") } } See FB20916929. Apparently objc autorelease pool is releasing the wrong address during context switch (across suspension points). I wonder why such obvious case has not been caught before.

Hi all, I have been trying to get Apple's assistive touch's snap to item to work for a unity game built using Apple's Core & Accessibility API. The switch control recognises these buttons however, eye tracking will not snap to them. The case in which it needs to snap is when an external eye tracking device is connected and utilises assistive touch & assistive touch's snap to item. All buttons in the game have a AccessibilityNode with the trait 'Button' on them & an appropriate label, which, following the documentation and comments on the developer forum, should allow them to be recognised by snap to item. This is not the case, devices (iPads and iPhones) do not recognise the buttons as a snap to target. Does anyone know why this is the case, and if this is a bug?

The code is pretty simple kernel void naive( constant RunParams *param [[ buffer(0) ]], const device float *A [[ buffer(1) ]], // [N, K] device float *output [[ buffer(2) ]], uint2 gid [[ thread_position_in_grid ]]) { uint a_ptr = gid.x * param->K; for (uint i = 0; i < param->K; i++, a_ptr++) { val += A[b_ptr]; } output[ptr] = val; } when uint a_ptr = gid.x * param->K, the code got 150 GFLops when uint a_ptr = gid.y * param->K, the code got 860 GFLops param->K = 256; thread per group: [16, 16] I'd like to understand why the performance is so different, and how can I profile/diagnose this to help with further optimization.

The “explore spatial accessory input on visionOS” presentation from WDC25 interests me. I bought both the MUSE Logitech stylus and the PS VR2 sense controllers to try out with the sculpting app presented by the author, engineer Amanda Han. Unfortunately the app itself was not included. Could the app be made available for downloading as well as the Xcode project? I appreciate any assistance the author and your team could provide. Thank you.

Hello, I am trying to read video frames using AVAssetReaderTrackOutput. Here is the sample code: //prepare assets let asset = AVURLAsset(url: some_url) let assetReader = try AVAssetReader(asset: asset) guard let videoTrack = try await asset.loadTracks(withMediaCharacteristic: .visual).first else { throw SomeErrorCode.error } var readerSettings: [String: Any] = [ kCVPixelBufferIOSurfacePropertiesKey as String: [String: String]() ] //check if HDR video var isHDRDetected: Bool = false let hdrTracks = try await asset.loadTracks(withMediaCharacteristic: .containsHDRVideo) if hdrTracks.count > 0 { readerSettings[AVVideoAllowWideColorKey as String] = true readerSettings[kCVPixelBufferPixelFormatTypeKey as String] = kCVPixelFormatType_420YpCbCr10BiPlanarFullRange isHDRDetected = true } //add output to assetReader let output = AVAssetReaderTrackOutput(track: videoTrack, outputSettings: readerSettings) guard assetReader.canAdd(output) else { throw SomeErrorCode.error } assetReader.add(output) guard assetReader.startReading() else { throw SomeErrorCode.error } //add writer ouput settings let videoOutputSettings: [String: Any] = [ AVVideoCodecKey: AVVideoCodecType.hevc, AVVideoWidthKey: 1920, AVVideoHeightKey: 1080, ] let finalPath = "//some URL oath" let assetWriter = try AVAssetWriter(outputURL: finalPath, fileType: AVFileType.mov) guard assetWriter.canApply(outputSettings: videoOutputSettings, forMediaType: AVMediaType.video) else { throw SomeErrorCode.error } let assetWriterInput = AVAssetWriterInput(mediaType: .video, outputSettings: videoOutputSettings) let sourcePixelAttributes: [String: Any] = [ kCVPixelBufferPixelFormatTypeKey as String: isHDRDetected ? kCVPixelFormatType_420YpCbCr10BiPlanarFullRange : kCVPixelFormatType_32ARGB, kCVPixelBufferWidthKey as String: 1920, kCVPixelBufferHeightKey as String: 1080, ] //create assetAdoptor let assetAdaptor = AVAssetWriterInputTaggedPixelBufferGroupAdaptor( assetWriterInput: assetWriterInput, sourcePixelBufferAttributes: sourcePixelAttributes) guard assetWriter.canAdd(assetWriterInput) else { throw SomeErrorCode.error } assetWriter.add(assetWriterInput) guard assetWriter.startWriting() else { throw SomeErrorCode.error } assetWriter.startSession(atSourceTime: CMTime.zero) //prepare tranfer session var session: VTPixelTransferSession? = nil guard VTPixelTransferSessionCreate(allocator: kCFAllocatorDefault, pixelTransferSessionOut: &session) == noErr, let session else { throw SomeErrorCode.error } guard let pixelBufferPool = assetAdaptor.pixelBufferPool else { throw SomeErrorCode.error } //read through frames while let nextSampleBuffer = output.copyNextSampleBuffer() { autoreleasepool { guard let imageBuffer = CMSampleBufferGetImageBuffer(nextSampleBuffer) else { return } //this part copied from (https://developer.apple.com/videos/play/wwdc2023/10181) at 23:58 timestamp let attachment = [ kCVImageBufferYCbCrMatrixKey: kCVImageBufferYCbCrMatrix_ITU_R_2020, kCVImageBufferColorPrimariesKey: kCVImageBufferColorPrimaries_ITU_R_2020, kCVImageBufferTransferFunctionKey: kCVImageBufferTransferFunction_SMPTE_ST_2084_PQ, ] CVBufferSetAttachments(imageBuffer, attachment as CFDictionary, .shouldPropagate) //now convert to CIImage with HDR data let image = CIImage(cvPixelBuffer: imageBuffer) let cropped = "" //here perform some actions like cropping, flipping, etc. and preserve this changes by converting the extent to CGImage first: //this part copied from (https://developer.apple.com/videos/play/wwdc2023/10181) at 24:30 timestamp guard let cgImage = context.createCGImage( cropped, from: cropped.extent, format: .RGBA16, colorSpace: CGColorSpace(name: CGColorSpace.itur_2100_PQ)!) else { continue } //finally convert it back to CIImage let newScaledImage = CIImage(cgImage: cgImage) //now write it to a new pixelBuffer let pixelBufferAttributes: [String: Any] = [ kCVPixelBufferCGImageCompatibilityKey as String: true, kCVPixelBufferCGBitmapContextCompatibilityKey as String: true, ] var pixelBuffer: CVPixelBuffer? CVPixelBufferCreate( kCFAllocatorDefault, Int(newScaledImage.extent.width), Int(newScaledImage.extent.height), kCVPixelFormatType_420YpCbCr10BiPlanarFullRange, pixelBufferAttributes as CFDictionary, &pixelBuffer) guard let pixelBuffer else { continue } context.render(newScaledImage, to: pixelBuffer) //context is a CIContext reference var pixelTransferBuffer: CVPixelBuffer? CVPixelBufferPoolCreatePixelBuffer(kCFAllocatorDefault, pixelBufferPool, &pixelTransferBuffer) guard let pixelTransferBuffer else { continue } // Transfer the image to the pixel buffer. guard VTPixelTransferSessionTransferImage(session, from: pixelBuffer, to: pixelTransferBuffer) == noErr else { continue } //finally append to taggedBuffer } } assetWriterInput.markAsFinished() await assetWriter.finishWriting() The result video is not in correct color as the original video. It turns out too bright. If I play around with attachment values, it can be either too dim or too bright but not exactly proper as the original video. What am I missing in my setup? I did find that kCVPixelFormatType_4444AYpCbCr16 can produce proper video output but then I can't convert it to CIImage and so I can't do the CIImage operations that I need. Mainly cropping and resizing the CIImage

Hello Apple team, I'm working on an iOS AR app using SwiftUI and RealityKit, and I was wondering if the Cinematic API can be used with a RealityKit scene. I’d like to achieve a shallow depth of field while keeping the 3D asset in focus, and vice versa. Thanks!

Create the QRCode CIFilter<CIBlendWithMask> *f = CIFilter.QRCodeGenerator; f.message = [@"Message" dataUsingEncoding:NSASCIIStringEncoding]; f.correctionLevel = @"Q"; // increase level CIImage *qrcode = f.outputImage; Overlay the icon CIImage *icon = [CIImage imageWithURL:url]; CGAffineTransform *t = CGAffineTransformMakeTranslation( (qrcode.extent.width-icon.extent.width)/2.0, (qrcode.extent.height-icon.extent.height)/2.0); icon = [icon imageByApplyingTransform:t]; qrcode = [icon imageByCompositingOver:qrcode]; Round off the corners static dispatch_once_t onceToken; static CIWarpKernel *k; dispatch_once(&onceToken, ^ { k = [CIWarpKernel kernelWithFunctionName:name fromMetalLibraryData:metalLibData() error:nil]; }); CGRect iExtent = image.extent; qrcode = [k applyWithExtent:qrcode.extent roiCallback:^CGRect(int i, CGRect r) { return CGRectInset(r, -radius, -radius); } inputImage:qrcode arguments:@[[CIVector vectorWithCGRect:qrcode.extent], @(radius)]]; …and this code for the kernel should go in a separate .ci.metal source file: float2 bend_corners (float4 extent, float s, destination dest) { float2 p, dc = dest.coord(); float ratio = 1.0; // Round lower left corner p = float2(extent.x+s,extent.y+s); if (dc.x < p.x && dc.y < p.y) { float2 d = abs(dc - p); ratio = min(d.x,d.y)/max(d.x,d.y); ratio = sqrt(1.0 + ratio*ratio); return (dc - p)*ratio + p; } // Round lower right corner p = float2(extent.x+extent.z-s, extent.y+s); if (dc.x > p.x && dc.y < p.y) { float2 d = abs(dc - p); ratio = min(d.x,d.y)/max(d.x,d.y); ratio = sqrt(1.0 + ratio*ratio); return (dc - p)*ratio + p; } // Round upper left corner p = float2(extent.x+s,extent.y+extent.w-s); if (dc.x < p.x && dc.y > p.y) { float2 d = abs(dc - p); ratio = min(d.x,d.y)/max(d.x,d.y); ratio = sqrt(1.0 + ratio*ratio); return (dc - p)*ratio + p; } // Round upper right corner p = float2(extent.x+extent.z-s, extent.y+extent.w-s); if (dc.x > p.x && dc.y > p.y) { float2 d = abs(dc - p); ratio = min(d.x,d.y)/max(d.x,d.y); ratio = sqrt(1.0 + ratio*ratio); return (dc - p)*ratio + p; } return dc; }

I am using Unity's GameKit to implement a turnbase game. I want to make a UI in Unity to show all the games I can join. I tried using var matches = await GKTurnBasedMatch.LoadMatches(); to get all the open matches. But it seems that I can only get the matcm related to the current apple account. Can you help me get all the matches? ALSO I used var match = await GKTurnBasedMatchmakerViewController.Request(request); to exit the gamecenter interface and start a game (automatic matching, no one was invited) Another device used var match = await GKTurnBasedMatch.Find(request); to find the game, but it did not find the game, but it start a new game (automatic matching). Can you help me solve these problems?

I have run into an issue where I am trying to use atomic_float in a swift package but I cannot get things to compile because it appears that the Swift Package Manager doesn't support Metal 3 (atomic_float is Metal 3 functionality). Is there any way around this? I am using // swift-tools-version: 6.1 and my Metal code includes: #include <metal_stdlib> #include <metal_geometric> #include <metal_math> #include <metal_atomic> using namespace metal; kernel void test(device atomic_float* imageBuffer [[buffer(1)]], uint id [[ thread_position_in_grid ]]) { } But I get an error on the definition of atomic_float . Any help, one more importantly, where I could have found this information about this limitation, would be helpful. -RadBobby