Hi, recently some users of my app upgrade MacOS to 15, and encounter hang when trying to get screen capture permission. After looking into this issue, I find out a repoducible way. If two processes of a same program both call CGDisplayStreamCreate/CGDisplayImageCreate API, the later one will hang(No need to be at almost same time). Here are my demo codes. #import <AVFoundation/AVFoundation.h> #import <CoreGraphics/CoreGraphics.h> #import <Foundation/Foundation.h> int main(int argc, const char *argv[]) { @autoreleasepool { CGDirectDisplayID displayID = CGMainDisplayID(); CGDisplayStreamRef stream = CGDisplayStreamCreate( displayID, 1, 1, kCVPixelFormatType_32BGRA, nil, ^(CGDisplayStreamFrameStatus status, uint64_t displayTime, IOSurfaceRef frameSurface, CGDisplayStreamUpdateRef updateRef){ }); if (!stream) { NSLog(@"Failed to create display stream!"); return 1; } else { NSLog(@"Create display stream success!"); CFRelease(stream); } } sleep(10); return 0; } The call of sleep is just to ensure I can run a second process of this program before the first one ends. It turns out that the first process checks the permission correctly and sleeps successfully. However the second process won't be able to print anything because it hangs when calling CGDisplayStreamCreate. Here is the output of sample: Call graph: 2546 Thread_1315735 DispatchQueue_1: com.apple.main-thread (serial) + 2546 start (in dyld) + 2840 [0x196f8f274] + 2546 main (in testapp) + 76 [0x100ef7e38] + 2546 SLDisplayStreamCreate (in SkyLight) + 316 [0x19d38163c] + 2546 SLSDisplayStreamCreateProxying (in ScreenCaptureKit) + 816 [0x231e2d614] + 2546 _dispatch_semaphore_wait_slow (in libdispatch.dylib) + 76 [0x19715e278] + 2546 _dispatch_sema4_timedwait (in libdispatch.dylib) + 64 [0x19715dc78] + 2546 semaphore_timedwait_trap (in libsystem_kernel.dylib) + 8 [0x1972ceda8] 2546 Thread_1315737 2546 start_wqthread (in libsystem_pthread.dylib) + 8 [0x19730b0f0] 2546 _pthread_wqthread (in libsystem_pthread.dylib) + 364 [0x19730c424] 2546 __workq_kernreturn (in libsystem_kernel.dylib) + 8 [0x1972d0a64] If I copy the program and run two different programs, the second program won't hang. This issue doesn't happen on MacOS early version. Can anybody help?

Start carpture window A, it works fine. After some time, I want to switch to capture window B without stop the scstream. My code is like this : content_filter = [[SCContentFilter alloc] initWithDesktopIndependentWindow:winID_B]; [scStream_ updateContentFilter:content_filter completionHandler:^(NSError *_Nullable error) { if (error) { "some error log" } else { "update OK" } }]; I cant get the "update OK" log, but sometimes it still output frame of the old window A. Is there any special operation need do before call updateContentFilter ? Or, is my usage is not correct ?

I'm trying to capture audio samples from the selected output device on macOS using ScreenCaptureKit? Thank you

I am trying to take a screenshot of each running window on the screen independently (even if not in the foreground). I am using SCScreenshotManager's captureImageWithFilter and SCContentFilter's initWithDesktopIndependentWindow. When I do this for every SCWindow in SCShareableContent's windows, I do get all the actual windows, but also get many mostly blank windows or windows that no ordinary user would consider to be an actual window. These include dozens of windows with no title, some with "Focus Proxy" title, some for menu bar icons, the wallpaper, the desktop icons, etc. I've implemented a naive solution that filters all windows that have no title, owningApplications with no title or bundle ID in a hardcoded blocklist (e.g. "com.apple.controlcenter", "com.apple.wallpaper.agent") and that helps, but is far from robust and is naturally fragile. Is there a recommended way to distinguish actual application windows or an overall better approach here? Thanks!

I am capturing a screenshot with SCScreenshotManager's captureImageWithFilter. The resulting PNG has the same resolution as the PNG taken from Command-Shift-3 (4112x2658) but is 10x larger (14.4MB vs 1.35MB). My SCStreamConfiguration uses the SCDisplay's width and height and sets the color space to kCGColorSpaceSRGB. I currently save to file by initializing a NSBitmapImageRep using initWithCGImage, then representing as PNG with representationUsingType NSBitmapImageFileTypePNG, then writeToFile:atomically. Is there some configuration or compression I can use to bring down the PNG size to be more closely in-line with a Command-Shift-3 screenshot. Thanks!

I use ScreenCaptureKit, CoreVideo, CoreImage, CoreMedia frameworks to capture screenshots on macOS 14.0 and higher. Example of creating CGImageRef: CVImageBufferRef cvImageBufferRef = ..; CIImage* temporaryImage = [CIImage imageWithCVPixelBuffer:cvImageBufferRef]; CIContext* temporaryContext = [CIContext context]; CGImageRef imageRef = [temporaryContext createCGImage:temporaryImage fromRect:CGRectMake(0, 0, CVPixelBufferGetWidth(cvImageBufferRef), CVPixelBufferGetHeight(cvImageBufferRef))]; I have the next results of profiling with XCode Instruments Memory Leaks & Allocations: there is constantly increasing memory usage, but no memory leaks are detected, and there are many calls to create IOSurface objects, that have been never released. The most part of memory - All Anonymous VM - VM: IOSurface. The heaviest stack trace: [RPIOSurfaceObject initWithCoder:] [IOSurface initWithMachPort:] IOSurfaceClientLookupFromMachPort I don't have any of IOSurface objects created by myself. There are low-level calls to it. In Allocation List I can see many allocations of IOSurface objects, but there are no info about releasing it. Due to this info, how can I release them to avoid permanent increasing memory consumption?

I am using Screen Capture Kit to capture the windows and record it. But from macOS Sanoma onwards I see a wired behaviour when I try to capture the window which is in Full screen mode. The CMSampleBuffer returned by Screen capture kit has empty space at the top of the full screen window content. The ContentRect attachment in CMSampleBuffer includes this empty space. So there is no way to know what is the actual window content in the CMSampleBuffer. In the CaptureCample sample code provided by Apple it does not enumerate the Full screen windows. I made a change in that to enumerate full screen windows. The issue is reproduced in that also. Attaching the Image of showing the empty space. Has anybody encountered this issue?

I am working on ScreenCaptureKit sample with SCContentSharingPickerObserver. My Target is SwiftUI based and calling Objective-C class method. I added [MyTarget]-Bridging.h and [MyTarget]-Swift.h I got compile error of unknown class name SCContentSharingPickerObserver in [MyTarget]-Swift.h. But I do not know how to fix this error since [MyTarget]-Swift.h is Xcode generated file. I set macOS target is 14.0, swift language ver is 5 Anyone know how to fix this error or waiting for Xcode update?

I have old ScreenCaptureKit sample downloaded on Oct 2022. That sample worked on Oct 2022. But it does not work on Apr 2024 on Sonoma 14.4.1 M1 MacBook. It only shows black screen. I also download updated ScreenCaptureKit sample and test it. It works on Sonoma 14.4.1 M1 MacBook. I noticed latest sample have SCContentSharingPicker and other changes. I have my screen capture application based on old ScreenCaptureKit sample. My app only shows black screen. Do I have to add SCContentSharingPicker and SCContentSharingPickerObserver on my application for capturing screen on Sonoma? Old way of screen capture without SCContentSharingPicker is not supported anymore on Sonoma?

Hello! I'm trying to find my own window or SCWindow in a SCDisplay. This is so I can automatically change the SCDisplay to record when I drag a window from a display to another. Is there any way to check the windows that are contained in every SCDisplay? Thank you in advance!

Hi, I'm new to AVAudioEngine(and macOS programming in general). I'm trying to mix microphone audio with ScreenCaptureKit audio using AVAudioEngine without playing it back. I've created a AVAudioPlayerNode and scheduling buffers in my SCStream handler: playerNode.scheduleBuffer(samples) and have connected the playerNode to the mainMixerNode. audioEngine.connect(audioEngine.inputNode, to: audioEngine.mainMixerNode, format: micFormat) audioEngine.connect(playerNode, to: audioEngine.mainMixerNode, format: format) The problem is that mainMixerNode plays the audio to the speaker creating a feedback loop. How can I prevent the mixer output from being played back. Also: Is this the best way of mixing microphone input with some other input? I ran into AVAudioEngine's manual rendering mode, which seems like the way to go for mixing audio without playing it back. However, I couldn't figure out how to connect microphone input to the AVAudioEngine in manual rendering mode?

If someone in Apple WWDR sees this, please take the feedback to heart and report it up the chain: When you announce that a technology is being deprecated — such as CGDisplayStream — and also publish WWDC sessions about the intended replacement — ScreenCaptureKit — then you also need to give third-party developers a clear deadline by which this technology will be deprecated so that they can plan engineering efforts around implementing the new feature, and have ample time to communicate this to their customers. If it's important for third-party developers to get on board with this change, you should use every available means to communicate this to them, including multiple email alerts to their registered email address. Additionally, if you plan to make a BREAKING change in a framework that results in a wildly different user experience, you should probably hold that off until the summer release for the next major OS. What you should definitely NOT do is roll out a new privacy prompt in a mid-year release of macOS; or give your developers, customers, and AppleSeed program participants zero advance notice that this alert is coming, ignore your own Human Interface Guidelines when designing said prompt, and perform no user experience design testing (aka "putting on your customer hat") during a presumed internal alpha testing cycle to refine the experience and still find the most effective and least annoying way to present this additional prompt and spur change with your third-party developers. Oh, wait, you've done exactly all those things the wrong way with respect to ScreenCaptureKit. Right now, a host of Apple device administrators and client platform engineers are sending mountains of feedback to you, and they're also scrambling to contact third-party developers to let them know this is coming. Most of the vendors being discussed in private forums are said to be caught off guard by this change. We anticipate that users are not going to like this, and there is no way we can manage it with MDM or configuration profiles. In short, the current experience is a ghastly mess. WE, the administrators, will get blamed for this, not the third-party developers. WE will have to explain to our leadership why this experience is terrible and cannot be managed. Engineers need deadlines to help plan their work and prioritize tasks. In this case, vendors have had no firm deadline for this effort. There's already precedence for Apple announcing estimated deadlines for deprecations and feature removals. You do your developers and customers a great disservice by not communicating schedules to them. Please do better. P.S.: Feedback filed as FB13619326.

Hello, I'm playing with the ScreenCaptureKit. My scenario is very basic: I capture frames and assign them to a CALayer to display in a view (for previewing) - basically what Apple's sample app does. I have two screens, and I capture only one of them, which has no app windows on it. And my app excludes itself from capture. So, when I place my app on the screen which is not being captured, I observe that most didOutputSampleBuffer calls receive frames with Idle status (which is expected for a screen which is not updating). However, if I bring my capturing app (which, to remind, is excluded from capture) to the captured screen, the frames start coming with Complete status (i.e. holding pixel buffers). And this is not what I expect - from capture perspective the screen is still not updating, as the app is excluded. The preview which the app displays proves that, showing empty desktop. So it seems like updates to a CALayer triggers screen capture frame regardless of the app being included or excluded. This looks like a bug to me, and can lead to noticable waste of resources and battery when frames are not just displayed on screen, but also processed somehow or/and sent over a network. Also, I'm observing another issue due to this behavior, where capture hangs when queueDepth is set to 3 in this same scenario (but I'll describe it separately). Please, advise if I should file a bug somewhere, or maybe there is a rational explanation of this behavior. Thank you

I found that didOutputSampleBuffer would not be called for long time when the screen is no change. It sometimes make me confused for if there something wrong. In my design, it will change to other screen shot method, such as CGWindowListCreateImage, when long time no data. But this is not what I expected. I set the minimumFrameInterval to 30 but it seems no work. [config setMinimumFrameInterval:CMTimeMake(1, 30)]; Is there any settings that can let me get a didOutputSampleBuffer, even given a CMSampleBufferRef with status SCFrameStatusIdle, called atlest one time per second? Which would make me think it works fine without any exception.

I notice from macOS Sonoma System Settings, we have "Screen & System audio Recording". I'm an macOS app developer and want to request only Audio permission, I browse the document for a while and WWDC code demo, but still have no idea of how to request "System Audio Recording Only" permission? All the demo and doc I can find is request "Screen Recording & System Audio"

When I try to initialize SCContentFilter(desktopIndependentWindow: window), I get a very weird error: Assertion failed: (did_initialize), function CGS_REQUIRE_INIT, file CGInitialization.c, line 44. This is a brand new CLI project on Sonoma 14.2.1. Any ideas?

Hello, I develop an application called MiniMeters and am using ScreenCaptureKit to get the desktop audio on macOS. As of macOS 14.2, a few users began noticing that the volume of the incoming audio differed depending on the audio device connected. One user's Apogee Symphony Desktop is trimmed -14dB, another user's UAD Apollo Twin is -14dB as well, my MOTU M4 is -6dB, and my MacBook Pro's internal speakers show 0dB. This does not change with changing the output volume on the interface (obviously), nor digitally in the system. I cannot seem to find anything documenting this change. It also affects other applications that use ScreenCaptureKit such as OBS. Does anyone have any idea what that could correlate to so I could potentially compensate? Thanks.

I try capture screen on wingon. For start app on winlogon i register LaunchAgent to start on LoginWindow. Application is started, but callback getShareableContentExcludingDesktopWindows newer called, and application just stuck.

So I'm building a colour sampler tool, similar to ColorSlurp and since CGWindowListCreateImage (which I am using for below macOS 14.0) is deprecated since Sonoma, I am wondering what's the best approach to replacing it. The way I use CGWindowListCreateImage currently is to take a screenshot of a specified area around the mouse pointer every time the mouse moves. Which works perfectly fine without issues. Now I've tried replacing CGWindowListCreateImage with SCScreenshotManager.createImage which is an async function and as you might expect, running async functions on mouse movements doesn't quite work out that well. It is lagging behind, heavily. So my question would be what's the appropriate ScreenCaptureKit methods to replace the functionality already created with CGWindowListCreateImage? Should I create a SCStream instead? Then I am worried about the fact that I would need to stream the whole screen instead of just the area around the mouse pointer since updating stream configs is an async function as well. I'd greatly appreciate any sort of direction!

Guys, In my main application bundle, I have included a helper bundle in its Resources. When the helper requests Accessibility permission, the system modal window displays what the helper is requesting permission for. However, when the helper requests permission for Screen Recording, the system modal window displays that the main application bundle is requesting permission, which includes the helper. This issue seems to be specific to Ventura, as both requests are displayed on behalf of the helper in Monterey. I'm wondering if this is a known issue or limitation or if there is a way to make the permission request specifically from the helper.