将 Metal MTKView 实时捕获为电影？答案

【问题标题】：Capture Metal MTKView as Movie in realtime?将 Metal MTKView 实时捕获为电影？
【发布时间】：2017-10-05 21:24:22
【问题描述】：

从MTKView 捕获帧的最有效方法是什么？如果可能的话，我想实时保存帧中的 .mov 文件。是否可以渲染成 AVPlayer 框架之类的？

目前正在使用此代码进行绘制（基于@warrenm PerformanceShaders project）：

func draw(in view: MTKView) {
     _ = inflightSemaphore.wait(timeout: DispatchTime.distantFuture)
             updateBuffers()

    let commandBuffer = commandQueue.makeCommandBuffer()

    commandBuffer.addCompletedHandler{ [weak self] commandBuffer in
        if let strongSelf = self {
            strongSelf.inflightSemaphore.signal()
        }
    }
   // Dispatch the current kernel to perform the selected image filter
    selectedKernel.encode(commandBuffer: commandBuffer,
        sourceTexture: kernelSourceTexture!,
        destinationTexture: kernelDestTexture!)

    if let renderPassDescriptor = view.currentRenderPassDescriptor, let currentDrawable = view.currentDrawable
    {
        let clearColor = MTLClearColor(red: 0, green: 0, blue: 0, alpha: 1)
        renderPassDescriptor.colorAttachments[0].clearColor = clearColor

        let renderEncoder = commandBuffer.makeRenderCommandEncoder(descriptor: renderPassDescriptor)
        renderEncoder.label = "Main pass"

        renderEncoder.pushDebugGroup("Draw textured square")
        renderEncoder.setFrontFacing(.counterClockwise)
        renderEncoder.setCullMode(.back)

        renderEncoder.setRenderPipelineState(pipelineState)
        renderEncoder.setVertexBuffer(vertexBuffer, offset: MBEVertexDataSize * bufferIndex, at: 0)
        renderEncoder.setVertexBuffer(uniformBuffer, offset: MBEUniformDataSize * bufferIndex , at: 1)
        renderEncoder.setFragmentTexture(kernelDestTexture, at: 0)
        renderEncoder.setFragmentSamplerState(sampler, at: 0)
        renderEncoder.drawPrimitives(type: .triangleStrip, vertexStart: 0, vertexCount: 4)

        renderEncoder.popDebugGroup()
        renderEncoder.endEncoding()

        commandBuffer.present(currentDrawable)
    }

    bufferIndex = (bufferIndex + 1) % MBEMaxInflightBuffers

    commandBuffer.commit()
 }

【问题讨论】：

这当然是可能的。由于在 iOS 上从 IOSurface 创建 CVPixelBuffer 不是受支持的操作，因此最快的方法可能是使用 AVAssetWriterInputPixelBufferAdaptor（包装像素缓冲池），您可以在其中读取内容每一帧的drawable的纹理。有一些示例说明如何使用浮动的 OpenGL ES 执行此操作。我最终会尝试发布一个包含所有细节的更完整的答案，但也许这会对你有所帮助。另一个要考虑的选择是 ReplayKit，如果它足够灵活，可以做你想做的事。

标签： ios swift3 metal avcapture

【解决方案1】：

这是一个小类，它执行编写捕获 Metal 视图内容的电影文件的基本功能：

class MetalVideoRecorder {
    var isRecording = false
    var recordingStartTime = TimeInterval(0)

    private var assetWriter: AVAssetWriter
    private var assetWriterVideoInput: AVAssetWriterInput
    private var assetWriterPixelBufferInput: AVAssetWriterInputPixelBufferAdaptor

    init?(outputURL url: URL, size: CGSize) {
        do {
            assetWriter = try AVAssetWriter(outputURL: url, fileType: .m4v)
        } catch {
            return nil
        }

        let outputSettings: [String: Any] = [ AVVideoCodecKey : AVVideoCodecType.h264,
            AVVideoWidthKey : size.width,
            AVVideoHeightKey : size.height ]

        assetWriterVideoInput = AVAssetWriterInput(mediaType: .video, outputSettings: outputSettings)
        assetWriterVideoInput.expectsMediaDataInRealTime = true

        let sourcePixelBufferAttributes: [String: Any] = [
            kCVPixelBufferPixelFormatTypeKey as String : kCVPixelFormatType_32BGRA,
            kCVPixelBufferWidthKey as String : size.width,
            kCVPixelBufferHeightKey as String : size.height ]

        assetWriterPixelBufferInput = AVAssetWriterInputPixelBufferAdaptor(assetWriterInput: assetWriterVideoInput,
                                                                           sourcePixelBufferAttributes: sourcePixelBufferAttributes)

        assetWriter.add(assetWriterVideoInput)
    }

    func startRecording() {
        assetWriter.startWriting()
        assetWriter.startSession(atSourceTime: .zero)

        recordingStartTime = CACurrentMediaTime()
        isRecording = true
    }

    func endRecording(_ completionHandler: @escaping () -> ()) {
        isRecording = false

        assetWriterVideoInput.markAsFinished()
        assetWriter.finishWriting(completionHandler: completionHandler)
    }

    func writeFrame(forTexture texture: MTLTexture) {
        if !isRecording {
            return
        }

        while !assetWriterVideoInput.isReadyForMoreMediaData {}

        guard let pixelBufferPool = assetWriterPixelBufferInput.pixelBufferPool else {
            print("Pixel buffer asset writer input did not have a pixel buffer pool available; cannot retrieve frame")
            return
        }

        var maybePixelBuffer: CVPixelBuffer? = nil
        let status  = CVPixelBufferPoolCreatePixelBuffer(nil, pixelBufferPool, &maybePixelBuffer)
        if status != kCVReturnSuccess {
            print("Could not get pixel buffer from asset writer input; dropping frame...")
            return
        }

        guard let pixelBuffer = maybePixelBuffer else { return }

        CVPixelBufferLockBaseAddress(pixelBuffer, [])
        let pixelBufferBytes = CVPixelBufferGetBaseAddress(pixelBuffer)!

        // Use the bytes per row value from the pixel buffer since its stride may be rounded up to be 16-byte aligned
        let bytesPerRow = CVPixelBufferGetBytesPerRow(pixelBuffer)
        let region = MTLRegionMake2D(0, 0, texture.width, texture.height)

        texture.getBytes(pixelBufferBytes, bytesPerRow: bytesPerRow, from: region, mipmapLevel: 0)

        let frameTime = CACurrentMediaTime() - recordingStartTime
        let presentationTime = CMTimeMakeWithSeconds(frameTime, preferredTimescale: 240)
        assetWriterPixelBufferInput.append(pixelBuffer, withPresentationTime: presentationTime)

        CVPixelBufferUnlockBaseAddress(pixelBuffer, [])
    }
}

在初始化其中一个并调用 startRecording() 后，您可以将调度的处理程序添加到包含渲染命令的命令缓冲区并调用 writeFrame（在结束编码之后，但在呈现可绘制对象或提交缓冲区之前）：

let texture = currentDrawable.texture
commandBuffer.addCompletedHandler { commandBuffer in
    self.recorder.writeFrame(forTexture: texture)
}

录制完成后，只需拨打endRecording，视频文件将完成并关闭。

注意事项：

此类假定源纹理为默认格式.bgra8Unorm。如果不是，您将遇到崩溃或损坏。如有必要，使用计算或片段着色器转换纹理，或使用 Accelerate。

该类还假设纹理与视频帧的大小相同。如果不是这种情况（如果可绘制大小发生变化，或者您的屏幕自动旋转），输出将被破坏并且您可能会看到崩溃。根据您的应用程序需要，通过缩放或裁剪源纹理来缓解此问题。

【讨论】：

你好沃伦。我有一个问题可能是因为pixelFormat。我在这里有一个问题：stackoverflow.com/questions/46146593/… 和 github 上的一个问题：github.com/svtek/SceneKitVideoRecorder/issues/3
你为什么用'addScheduledHandler'而不是'addCompletedHandler'？
我认为使用调度处理程序是不正确的，因为该块将在 GPU 执行命令缓冲区之前运行。您需要使用完成处理程序。
还应该注意MTKView的属性framebufferOnly应该设置为false。由于这个原因，我最近遇到了问题，并且在视频中出现了伪影/崩溃
我在texture.getBytes(...) 收到Thread 1: EXC_BAD_ACCESS。有任何想法吗？花了几天时间尝试调试。

【解决方案2】：

升级到 Swift 5


import AVFoundation

class MetalVideoRecorder {
    var isRecording = false
    var recordingStartTime = TimeInterval(0)

    private var assetWriter: AVAssetWriter
    private var assetWriterVideoInput: AVAssetWriterInput
    private var assetWriterPixelBufferInput: AVAssetWriterInputPixelBufferAdaptor

    init?(outputURL url: URL, size: CGSize) {
        do {
          assetWriter = try AVAssetWriter(outputURL: url, fileType: AVFileType.m4v)
        } catch {
            return nil
        }

      let outputSettings: [String: Any] = [ AVVideoCodecKey : AVVideoCodecType.h264,
            AVVideoWidthKey : size.width,
            AVVideoHeightKey : size.height ]

      assetWriterVideoInput = AVAssetWriterInput(mediaType: AVMediaType.video, outputSettings: outputSettings)
        assetWriterVideoInput.expectsMediaDataInRealTime = true

        let sourcePixelBufferAttributes: [String: Any] = [
            kCVPixelBufferPixelFormatTypeKey as String : kCVPixelFormatType_32BGRA,
            kCVPixelBufferWidthKey as String : size.width,
            kCVPixelBufferHeightKey as String : size.height ]

        assetWriterPixelBufferInput = AVAssetWriterInputPixelBufferAdaptor(assetWriterInput: assetWriterVideoInput,
                                                                           sourcePixelBufferAttributes: sourcePixelBufferAttributes)

        assetWriter.add(assetWriterVideoInput)
    }

    func startRecording() {
        assetWriter.startWriting()
      assetWriter.startSession(atSourceTime: CMTime.zero)

        recordingStartTime = CACurrentMediaTime()
        isRecording = true
    }

    func endRecording(_ completionHandler: @escaping () -> ()) {
        isRecording = false

        assetWriterVideoInput.markAsFinished()
        assetWriter.finishWriting(completionHandler: completionHandler)
    }

    func writeFrame(forTexture texture: MTLTexture) {
        if !isRecording {
            return
        }

        while !assetWriterVideoInput.isReadyForMoreMediaData {}

        guard let pixelBufferPool = assetWriterPixelBufferInput.pixelBufferPool else {
            print("Pixel buffer asset writer input did not have a pixel buffer pool available; cannot retrieve frame")
            return
        }

        var maybePixelBuffer: CVPixelBuffer? = nil
        let status  = CVPixelBufferPoolCreatePixelBuffer(nil, pixelBufferPool, &maybePixelBuffer)
        if status != kCVReturnSuccess {
            print("Could not get pixel buffer from asset writer input; dropping frame...")
            return
        }

        guard let pixelBuffer = maybePixelBuffer else { return }

        CVPixelBufferLockBaseAddress(pixelBuffer, [])
        let pixelBufferBytes = CVPixelBufferGetBaseAddress(pixelBuffer)!

        // Use the bytes per row value from the pixel buffer since its stride may be rounded up to be 16-byte aligned
        let bytesPerRow = CVPixelBufferGetBytesPerRow(pixelBuffer)
        let region = MTLRegionMake2D(0, 0, texture.width, texture.height)

        texture.getBytes(pixelBufferBytes, bytesPerRow: bytesPerRow, from: region, mipmapLevel: 0)

        let frameTime = CACurrentMediaTime() - recordingStartTime
        let presentationTime = CMTimeMakeWithSeconds(frameTime, preferredTimescale:   240)
        assetWriterPixelBufferInput.append(pixelBuffer, withPresentationTime: presentationTime)

        CVPixelBufferUnlockBaseAddress(pixelBuffer, [])
    }
}

【讨论】：

【解决方案3】：

我正在使用这篇文章来记录自定义金属视图，但我遇到了一些问题。当我开始录制时，我在 iPhone 12 Pro Max 上从 60fps 提高到 ~20fps。在 Profiling 之后，减慢一切的函数是 texture.getBytes，因为它正在将缓冲区从 GPU 抓取到 CPU。

另一个问题是视频和音频不同步，不确定是否会导致这种情况。我不确定我是否应该进入信号量路线来解决这个问题，或者还有其他潜在的解决方法。

在我的例子中，纹理大小与屏幕大小一样大，因为我从相机流创建它，然后通过几个 CIFilter 对其进行处理。我不确定问题是否在于它太大，所以 getBytes 无法实时支持这种大小的纹理。

如果我需要定义优先级，我的第一要务是解决音频和视频之间的不同步问题。任何想法都会非常有帮助。

【讨论】：

@warrenm 也许你对此有一些想法
您似乎发布了一个问题作为答案；您可能应该发布自己的问题。