求助：如何为自定义相机添加人像拍摄模式？AVPhotoCapture开启深度无效果

Hey there! Just turning on image depth in AVPhotoCapture is only the first piece of the puzzle—there are several critical steps you’re probably missing to get that iPhone X-style portrait blur working. Let’s break this down step by step:

1. First, Confirm Your Device Supports Portrait Mode #

Not all iOS devices can capture portrait effects. You need a device with a dual-camera system (or LiDAR) that supports depth data and portrait matte delivery. Check this upfront with code:

guard let device = AVCaptureDevice.default(.builtInWideAngleCamera, for: .video, position: .back) else { return }
if device.isDepthDataDeliverySupported && device.isPortraitEffectsMatteDeliverySupported {
    // Your device can handle portrait mode
} else {
    // Portrait mode isn't supported here
}

Note: iPhone 7 Plus and later (excluding SE models), iPad Pro 11-inch (2nd gen+) and 12.9-inch (4th gen+) support this.

2. Enable Depth Data Delivery in Your Capture Session #

You can’t just toggle depth in capture settings—you need to enable it at the session level first. When configuring your AVCaptureSession, make sure to:

• Set depthDataDeliveryEnabled to true on your capture device input
• Select a device format that supports both video and depth data

Here’s a snippet to set this up:

let session = AVCaptureSession()
session.sessionPreset = .photo

guard let device = AVCaptureDevice.default(.builtInWideAngleCamera, for: .video, position: .back) else { return }
do {
    let input = try AVCaptureDeviceInput(device: device)
    if session.canAddInput(input) {
        session.addInput(input)
    }

    // Find a format that supports depth data
    let availableFormats = device.formats.filter { format in
        let supportsDepth = format.isDepthDataDeliverySupported
        let supportsVideo = format.mediaType == .video
        return supportsDepth && supportsVideo
    }
    // Pick the highest resolution format (adjust as needed)
    if let bestFormat = availableFormats.sorted(by: { $0.highResolutionStillImageDimensions.width > $1.highResolutionStillImageDimensions.width }).first {
        try device.lockForConfiguration()
        device.activeFormat = bestFormat
        device.depthDataDeliveryEnabled = true
        device.unlockForConfiguration()
    }
} catch {
    print("Error configuring device: \(error)")
}

3. Configure Capture Settings Properly #

When setting up AVPhotoCaptureSettings, don’t forget to enable both depth data and portrait matte delivery—this is what lets you generate the blur effect:

let settings = AVCapturePhotoSettings()
settings.isDepthDataDeliveryEnabled = true
settings.isPortraitEffectsMatteDeliveryEnabled = true
// Prioritize quality to ensure depth data isn't skipped
settings.photoQualityPrioritization = .quality

// Trigger capture
photoOutput.capturePhoto(with: settings, delegate: self)

4. Process the Captured Data to Generate Portrait Blur #

The system doesn’t automatically apply the blur for you—you need to use the depth data and portrait matte with Core Image to create the effect. In your AVCapturePhotoOutputDelegate:

func photoOutput(_ output: AVCapturePhotoOutput, didFinishProcessingPhoto photo: AVCapturePhoto, error: Error?) {
    guard error == nil else {
        print("Capture error: \(error!)")
        return
    }

    // Get the original image, depth data, and portrait matte
    guard let imageData = photo.fileDataRepresentation(),
          let originalImage = UIImage(data: imageData),
          let depthData = photo.depthData,
          let portraitMatte = photo.portraitEffectsMatte else {
        print("Missing required data for portrait effect")
        return
    }

    // Convert to CIImage for processing
    let ciImage = CIImage(image: originalImage)!
    let depthCIImage = depthData.depthMap as CIImage
    let matteCIImage = portraitMatte.mattingImage as CIImage

    // Use CIPortraitEffectFilter to create the blur
    let filter = CIFilter(name: "CIPortraitEffectFilter")!
    filter.setValue(ciImage, forKey: kCIInputImageKey)
    filter.setValue(depthCIImage, forKey: kCIInputDepthDataKey)
    filter.setValue(matteCIImage, forKey: kCIInputPortraitEffectsMatteKey)
    // Adjust blur intensity (0.0 to 1.0)
    filter.setValue(0.7, forKey: kCIInputBlurRadiusKey)

    // Render the filtered image
    guard let filteredImage = filter.outputImage else { return }
    let context = CIContext(options: [.useSoftwareRenderer: false])
    if let cgImage = context.createCGImage(filteredImage, from: filteredImage.extent) {
        let portraitImage = UIImage(cgImage: cgImage)
        // Use portraitImage however you need (display, save, etc.)
    }
}

5. Check Environmental Conditions #

Even if all code is correct, portrait mode might fail if:

• The subject is too close or too far (most devices work best between 2–8 feet)
• Lighting is too dim (depth sensing relies on good light)
• The subject isn’t clearly separated from the background

Make sure you’re testing on a physical device (simulators don’t support depth capture) and that you’ve requested the appropriate privacy permissions (NSCameraUsageDescription in Info.plist).

内容的提问来源于stack exchange，提问作者Singh