Chapter 24: Detecting Images & 3D Objects

前言

之前的章節有做過辨識某個特定的2D圖片後，呈現相關AR內容。在新的ARKit 2.0中新增可辨識特定的3D物件另外可以對某個2D圖片進行追蹤。

大綱

• Image Tracking

  • Detecting a card

  • Highlighting a card

  • Tracking a card

  • A word on tracking configuration

  • Tracking a card, version 2.0

• Detecting 3D Objects

  • Creating a reference model

  • A brief overview of ARDock

  • Preparing the ARKit session

  • Highlighting the detected object

  • A few notes about quality and accuracy

Image Tracking

• Image detection和Image tracking的不同

  • Image detection is a static process that locates the reference image in the real world through the device’s camera.

  • Image tracking, on the other hand, is a dynamic process that begins immediately after an image is detected. It tracks the location of the reference image if it change

• ARKit 2.0 提供新的 configuration - ARImageTrackingConfiguration.

  • 相比ARWorldTrackingConfiguration最大的優勢less expensive in terms of processing power — CPUs and batteries。

Detecting a card

這節內容之前章節已經有做過了。偵測提供的內建的特殊圖片。

    // World tracking
    let configuration = ARWorldTrackingConfiguration()
    // 當ARKit偵測到這些圖片時，會透過ARSCNViewDelegate中的renderer(_:nodeFor:)回傳ARImageAnchor
    let triggerImages = ARReferenceImage.referenceImages(inGroupNamed: "deck", bundle: nil)!
    configuration.detectionImages = triggerImages

    sceneView.session.run(configuration)

////////////////////////////

extension DeckViewController : ARSCNViewDelegate {
  func renderer(_ renderer: SCNSceneRenderer, nodeFor anchor: ARAnchor) -> SCNNode? {

    if let imageAnchor = anchor as? ARImageAnchor {
      let overlayNode = createCardOverlayNode(for: imageAnchor)
      return overlayNode
    }

    return nil
  }
}

Highlighting a card

• 再偵測到撲克牌上覆蓋一個AR紅色透明平面。

extension DeckViewController {

  func createCardOverlayNode(for anchor: ARImageAnchor) -> SCNNode {
    let box = SCNBox(
      width: anchor.referenceImage.physicalSize.width,
      height: 0.0001,
      length: anchor.referenceImage.physicalSize.height,
      chamferRadius: 0)

    if let material = box.firstMaterial {
      material.diffuse.contents = UIColor.red
      material.transparency = 0.3
    }

    return SCNNode(geometry: box)
  }

A word on tracking configuration

• New tracking configuration ARImageTrackingConfiguration. This is specifically designed to detect and track 2D images without the burden of tracking a full 3D world.

  • It uses less CPU

  • It does it better: In world tracking, injected content is still tracked and displayed even if the detected image is no longer visible in the camera — you experienced that a few moments ago.

  • It’s more precise

  • It works in moving environments

Tracking a card, version 2.0

• 切換成新的configuration

-    // World tracking
-    let configuration = ARWorldTrackingConfiguration()
+    // Image tracking
+    let configuration = ARImageTrackingConfiguration()
     // 當ARKit偵測到這些圖片時，會透過ARSCNViewDelegate中的renderer(_:nodeFor:)回傳ARImageAnchor
     let triggerImages = ARReferenceImage.referenceImages(inGroupNamed: "deck", bundle: nil)!
-    configuration.detectionImages = triggerImages
+    configuration.trackingImages = triggerImages

Creating a reference model

就像2D要提供reference image, 3D就要提供reference object, 那要如何建立reference object，Apple在WWDC 2018有提供一個範例可以讓我們產生想要的reference object。ARKit Scanner。

Preparing the ARKit session

    // World tracking
    let configuration = ARWorldTrackingConfiguration()
    let models = ARReferenceObject.referenceObjects(inGroupNamed: "models", bundle: nil)!
    configuration.detectionObjects = models

    sceneView.session.run(configuration)

extension RubikViewController : ARSCNViewDelegate {
  func renderer(_ renderer: SCNSceneRenderer, nodeFor anchor: ARAnchor) -> SCNNode? {
    if let anchor = anchor as? ARObjectAnchor {
      print("Found 3D object: \(anchor.referenceObject.name!)")
    }
    return nil
  }
}