Why does ARKit MDLTransform randomly return 0 from camera transform?

Question

I'm very new to ARKit (2 days old to be precise) and I'm trying to place a node on my ARSCNView when tapping and I am able to successfully do so 50% of the time. The other half of the time, nothing appears on screen. I've narrowed it down to the following line of code:

func getCameraTransform(for camera: ARCamera) -> MDLTransform {
        return MDLTransform(matrix: camera.transform)
    }

When nothing is appearing on screen, the above function is returning a transform.translation of float3(0.0, 0.0, 0.0)

I've checked and the camera.transform is providing a value that looks like simd_float4x4([[0.98134, 0.0198354, -0.191256, 0.0)], [-0.0129627, 0.999227, 0.0371193, 0.0)], [0.191844, -0.0339474, 0.980838, 0.0)], [-0.482363, -0.00405288, -0.768801, 1.0)]]) transform.translation is 0.

Why does this return 0 sometimes but work perfectly fine other times I tap?

Here is the full code:

override func touchesBegan(_ touches: Set<UITouch>, with event: UIEvent?) {
        guard touches.first!.tapCount == 1 else { return }
        let camera = sceneView.session.currentFrame!.camera

    let transform = getCameraTransform(for: camera)
    let boxGeometry = SCNBox(width: 0.1, height: 0.1, length: 0.1, chamferRadius: 0)
    let cube = SCNNode(geometry: boxGeometry)
// SOMETIMES TRANSFORM.TRANSLATION IS JUST 0s
        let position = SCNVector3(
            transform.translation.x,
            transform.translation.y,
            transform.translation.z
            )
        cube.position = position
        sceneView.scene.rootNode.addChildNode(cube)
    }

func getCameraTransform(for camera: ARCamera) -> MDLTransform {
        return MDLTransform(matrix: camera.transform) // this is returning the 0 translation every other time
    }

score 0 · Accepted Answer · answered Mar 22 '18 at 23:57

I think it may be better to look at a different way to approach the issue. I haven't seen that example before and from having a test, it doesn't seem to be what you need.

As such, I will provide you with a few examples which should help you get started.

(1) Place An Object At A Certain Position Based On The Current Frame Of The Camera:

/// Adds An SCNNode 3m Away From The Current Frame Of The Camera
func addNodeInFrontOfCamera(){

 //1. Get The Current Frame Of The Camera
 guard let currentTransform = augmentedRealitySession.currentFrame?.camera.transform else { return }

 //2. Create An SCNNode
 let nodeToAdd = SCNNode()
 let boxGeometry = SCNBox(width: 0.1, height: 0.1, length: 0.1, chamferRadius: 0)
 boxGeometry.firstMaterial?.diffuse.contents = UIColor.red
 nodeToAdd.geometry = boxGeometry

 // Create A Transform With A Translation 1m In Front Of The Camera
 var translation = matrix_identity_float4x4

 //Change The X Value
 translation.columns.3.x = 0

 //Change The Y Value
 translation.columns.3.y = 0

 //Change The Z Value
 translation.columns.3.z = -1

nodeToAdd.simdTransform = matrix_multiply(currentTransform, translation)
augmentedRealityView?.scene.rootNode.addChildNode(nodeToAdd)

}

(2) Place an SCNNode Based On Any Available featurePoints In The Scene:

override func touchesBegan(_ touches: Set<UITouch>, with event: UIEvent?) {

    //1. Get The Current Touch Point
    guard let currentTouchPoint = touches.first,
        //2. Perform A Hit Test Against Any Feature Points
        let result = augmentedRealityView.hitTest(currentTouchPoint.location(in: augmentedRealityView), types: .featurePoint).last else { return }

    //3. Get The World Coordinates Of The Result
    let transform = SCNMatrix4(result.worldTransform)

    //4. Get The Positional Data From The Transform
    let vectorFromTransform = SCNVector3(transform.m41, transform.m42, transform.m43)

    //5. Create An SCNBox At The Point
    let boxNode = SCNNode()
    let boxGeometry = SCNBox(width: 0.1, height: 0.1, length: 0.1, chamferRadius: 0)
    boxNode.geometry = boxGeometry
    boxNode.position = vectorFromTransform
    augmentedRealityView.scene.rootNode.addChildNode(boxNode)

}

Apologies if it doesn't explain the results your are experiencing, but nonetheless, I hope it helps...

This is really helpful! How can the code be modified for the box to be placed where the user taps the screen regardless of feature points? For example, if I tap the screen while looking at a blank white wall, I'd like the box to have the X and Y coordinates of the tap, and then a hard-coded Z value of -0.8 for example. Is this possible? — Alan Scarpa, Mar 28 '18 at 01:55
Alan please check this: https://stackoverflow.com/questions/25150737/how-to-use-ios-swift-scenekit-scnscenerenderer-unprojectpoint-properly & Look At Crashalot's Answer :) — BlackMirrorz, Mar 28 '18 at 03:06

Why does ARKit MDLTransform randomly return 0 from camera transform?

1 Answers1