A few things :)

Firstly - we refer to the current in-market devices as "Windows motion controllers". In terms of naming convention, I would suggest 'motion controllers'.

It is unfortunately not possible to do an exact string match on device ID - the last section (045E-065B) will vary based on the specific connected device. As an example: if you use the Simulator, the controllers will actually come through as:

Spatial Controller (Spatial Interaction Source) 00FF-FF00

We recommend doing a string startsWith test for: 'Spatial Controller (Spatial Interaction Source) ' to determine that the device is a motion controller to choose button mappings.

The next piece of the puzzle is support for the devices pointing ray direction and position offset. You may notice that the device pose does not accurately reflect where you are pointing (this is similar to Oculus Touch controllers). You can really see this if you compare the pointing ray from the VRController to that in the Mixed Reality cliffhouse.

To make matters a bit more complicated - the ray offset and direction actually vary depending on what kind of motion controller is plugged in connected - which is discernable by looking at the code portion of the device ID. As an example, you can use the following regex to get the device code:

/([0-9a-zA-Z]+-[0-9a-zA-Z]+)$/

There are currently 2 distinct devices supported by Windows Mixed Reality, each having a slightly different form factor and pointing direction. (They have the same button mappings). GLB (binary glTF) files of these two devices are available at this fork of the a-frame assets repository:
https://github.com/leweaver/assets/tree/msft/new-controller-model/controllers/microsoft

The pointing pose offset can actually be found in those glTF files, on the node named POINTING_POSE. That node is transformed such that it faces outward in the direction of the pointing pose. It is a local transform, so an attached line would be oriented correctly (ie, it relies on the transform accumulation of the parent nodes).

Happy to provide more info!
Lewis

I know this is not specifically mentioned as supported, but I tried out the GearVR Controller with a GearVR and a Samsung S6.

Trying the demo page on this project, I got the following results:

• Using the new(ish) built in Internet browser on Oculus home - the controller is visible and tracks/points well, but none of the buttons work - I can't interact with the UI elemental via clicking.
• Using Samsung Internet with WebVR enabled, I get a black screen when loading the page.

Let me know if there is any useful debugging info I can provide for this controller.

Cheers! - James

Can only see them when I stretch my arms out 100% on oculus.

Xbox ONE controller returns with a null gamepad.pose field. This is when running the demonstration link. The axis and button fields seem valid.

Is this a known limitation?

Details:
Windows 10
id"Xbox 360 Controller (XInput STANDARD GAMEPAD)"
Google Chrome | 65.0.3325.181 (Official Build) (64-bit) (cohort: Stable)
Chrome flags: chrome://flags/#enable-webvr, chrome://flags/#enable-gamepad-extensions, chrome://flags/#openvr
No VR display hardware attached

I am going to try Firefox.

When we rotate our head, we expect controller is static (like native application).

Hi. First of all thanks for the VRController. So far it's working great.

Having a big problem with the mesh attached to the controller object, as it is not travelling with the controller object. I am not a threejs expert so I might be missing something obvious :-/

I am attaching (your) mesh to my controller as per the example...

 controllerMesh.add(handleMesh);
 controller.add(controllerMesh);

I am attempting to shuttle my camera about using thumbsticks by adjusting the scene position...

 controller.addEventListener('thumbstick axes changed', (event) => {
      this.scene.position.sub(new Vector3(event.axes[0], 0, event.axes[1])); 
 })

For the most part this is working great. Unfortunately the controllerMesh stays put.
I have examined my controller position and it does not change (the scene does). The controller mesh is a child of the controller, so I don't know what I could be missing.
Thanks.

Edit....
I should add that I initially attempted to update the controller object position within the scene and could not seem to update it. After looking at your "space rocks" example it appeared to me that you were moving the "world" instead of the controller/camera.

I have setup a basic example of using the controller with a RayCaster with a laser line pointer to click on intersected objects.

The intersect point is either above or below the object depending what version of three.js is used. I have integrated the standing matrix fix in r90. The intersect point is now above the object.

When testing against the daydream controller example in three.js the intersect is exact. There seems to be some offsetting and not sure how to remove it perhaps ?

http://dev.electroteque.org/vrcontroller/raycast.html
http://dev.electroteque.org/vrcontroller/daydream.html

When you user a Daydream viewer with a controller, camera is on 0,0,0 perhaps is related with this three.js issue > mrdoob/three.js#11619

I can't seem to find an example here to get the controller clicking objects with a raycaster. It for some reason is using Dat gui only.

Any ideas ?

I upgraded threejs to r89, the controller will not work. May I ask you to see this issue? Thank you so much!

Any interest in making this awesome tool available as an npm package?

I am not sure what the story is with Occulus. But the Gamepad instance changes when exiting then re-entering presentation mode. The timestamp is different which can be detected.

A new gamepad instance is required to be updated. I have tried to modify myself but it's not detecting button changes when updating to it yet. I haven't uploaded my tests with a working example yet.

Something like this could detect the change and either update the gamepad or disconnect the controller.

if (this.controllers[ i ].gamepad.timestamp !== gamepad.timestamp) {

In GearVR / Daydream taking off the headset does not dispatch a deactivate event to possible manage gamepad disconnection that way. Only exit presentation events would work.

I guess it has to be disconnected and reconnected ?

Hello,

I wanna first thank you for the hard work you have put in this project this is truly useful and contribute a lot to the development of VR.
Anyway I'm looking for a way to scale the user according to the scene my character is way too little compared to the model I draw on my scene.
I have achieved to scaled up the controller (hand) but neither my size or camera.
Is their a way I could do this?
Thank you.

I needed access to original isPressed boolean in order to fly around the scene with the mouse press. So I have to add this in order to expose it.

if ( button.isPressed && isPrimary ) {
    controller.dispatchEvent({ type: 'primary pressed' })
}

I don't quite understand why it was not exposed original. It is such a basic thing to have.

I can make an PR for that if necessary.

Hey, great repo - any development with NOLO VR controllers? through SteamVR or standalone?