Inside-Out Tap Gesture Recognition

Nuka_DiLucca

I am currently implementing a system that needs to look for the tap gesture itself.
I don't want to attach a script to every single object if all it will do is return "True" if the tap gesture was performed while the object was in focus.

One example is if I tap in open space I would like a cube to appear 10 Meters away from my face where I tapped.

Another example is when I tap a cube that ONLY has a transform component attached, I want to destroy it.

All I am looking to do is Inside-Out sensing rather than Outside-In.

Thank You

stepan_stulov

Hey, @Nuka_DiLucca

I think your confusion comes from the HoloToolkit that has a bit more wiring of gestures and gaze which you seem to be trying to unwire, so to say. I'm not fully aware of how it works in the HoloToolkit so I cannot advise on that, although I believe it's possible too.

If you use pure Unity APIs for HoloLens without the HoloToolkit, you can simply use GestureRecognizer class, as I already said. Simply create an instance of it, set the list of recognizable gesture and start capturing. When a tap (or any other desired gesture) happens, your event handler will be called. That's it! No gaze, no objects, nothing else is required. An empty scene with a camera and an instance of GestureRecognizer created anywhere. You can look at empty space, at a collider or at Mona Lisa, it will behave the same:)

Tracking on HoloLens is always inside-out regardless of your application code. What you refer to is capturing gestures regardless of anything gaze related, which is what I described above. In fact gestures ARE global and require no object to be gazed at. It's just that HoloToolkit combines these features for easier application development so you can forget about gaze, gestures, etc. and focus on scene's interactive content, "tappables", "manipulatables", etc.

There should be something like "global gestures" somewhere in the HoloToolkit though.

Does this make it clearer?

Cheers

stepan_stulov

1. To register a tap gesture regardless of what you'r looking at you can simply use GestureRecognizer. Set its active gestures, subscribe to the one you need and receive the event. No need to have any objects in the scene at all.
2. To destroy the cube you're looking at you need to know that you're looking at it. This can be achieved easily by raycasting a ray from the camera and finding a collider that's hit. Physics' raycasting works with colliders only. If you for some strange reason don't want a collider then you can re-implement raycasting yourself and see whether the cube or at least its centre is in the camera's frustum. But why doing that if the collider way is much easier.
3. What do you mean by "inside out sensing"?

Nuka_DiLucca

I am trying to understand the event system a little better. If I want to look for a tap event when GazeManager.Instance.HitObject == null do I have to grab the event handler from GestureRecognizer and look to see when an event is subscribed to public event TappedEventDelegate TappedEvent;?

I'm just not understanding the process to go about receiving an empty tap.

i.e. In Unity if I tap empty space print "Hello World."

3A: When I say inside out sensing I am referring to the way tapping is recognized. Similar to VR where you have the HTC Vive with base stations that is considered outside in sensing. What I'm trying to simply phrase is the process of NOT having objects determine whether or not they have been tapped, but instead have a listener look for when an object it tapped.

i.e. An object looks for a tap gesture by using
public void OnInputClicked(InputClickedEventData eventData)
A listener looks for a tap gesture by using ????GestureRecognizer????

stepan_stulov

Hey, @Nuka_DiLucca

I think your confusion comes from the HoloToolkit that has a bit more wiring of gestures and gaze which you seem to be trying to unwire, so to say. I'm not fully aware of how it works in the HoloToolkit so I cannot advise on that, although I believe it's possible too.

If you use pure Unity APIs for HoloLens without the HoloToolkit, you can simply use GestureRecognizer class, as I already said. Simply create an instance of it, set the list of recognizable gesture and start capturing. When a tap (or any other desired gesture) happens, your event handler will be called. That's it! No gaze, no objects, nothing else is required. An empty scene with a camera and an instance of GestureRecognizer created anywhere. You can look at empty space, at a collider or at Mona Lisa, it will behave the same:)

Tracking on HoloLens is always inside-out regardless of your application code. What you refer to is capturing gestures regardless of anything gaze related, which is what I described above. In fact gestures ARE global and require no object to be gazed at. It's just that HoloToolkit combines these features for easier application development so you can forget about gaze, gestures, etc. and focus on scene's interactive content, "tappables", "manipulatables", etc.

There should be something like "global gestures" somewhere in the HoloToolkit though.

Does this make it clearer?

Cheers