Missing CollisionObject input_event integration.

[Splizard]

I've found this project incredibly helpful for building the mobile/touch camera input for my 3D project. I wanted to use the SingleTouch/SingleTap events for selecting objects in the scene howevever the events don't appear to propogate through to the CollisionObject input_event signal.

https://docs.godotengine.org/en/stable/classes/class_collisionobject.html

I've had a look through the Readme here and also at the Godot documentation, there doesn't appear to be a simple way to enable this.

Here's two ideas on how this could be achieved:

1. emit a raycast from the primary camera whenever these events are triggered and then call _input_event on the collider.
2. provide a helper function for converting the Godot touch events inside of a _input_event handler (leveraging the existing raycasts), for example:

func _on_input(camera, event, position, normal, shape_idx):
    var gestures = InputGestureEvents(event)
    for gesture in gestures:
        // handle events

[Federico-Ciuffardi]

If I understand correctly this is similar to what happens with the buttons discussed in PR #19. I think this is something that should be fixed in the Godot engine itself. But for now I like your idea of using a raycast. I don't have much experience developing 3D games in godot but I put together something basic that should work, if you want to try it add the following code to the _emit method in InputManager.gd:

if "position" in val:
        var camera = get_viewport().get_camera()
        var from = camera.project_ray_origin(val.position)
        var to = from + camera.project_ray_normal(val.position) * (camera.far)
        var root = get_tree().get_root()
        if root is Node:
                var inter = root.get_world().direct_space_state.intersect_ray(from,to)
                if !inter.empty():
                        inter["collider"]._input_event(camera, val, inter["position"], inter["normal"], inter["shape"])

Also here is an example project I used for testing: example.zip

Some comments about the code:

• The ray must have a length. I used camera.far because I assume that is a reasonable length.
• I check if the root of the tree is a Node to check if the scene is 3D before trying to get the direct_space_state but I don't know if there is a better way to do it (there probably is).
• The code can also be placed in the _input or _unhandled_input method of another node where it must only be executed if the processed event is non-native. This has the advantage of not being specific to GDTIM custom events.

[JeffBusch]

I am running into this issue in my 2D project. I planned on using the SingleTouch/SingleTap events on an Area2D input_event() instead of using the button nodes (mouse emulation) since my project is mobile only. I'm assuming the raycast solution is only relevant in a 3D project?

Maybe I'm approaching this incorrectly but my goal was just to make my own buttons that use the GTIM events.

[Federico-Ciuffardi]

Mouse emulation and button nodes should work even if your project is mobile only, but maybe you have another reason.

The raycast solution, as far as I know, is not useful in 2D, but in 2D it's not that hard to do it manually. I mean you need the coordinates of the 4 vertices that define the Area2D (assuming it's a rectangle) and then with that you can check if the position of the event is inside the Area2D. I'm not sure if it makes a difference, but you should also be able to do this kind of manual detection with the button nodes, in fact all you need is 4 dots describing a rectangle.