2D screen point to 3D world point

I am trying to achieve converting a click on the screen to coordinates in the world on the plane x.y=0 (ie z=0, composed of the x and y axis).

I have found code to convert 3D -> 2D (https://github.com/enthought/mayavi/blob/master/examples/mayavi/advanced_visualization/mlab_3D_to_2D.py), however I am unsure how to go about reversing this as I have not covered those topics at university yet. I have attached a screenshot of my setup so far. It is basically a surface at z=0 on which you will be able to drag a 2D actor. For this, I need to find the point on the surface which was clicked and then handle collision detection using bounds of the 2D actor on the surface.

So far, I am able to get the point clicked in the scene window via:
```
def press_callback(vtk_obj, event):
x, y = vtk_obj.GetEventPosition()
...

self.scene.interactor.add_observer('LeftButtonPressEvent', press_callback)
```

This gives the point from the left-bottom point of the scene UI pane/canvas.

Would it be possible to transform this point to a point on the surface (gray)?


I have tried using a picker but I can't get the correct bounds to be picked when clicking, when I click on the red shape it is not picked, only when I click within the bounds of the black square I drew on in photoshop does the picker pick the shape. I have checked the `picker.pick_position`, and `picker.actor.bounds`, and they match up. So it seems to me like the `pick_position` is not correctly transformed to the world.

However, as you can see by the two white points plotted, the bounds of the red actor/shape are _outside_ the rectangle which is being picked up by the picker. This rectangle of detection bounds scales when you move in and out.

This is how I am calling `picker.pick`:

```
def press_callback(vtk_obj, event):
x, y = vtk_obj.GetEventPosition() # screen point in scene UI pane

result = picker.pick(x, y, 0, self.scene.scene.renderer)
print(picker.pick_position)
if result == 0: # successful pick
return
pickPosition = picker.pick_position
print([o for o in picker.actors])
print("pick", pickPosition, [picker.actor], picker.actor.bounds, picker.actor.position)
if picker.actor == sphere_actor:
print("Bingo")
```

Should I be supplying the `(x, y)` of the click relative to the left-bottom corner of the scene UI?

Any ideas how to progress? I am trying to achieve dragging functionality of the red 2d shape along the surface at z=0. I am just blocked on getting the `Screen Point -> World Point @ surface(z=0)` transformation as I can do the boundary detection myself, no need for the picker. Many thanks!