I have a screen-space canvas where I place all of my gestures. I have a transform gesture (translation only, projection type layer) and a screen transform gesture on this canvas. Then I have a FullscreenLayer on the canvas as well.
If I set the FullScreenLayer to Global in the editor, my translation gesture stops working, but the screen transform gesture works.
If FullscreenLayer is set to Main Camera, both gestures work. This setup works fine when I have one camera, but if I have two cameras (Each taking up 50% of the view rect), input only works on the first one. This is expected because Main Camera is only one of the cameras.
Finally, if I set FullscreenLayer to Main Camera in the editor, and then in a script on Start() set FullscreenLayer to Global, both of my gestures work on both screens. This is what I want, but the way to get there isn't obvious, nor do I know if that's how it's supposed to behave or if I'm exploiting a bug. So, this is working for me but thought I'd submit this strange behavior for your analysis.