It only recognizances a 3-D image.
it doesn't even work with a 3D image. It has to be a real face. They experimented with 3D imaging, no go. 3D imaging is still a 2D flat surface simulating a 3D object. Apple's implementation works with a specific wavelength of infrared light projected from an LED on the iPhone which maps dots onto the user's face, a sensor then measures the distances of those dots and compares them to a stored, encrypted map of the user's face. If it matches, and the user is paying attention in a species way, it unlocks. If not, it won't. It has to be intentional.