Perhaps it will use some form of code/time division so that the image is never displayed in full but is displayed in such a way as to exploit human vision idiosyncrasies (persistence of vision, inversion, colour saturation). You flash a couple of different saturated images over a small time period in order to give the impression of the image required, simple screencaps would fail [like how CRT uses the phosphor and the speed of the electron beam scanning to give the impression of a single image; and like rasterisation builds an image in two parts - but instead building it in many parts].

Not undefeatable, of course.

/spitballing