quick'n'dirty solve for transparentImage[ alpha channels ] => image models

this allows for focused depth masking of image-to-image

why pay extra for complicated models to do this for you?

uses photon to flip depth mask pixels to black based on transparent reference image