How to merge two images at runtime?

What I want to do is cull the difference between two images and prioritize the first...

Image A is a picture of something. Say, a kitchen. Image B is the same picture, only there's 'items' around the kitchen. More than one easily.

I want to be able to control what players can see in image B based on what they have taken and what they havent. For example, if there are 2 items visible, and I did the work manually (which I dont want), I would have to render out 4 individual images (A, B, B(item 1 gone, item 2 there), B(item 1 there, item 2 gone)). Doing the work manually means the moment I add any more items the workload increases exponentially. I need some kind of automated solution that can cull the difference in the two images back down to image A.

Ideally I want the workflow to decrease to needing only two images regardless of the amount of items in the scene (of course nothing can overlap), and a script handle the difference on it's own. Is what I'm asking make sense?