Leaving your obvious reading comprehension problem aside for a moment...
If you feed the values of a pair of pixels to the running difference program/script, I predict the program/script will add a particular value, probably 50% gray in the most common applications of the process, to the value of one pixel and subtract the value of the other pixel from the result. It will then generate a pixel of that resulting value and place it in the location on the output graph that corresponds to the location of the pixels taken from the input pixels as taken from the source images.
There was my quantitative prediction. Another prediction I'll make here is that you won't recognize it as a prediction, although all the other participants in the discussion will, and you'll lie again and say I haven't provided it.