Exactly. Suppose the color channel is red, what digital algorithm possibly guess the values for the green and blue channels. Lets do some guessing. Eight bits coukd be split into two for each color channel and two for black level. So the converter or some software opens the file and by some algorithm (mathematical procedure) creates two new bits for every one bit in the file. So how do we know that algorithm faithfully reproduces what we saw in the view finder?