Using logical reasoning, we can imagine each individual pixel RGB being read and not passing for the demosaicing process, which would result in three different values per pixel. Imagine now that an object is pure red, and so green and blue channels have signal near zero. When viewed closely, the image of this object (which was red) would be formed by a light spot (red channel) and two dark spots (green and blue channels). The result would probably be something like film grain black and white. Do not believe there would be a gain of sharpness, but one accrued granulation, which might seem more organic, or artistic.