The right slider is the black point. It indicates where the blackest black in the image will be — right now it’s zero but if you drag it to the left the black point will change, and what was dark gray before will become the “new” black. If you make it 5 instead of 0 then all pixels in the image whose value is 0,1,2,3,4 and 5 will all become value 0 when you press OK button. Similarly, the left slider is the white point and you can drag it to the right to make the pixels that were gray become pure white. By doing this, you’ll crop the scan’s value range, chopping off the unwanted “noise” data, and this means you are also reducing the image’s “bandwidth” in the same time. So remember that the less distance between the black and white points you leave, the fewer actual brightness levels will be left — and if you go too far you may end up with a rather coarse image. But for most scans it’s enough to cut off no more than 5% to 10% of the initial range, which isn’t much. The middle slider behaves differently: it’s the midpoint and is defined proportionally to the black and white points. You can use it to adjust the overall brightness of the image: pulling it to the right makes the scan darker, pillung it to the left makes it brighter without losing the clarity.

What we need do to our scan is to cut about 5 levels of black (to make the graphite yet darker) and cut off a lot of white — because we need to lose all the gray noise in the background. (Choose the midpoint to your liking.) Essentially you’ll cut off the whole big peak and leave only a bit of its left slope, like this.

After you press OK, the histogram of the resulting image will look like this (and typical for a clean scan of pencil sketch):
Notice the “comb” appearance of the histogram. This is precisely the result of too few actual brightness levels left after cropping. This can frequently be seen in processed images, and shouldn’t bother you too much. If it isn’t degrading the visual quality, you may safely ignore it. If it does, you should consider making a better scan (we’ll return to that yet) because your current one obviously uses too little bandwidth for meaningful data.

Continue...