This wiki page covers how to scan line art.
Note: Before making any edits to a scanned image, sometimes called a
raw image, it is recommended to work on a copy and not on the
original. This way if things go wrong, a re-scan isn't
If the black and white art was scanned as colour, then convert it to
grayscale. (There is no reason to set your scanner to grayscale, since
its sensor is colour by default – and Photoshop does a better job at
Select Arbitrary from the Image Rotation menu and press OK
(keeps the image as background layer) – or press Straighten Layer in
the context menu (converts background to a layer). Both will achieve
basically the same for our needs:
The goal here is to get the outer borders and an image aspect ratio that
are the same for each page. For that we're going to use a fixed aspect
To set the overall contrast level of the page we create a curves
'adjustment layer' by clicking on the icon at the bottom of the layers
toolbox. Its controls should pop up automatically – if they don't just
double click the curves icon on the newly made adjustment layer. We do
this in a control layer instead of in the image itself, so we can reset
and start anew if the result is unsatisfactory.
First we set the black point of the image. Choose the uppermost of the
three eyedropper tools in the pop-up. Set the page magnification to
around 400%, so you can easily pick a really dark pixel and sample it
with the eyedropper. Go back to 100% or smaller magnification and check
if the blacks look good now. Click the eye icon in the pop-up to
enable/disable preview. If dense line material, e.g. cross-hatching,
looks too dark, choose an even darker pixel with the eyedropper tool.
Next using the bottom most of the three eyedropper tools, we set the
white point of the image. Do not choose the brightest white pixel, but
an average, faintly grayish pixel. We want to get rid of paper
imperfections, pencil traces, and smears – and there is no artwork in
the highlights of the page to worry about.
Finally we do a 'shadow boost' on the curve itself. Create a new edit
point in the middle of the diagonal line and another one in the first
quarter from top. Gently nudge the latter edit point a bit to the right
and down, so the dark levels get a bit lighter – without influencing the
true black level or the white levels of the page.
From now on we're using 'destructive' editing techniques, so we
duplicate the image layer and continue working on the copy layer,
keeping the original background layer untouched. Right-click the
background and select duplicate:
Now that we have a good overall contrast, we're going to do a quick and
dirty removal of pencil traces and other leftover smudges. The Magic
Wand is our friend for this job. Make sure you're on the correct
While there's only two screenshots for this chapter, it's actually the
most time-consuming part of the whole procedure. We use a white brush to
clean all the specks and dots that haven't been caught by earlier
editing steps. Use 100% flow and opacity and a soft brush border. Work
at whatever magnification you find comfortable. Something between 100
and 400% should do nicely:
Conversely, if you have inked areas which are not dark enough, do the same, but with the Burn tool (set to shadows).
See this wiki page for more information on how to backup files.
First we need to resize the image. Usually we go by standard monitor resolutions. Nowadays a width of 1600px (portrait) or 2560px (landscape) seems reasonable. It should result in perfect readability, yet still manageable download sizes. Make sure Constrain Proportions is checked. For portrait images use a width of 1600px. For landscape images use a width of 2560px. The height will be calculated automatically.
For the resampling algorithm, choose one of the bicubic filters. For
downsizing images usually Bicubic automatic or Bicubic sharper
works best. In the end it depends on your taste, so do a bit of trial
and error and stick with what you like:
Next the goal is to produce high quality images with file sizes as small as possible.
Choose PNG8 with the Perceptual algorithm. In the Colors field, choose a number of colours that's high enough so the Optimized image looks as good as the original. Usually, black and white line art needs 8 to 16 colours to be indistinguishable from the original. Images with a few colours 16 to 32 should work. The maximum colours of 256 is most often a waste of resources.
Make sure to look at the preview window at 100% magnification and compare the Original to the Optimized file.
We can ignore all other export settings since we already resized the image
See here for a way to automate batch exporting of PNG files.