Tips about scanning

This article about scanning is quite technical. It assumes that the reader has a basic understanding of what's involved when using a scanner and Photoshop.

For equipment I have a rather old HP scanner, a HP ScanJet IIc. It's resolution is 400 dpi (dots per inch). I use the software that was shipped with HP ScanJet 4c, Deskscan II, v 2.5. It gives very good control over the scanner: there are ways to adjust color, highlight and shadow, levels, curves. I use Adobe Photoshop 4 to work on scans. It helps to have lots of RAM in your computer, when working with big scans it really speeds things up. I have calibrated the scanner to my screen, also in Photoshop I did a calibration of my screen (File\Color Settings\Monitor Setup\Calibrate).

Overview of the steps involved

The steps involved

1. Making the actual scan

Before making the scan I have adjusted the ScanJet software for best results. So the lamp is warmed up before each scan to get consistent quality. I've made a "print-path" that tells my scanner it should use no calibration ("Screen") and that it should use 400 dpi when scanning. This gives quite huge files, I need this resolution though to be able to get rid of the moiré pattern without losing too much sharpness. B&W images can be as big as 10 MB, color images are sometimes over 30 MB.

I do my best to use the controls to get true black on one end and true white on the other end. Then it's kind of an art to use the 256 grays in between for best results. I found out it's better to do this now while making the scan, than later with Photoshop. I like the "emphasis" that works like "curves" in Photoshop: I often use it to make the shadows a bit lighter, usually to see the faces (eyes) better.

Also I found out that it is faster to scan to a file than it is to scan to Photoshop through the Twain driver and save it from there. So usually when I start with a book I scan all images directly to disk until the book is done or until I'm running almost out of disk-space. I've over 700 MB reserved for this and since I'm usually doing B&W this is sufficient for most projects.


2. Gaussian blur (Filter\Blur\Gaussian Blur)

When scanning printed material it's almost impossible to avoid getting a moiré pattern on your scans. This is because the printer (printing office) uses a silk-screen to print photos in books and magazines and this screen is scanned as well. Because this pattern is very regular and the scanner also scans in a fixed pattern, it causes interference between the scan-lines and the silk-screen. Using Gaussian blur gets rid of the moiré pattern.

The reason I scan at 400 dpi is to have enough pixels to use a blur without losing too much sharpness. On the 400-dpi scans I usually have a blur with a radius of 1.3 pixels, this depends on the book. When I decided on the best blur I use it for the whole book. Although I've heard that some scanners have built-in "de-screeners" I doubt whether the results are that good. But I don't have to worry about that, my scanner doesn't have that option.


3. Resize (Image\Image Size) and save (File\Save As)

The only reason for a resize here is available disk-space. I resize to 1500 pix high, which is about double the size of the final image. So I can still crop the white space later and I rather don't use levels on the final size since it gives "holes" in the histogram. I save as highest quality JPEG (10 Maximum), a loss-less compression which you can use if you want to save the file over and over again.

I've combined this with step #2 in a Photoshop batch-job. This is very handy, these batch-jobs! I have all scans in one folder, start the job and when I come back an hour or two later all scans are blurred and resized and saved, saves me a lot of time.


4. Crop (Image\Crop) and Levels (Image\Adjust\Levels)

I hate these white lines at the border of scans. So I make sure I get rid of these first. If needed I first rotate the picture if the image was scanned a bit tilted. I usually want as much of the picture as possible, out of respect for the photographer's work.

"Levels" is what makes an image look good in my opinion. After doing levels the blacks are really dark and the whites do shine! It's impossible for me to explain how to do this, just see for yourself. Load an image and adjust the left "Input levels" slide to where the histogram starts. This makes the darkest gray show up as black. Adjust the right slide to the left till where the histogram ends, this makes the lightest gray turn white. Now the image has 256 gradients of grays. The hardest slide to adjust is the middle one. While adjusting the two other slides is very easy and straightforward, for the middle slide you need a good eye, you can screw up an image real bad or make it look ten times better and everything in between. I only do levels when it's dark outside, with the curtains closed so the lightning in my room is always the same. I usually have a reference image open in Photoshop so I can compare what I'm doing with an image I think looks just fine. Also I have a fixed setting for my monitor.


5. Resize (Image\Image Size)

After the levels I resize the image to its final size. Then I save it and complete the images later. I do steps #4 and #5 combined in one session and usually do these steps for a complete project in one evening or two. This makes the series appear consistently done. As for the size of the images: I like 768 pix high or 1024 pix wide since that is the size of my monitor. I don't see much use in bigger images although I know people that disagree. When I cannot get a good enough quality I make the images a bit smaller, like 600 pix high or 800 pix wide. Talking about pixels: after the scanning I never ever think about resolution again, this is because I scan for screens and not for printers.


6. Sharpen (Filter\Sharpen\Unsharp Mask)

Quite a tricky part. When not done right the images seem like they are cast in bronze. I try to watch the hair and not overdo it. What the "unsharp mask" really does is to make the difference in levels between two pixels bigger. This works especially good on borders or boundaries, the contrast on these is enhanced which gives an impression of sharpness. As for the settings, there are three that I'll try to explain. All settings depend very much on the size of the images, I always work on pictures that fit 1024x768.

I place "Threshold" on 12 levels. This means that Photoshop will only sharpen when two adjacent pixels have a difference in brightness of at least 12 levels. The lower this adjustment the more sharpening will appear, but the picture will look ugly with a setting that is too low.

I place "Radius" on 0.7 pixels. The higher this setting the wider the contrast line will be on boundaries.

"Amount" depends on how sharp the image already looks. I will use in between 100% and 200% most of the time.

One of the tricks Blueboy taught me is what I call a "selective" sharpening. The result is that only on places where you need sharpening you'll have the "Unsharp mask", and you can use pretty high settings without getting too much grain in parts where you don't want it. Here's how to do it:

. Do the Unsharp Mask.
. Make a snapshot of the image (Edit\Take Snapshot).
. Undo the Unsharp Mask (Ctrl-Z).
. Use the Clone tool (clone from snapshot) to "paint" in the parts you want to get sharpened.

This is very laborious work, but it's worth the effort.


7. Retouch

I use the clone-tool for doing the retouch. I zoom in to 400% and use a soft brush with a hardness of 0 and a diameter of 9 most of the time. At this point the image is already looking pretty good with all the work that has been done, but this step makes it really look good. I get rid of all the spots, scratches and hairs and hardly ever manage to do more than 10 on an evening. I turn up the brightness of my monitor a bit so that I can see all faults in the image.


8. Selective blur

When I see that a picture is very grainy I use the technique as described in step #6 to selectively blur the image. I usually first take a snapshot, then do a blur of .3 pix, then use the clone tool (clone from snapshot) to restore the sharpness where I want it.


9. Final save as

The last step is to save the image in a more compressed file. I save as "6 quality high" JPEG's, when the files are too big I save as "5 quality high"


Home