The Right Way to Scan

Maximize information for a great restoration or print copies

By Ctein Back to

Ctein_JF_2008_1

A proper scan is the first and most important step toward a good photo restoration (or copy of a print for which the negative is lost). A scan that faithfully reproduces a faded photograph is rarely useful.The upper photo in Figure 1 is a “normal” 8-bit scan from a faded black-and-white print.This is not a good basis for a restoration.The upper histogram in Figure 2 shows why. Only half the total range of values available is actually being used in this scan. A good photograph almost always has a full range of tones from black to white (or nearly so). A good scan of a photograph (whether it’s deteriorated or not) should span the range of values from near-black to near-white.

A high-quality scan’s histogram will look like the middle one in Figure 3; you don’t want it to look like the top one (which makes poor use of the range of available values) or bottom one (which loses some of the tones near white and black). The bottom photo in Figure 1 is from a corrected scan that produces an image that had a much more complete and neutral range of tones. I could expand the tonal range of the normal scan in Photoshop to produce similar contrast and color, but if I do that I get one of those “picket-fence” histograms (Figure 2, bottom). There are many gaps in the tonal scale that will show up as discontinuous-tone steps in the print (Figure 4, left). Compare that with Figure 4, right, enlarged from my corrected scan.

A few gaps in the histograms really won’t be visible in a print; this example is extreme so that I can make it clear that many gaps are not a good thing. You can have a pretty ratty-looking histogram and still see excellent tonality in the final print. Unless I’ll be expanding or compressing the tonal scale by more than 25% when I work on the file, I don’t worry too much about gaps. You can ensure good tonality in two ways. The first is to use the levels and curves controls in the scanner software to produce a good range of data in the scan. All scanners collect extra value levels internally, and they are available to fill in any gaps in the histogram when you expand or compress parts of the tonal range. Levels adjustments are easy; just bracket the range of tones in the histogram with the black and white sliders, as in Figure 5. Allow yourself some safety margin. Try to keep the darkest pixels in the scan at values of 10 to 20 and the lightest pixels around 240. That way you’ll avoid accidentally clipping the highlights or shadows.

The same principles apply to scanning black-and-white and color prints and slides (but not color negatives). Strive for a scan that makes good use of the full range of values in each channel. Frequently this will get you close to good color and tone (Figure 6). If the resulting image has an overall color cast, you can correct much of that in the scan by shifting the midtone sliders in each channel’s levels adjustment until the average color looks good. This scan originally had a blue color cast. I adjusted the midpoint for the Blue channel to make the scan more yellow.

 

16-bits—more than twice as good

Even when the scan has a very narrow tonal range, a 16-bit scan captures enough gray levels to avoid the picket-fence histogram when the tonal range is expanded.

Ideally you should adjust the scanner levels settings and do a 16-bit scan, but if in 16-bit mode, you can be a little sloppy about the scanner settings. I scan batches of photographs in 16-bit mode using the same scanner settings.

With extra bits of tonal depth to play with, there are plenty of data for me to fine-tune the photo later, as in Figure 7. The left-hand photograph and histograms come from a straight 16-bit scan with no corrections. Less than half of the range of possible tones appears in the photo; the spikes at the right side of the histograms are just from the white paper border. The curves settings shown in Figure 8 produce the much better looking photograph on the right side of Figure 8. These are very extreme corrections. Even so, I’m assured good, continuous-tone quality in the finished restoration.

Sometimes 16-bit data are a must. When the condition of the original is so uneven that no overall set of corrections can produce good results, you’ll need those extra bits. The severely degraded glass plate in Figure 9 is almost entirely bleached out, and the scanned density isn’t anything close to even. Until I can manipulate the file into a uniform-looking image, I need to work in 16 bits so that I can apply different strong corrections on various parts of the scan and still have well-filled histograms of data.

Because 16-bit files are twice as big, computations take twice as long—much longer if you run out of RAM and start swapping scratch files to disk. It’s something to keep in mind. It’ll really slow you down if doing a 16-bit scan results in your image-processing program running to the hard drive every time you perform an operation on the file. If your computer really isn’t up to working on 16-bit files for the whole restoration process, compromise. Do as much of the color and tone correction as you can in 16-bit mode. Then convert the file to 8-bit mode. That minimizes the image-quality problems of working with 8-bit files. The closer you get to the finished restoration, the less having 16-bit data matters.

Scanning color film

Scanning old color negative film is more of a challenge; the “white-black” rule I gave at the start of this article doesn’t work as well. The most successful approach is to capture every bit of information you can from the negative and massage it on the computer. Most film scanners should have no trouble capturing a good range of densities in a negative. Because a negative’s contrast is naturally low, it gets boosted substantially when it’s printed, even more so if the negative is faded. So do your scans in 16-bit mode; if you start with a mere 8-bit scan, you run the risk of winding up with a picket-fence histogram. If your scanner has it, 16-bit linear mode is best; it passes on the data with the minimum amount of software interpretation.

Use Slide setting

Most scanner software assumes you want a positive image of normal contrast, and doesn’t capture the full density range of the negative. You want a scan that contains as much information from the negative as possible. Since scanner software is designed to retain a greater density range when scanning slides, I scan negatives as if they were slides. This pulls the maximum density range from the film.

The image you see on your screen will then, of course, be a negative, but a simple inversion operation will give you a positive image. The color will be wrong, the contrast will be flat, and it may appear much lighter or darker than you would want a good-looking photograph to be. That’s all correctable. Not correctable would be highlights and shadows that are entirely lost because your scanner threw away some of the detail in the photograph.

Slides typically have very contrasty midrange tones, and flat highlights and shadows. Scanning slides in 16-bit mode ensures that you have plenty of well-distinguished tonal information in the highlights and shadows so that you won’t see posterization and banding when you improve the overall contrast characteristics of the photograph. Faded slides can be particularly tricky. Often one dye layer is hardly faded at all, while there is an overall buildup of stain. The result is that even though the slide looks washed out, it may actually have higher-than-normal density in one or more channels. A scanner’s 16-bit linear mode may bring in those extra-high-density data, but sometimes even that won’t be enough.

When that happens, the answer is to make two or even three scans at very different exposure settings and combine them (Figure 10). Photoshop offers the Merge to HDR feature to combine several differently exposed files into a single image. Picture Window does this better and more intuitively; it has a simple tutorial explaining how to do this using its Stack Image operation.

B&W film and glass plates

Most scanners see silver image densities as higher than dye image densities. A scanner may have no trouble with high densities in color film, but hit the wall with silver-based black- and-white negs. Many pre-WWII films and plates are especially dense and contrasty. The only times when you’re not likely to see high black-and-white densities are when the original is severely faded or bleached, as in Figure 9. Scan in 16-bit mode, using whatever scanner settings let you capture the maximum density range. As with color negatives, you’ll probably find that you’re best off telling the scanner software that you’re working with a color transparency.

Whenever you scan contrasty and high-density originals, be sure to mask off the unexposed areas at the edges of the film, film sprocket holes, and the clear parts of the film carrier or platen. Black construction paper’s simple, but effective.

Resolution decisions

Usually you won’t have to scan prints at extremely high resolutions. It’s rare to find an old print that has 1200-ppi-worth of fine detail. Usually 8×10-inch and larger originals don’t even have 600-ppi-worth of detail in the photograph. Often a mere 300-ppi scan can capture all of the real detail to be had. This is true even of contact prints from glass plates and large- format film negatives.

If you believe that a particular photo merits a high-resolution scan, do some test scans of a small section of it at different resolutions: 300, 600, 1200 ppi, and so on, to determine the point where increasing the scan resolution further doesn’t get you any more picture detail. Compare the scans on your monitor at 200% scale. If you find that the pixelation in the lower- resolution scans makes it difficult for you to compare the amount of image detail with the higher resolution scans, resample all of the files to the same ppi. Scanning at unnecessarily higher resolutions expands your file size dramatically; a 600-ppi scan is four times as large as a 300-ppi scan.

High-resolution scans of low-resolution prints can be useful when there’s physical damage with sharply defined, clear edges (Figure 11). Scanning at higher resolutions spreads out real image detail over many more pixels, while the damage boundaries remain pixel-sharp. This makes it easier to use edge-finding filters and similar tools to extract the damaged areas. When damaged edges and fine-image details are both only 1 or 2 pixels wide, it’s hard for software to distinguish between them. When the finest image detail is 5 to 10 pixels wide, you can do a pretty good job of masking that selects for the damage.

Adapting to circumstances

Even when working with a black-and-white original, I scan in full-color mode. One reason is that most scanners’ black-and- white mode uses only one channel (green) of data. That yields noisier scans than scanning in full-color mode. It’s easy enough to convert the image to monochrome once it’s in your computer, and the quality will be better.

Don’t worry about the exact image color when you’re scanning monochrome originals. Once you’ve got the scan in the computer, you’ll desaturate it and neutralize any color casts (when you print the restored image, you can adjust the color to replicate the look of the original). Avoid Photoshop’s Desaturate adjustment or Grayscale mode conversion, however. Desaturate simply mixes equal amounts of all three color channels. That’s a good choice only when no channel is significantly noisier, or shows more dirt and stains than the others. Badly deteriorated black-and-white photographs usually don’t look like that. For example, in Figure 12 the stains are primarily orange-yellow in color. The red channel is by far the most useful for performing a clean restoration, because the damage is substantially less visible there.

Grayscale-mode conversion mixes about 60% green, 30% red, and 10% blue. This will rarely be the optimum mix. Instead, use the Channel Mixer with the Monochrome option selected. Channel Mixer lets you specify how much of each color channel goes into the grayscale version of the photo- graph. Eliminate color channels that emphasize defects. Conversely, a conversion that emphasizes the channels that most clearly show damage is useful for constructing selections and masks to isolate the damaged areas for repair.

Color can be a valuable tool for restoring black-and-white photographs. Differences in color in different parts of the photograph are evidence of damage. You can use those differences when creating masks that select especially damaged areas for restorative work. There are even advantages to scanning monochrome photographs with exaggerated and unrealistic color to make masking easier. Once selectively masked, you can apply corrections to those areas separate from the rest of the image or use differential color information to “subtract out” damage and stains in a photograph.

Follow my scanning principles and you’ll discover that a good scan does half the work of making a high-quality restoration for you.


About the Author

Ctein
Ctein
Ctein is a technical writer and expert printmaker. He is also the author of Digital Restoration and Post Exposure—Advanced Techniques for the Photographic Printer.