Old silver photographs are prone to tarnish and bronzing—shiny or iridescent metallic-looking spots on the surface of the print, especially in the shadows. Tarnish is a vexing problem in restoration because it casts a veil over everything, but with the right techniques it is very easy to fix.

Figure 8-42, left, shows a B&W photograph that’s tarnishing out. There’s a bluish sheen to the blacks, and much of the print has a low-contrast, milky haze over it. The reduced contrast was easy to fix with a good scan (middle photograph); I got rid of some of the yellow in the highlights and improved the density in the shadows. It didn’t get rid of the tarnish, but it did make it clearer.

Fig. 8-42  <FC>The photograph on the left is not badly faded, but it is yellowed and severely tarnished. The middle figure shows the scan I made of this photograph that eliminated the stain and made the photographic image less yellow. This made the tarnish bluer. I accentuated the color of the tarnish with the Hue/Saturation settings in Figure 8-43 to produce the figure on the right.

The way to attack tarnish is to isolate it with a mask. I applied the Hue/Saturation settings in Figure 8-43 to exaggerate the color of the tarnish so that it would be easier to pick out. That produced Figure 8-42, right.

Fig. 8-43  These Hue/Saturation settings exaggerate the color difference between the tarnish and the underlying photograph, so that it will be easier to prepare a tarnish-selecting mask. The Master adjustment of +40 points exaggerates both the slightly yellow tone of the photograph and the blue color of the tarnish. The Cyan adjustment of +59 points kicks up the saturation in the tarnish even more.

Many Photoshop tools can create a mask out of color distinctions: the Color Range selection, the Channel Mixer, or Image Calculations ( see Chapter 7, Making Masks, page XX). The Asiva Selection plug-in is a favorite of mine. Figure 8-44 shows the Asiva Selection control panel with the curves I used to isolate the tarnish. The top curve restricted the selection to colors in the blue part of the spectrum, excluding undamaged parts of the photograph that were yellowish. I set the saturation curve to reject areas than had very little or no saturation, as those were untarnished. The resulting selection is shown in Figure 8-45.

Fig. 8-44  This screenshot of the Asiva Selection plug-in at work shows the settings I used to create the tarnish-selecting mask in Figure 8-45. I restricted the range of selected colors to the blue tones characteristic of the tarnish. I also adjusted the Saturation curve to reject any tones that were not at least moderately saturated. This did a good job of isolating the tarnish for repair.

Fig. 8-45  This is the finished mask that I used to repair the photograph in Figure 8-47.

I created a Curves adjustment layer with that selection. It would be just as effective to apply the curves directly to the selection, but the adjustment layer left me free to try out different curves that I could revise later. Knowing that the tarnish was greenish blue and lighter than the surrounding areas, I developed my curves by trial-and-error (Figure 8-46). The RGB curve doesn’t have a big effect; it darkens the shadows some and ensures that the darkest tones really are near-black. The green curve shifts the color substantially to the magenta, having the least effect in the highlights and the greatest effect in the shadows. The really big change occurs in the blue curve. I dragged the black point way in and added another control point, which pulled the whole curve down, so that everything except for the very whitest tones was shifted very strongly to the yellow.

Fig. 8-46  This Curves adjustment eliminates the tarnish from Figure 8-42 when it’s used in conjunction with the tarnish-selecting mask from Figure 8-45. The RGB curve slightly darkens the shadows, the green curve makes the tarnish more magenta, and the blue curve makes it much yellower. That eliminates the light blue/cyan cast that the tarnish had and makes it blend in perfectly with the photograph (Figure 8-47, left).

The resulting photograph in Figure 8-47, left, is tarnish-free. The details in the hair and the texture of the moire silk blouse are once again rich and clear. The restoration isn’t complete; there are yellow stains and marks on the surface of the print, and the paper texture is distracting, but I’ve fixed the worst problem. In the next section, I get rid of the stains and marks. Then I’ll subdue the annoying paper texture, which will finish up this restoration.

Fig. 8-47  The figure on the left shows the photograph from Figure 8-42 after tarnish repair. Tone and contrast are good, but there’s some yellow-orange stain, and the paper texture is distracting. The stain is least visible in the red channel, so I copied that into the middle figure and used the Clone tool to repair the minimal amount of residual damage. The figure on the right shows the fully restored photograph after I removed the paper texture with Neat Image (Figure 8-50).

In Chapter 7, page XX, I created masks using channels and the Channel Mixer to exploit color differences between B&W photographs and the damage inflicted on them. I took advantage of colors or combinations of colors that especially emphasized the damage over the photograph.

Those same tricks can be inverted to minimize the damage, using complementary colors. For example, damage that is especially visible in the green channel will have a strong magenta component. That means it will not be very visible in the “magenta” (red + blue) channel. I would use the green channel to build a mask from because it shows the damage so clearly. Conversely, if I wanted to suppress that damage, I would use the Channel Mixer to combine the red and blue channels and exclude the green channel.

I used the colors green and magenta merely as examples. The precise color that works best will depend on the photograph and the damage. If the damage looks reddish compared to the photograph, check out the red channel; if it looks cyan, then check out the green and blue channels. What you can usually count on when restoring a B&W photograph is that one channel will show the damage and stains on the photograph less than the other channels.

That’s especially true when you’ve done a good scan that produces a reasonably neutral-toned photograph. Stains and other defects usually have a different color from the photograph proper. Once you’ve finished with the kinds of repairs that require a full-color image (like the tarnish-reducing work I did in the previous section), look at the individual color channels for the photograph and select the one that looks the cleanest for further restoration work.

In the portrait in Figure 8-47 the stains are yellow-orange in color. When I inspected each color channel, I saw that the red channel displayed hardly any of the stains. I copied that channel into a new file that is shown in the middle column of Figure 8-47. I used the Clone tool to clear out the white marks of the top and remove a couple of small dark spots from the picture; it literally took just a few minutes work to make it look this good.

Dealing with Textured Prints

Paper texture isn’t really damage, because it was an intentional part of the original print. Print textures, though, usually look bad when they are reproduced on a flat-finish paper. If you want to restore an original, textured photograph to a fresh textured print, you will get a much better-looking print if you print a clean image on textured paper than if you try to print the illusion of texture. Consequently, I treat texture as if it were widespread damage—something I want to erase from the prints while doing as little destruction of photographic detail as possible.

Like tarnish and other surface blemishes, texture tends to get enhanced in scans. I talked in Chapter 4, Getting the Photo into the Computer, about rephotographing textured prints on a copy stand as one way to get around the paper texture problem. Here, I deal with getting rid of paper texture in scans.

The print in Figure 8-47 cleaned up nicely with the tarnish mask. To get rid of its texture I pulled up Neat Image, a very powerful noise reduction plug-in (see Chapter 3, Software for Restoration). Neat Image profiled a section of the background and created a filter that could cancel out the texture just as if it were noise (Figure 8-50). The filter was so effective that it completely eliminated the paper texture at its default settings, but it also softened the finest detail in the photograph.

Fig. 8-50  The Neat Image plug-in for Photoshop can clean up paper texture in a scan just as well as noise. I eliminated the texture from Figure 8-47 without compromising fine photographic detail by adjusting the high-frequency Noise Reduction Amount. Turning this all way down to 0% ensured that fine details like the eyelashes and catch lights wouldn’t be filtered out. Most of the paper texture was still eliminated, resulting in a very good restoration (Figure 8-47, right).

I wanted to retain the fine detail. I did that by using the noise reduction sliders. I set the sliders’ high-frequency Noise Reduction Amount” to 0%. I left the mid and low-frequency settings at 100%. That meant that the filter still had full effect on the coarse paper texture but left fine detail (and texture) alone.

Applying the filter produced the photograph in Figure 8-47, right. All of the image detail is still there, but almost all of the paper texture was removed. There’s a very fine “tooth” to the photograph, from the high-frequency texture I allowed to get through, but it’s not distracting or intrusive.

The half-century-old color print shown in Figure 5-54 is the most badly faded photograph I’ve ever worked on. Although there is considerable shadow density, the midtones and highlights have almost completely disappeared. This is also apparent in the scanner software histograms (Figure 5-55), which show that most of the tones fall into a huge peak near the highlights.

Fig. 5-54  This 55-year-old color snapshot is yellowed and faded to the point of near-invisibility.

Fig. 5-55  The scanner software histograms for Figure 5-54 confirm the overall fading problem but indicate that there is still information in the faded areas, as shown by the broadened peaks to the right in the histograms.

The first step in recovering information from a photograph like this is making the scan. Because I’ll need to make truly radical changes to the tonal distribution to produce a good restoration of this photograph, a 16-bit scan is a must. I set the levels controls in my scanner software to maximize the range of values in the file, which got me Figure 5-56. It’s still bad, but now at least I could make out some more detail in the baby and the crib. Because tones were so lacking and badly distorted, I felt it necessary to do some color correction before I made major changes in the tonal distribution or I’d risk losing data from the most badly faded red channel. Running the photo through DIGITAL ROC (see Chapter 6, Restoring Color) got me to Figure 5-57. Neutrality was restored, but the photograph was still extremely faint.

Fig. 5-56  A new scan of the snapshot using the levels settings from Figure 5-55 shows much more detail in the baby and the crib. This photograph may not be a hopeless case.

Fig. 5-57  Running the photograph in Figure 5-56 through the DIGITAL ROC plug-in restores even more detail and brings the faint beginnings of correct color to this photograph.

I’ve two approaches for handling images like this. The first is, as usual, Curves. I created the curve in Figure 5-58 for this photograph. It’s an unusual curve for photo restoration; I gave three-quarters of the darker tones very low contrast so that I could greatly exaggerate the density range in highlight tones. I moved the white point in a bit to eliminate the residual highlight stain.

Fig. 5-58  This Curves adjustment turns Figure 5-57 into Figure 5-59a. The curve drastically darkens the highlights and midtones, and it increases contrast in those tones by a factor of four.

Fig. 5-59  There are many ways to intensify detail in a faded photograph. All the photographs here are derived from Figure 5-57. Figure (a) is the result of applying the Curves adjustment in Figure 5-58. To create (b) I took Figure 5-57, duplicated the background layer, and blended it in using the Multiply mode. I produced (c) by making two duplicate layers and blending them both in using Multiply. Figure (d) is the same as (c), except that I desaturated the second duplicate layer, so that I was multiplying luminosity but not color.

The resulting photograph, in Figure 5-59a, could not remotely be considered a good photograph. A lot of work needs to be done to turn this into the acceptable restoration seen below; see Chapter 10, Examples, page XX, for a description of how I completely restored this photo. What’s important, though, is that a good scan and a serious dose of Curves were sufficient to turn nearly blank photographic paper into a recognizable, acceptably detailed image.

Good scans and Curves are techniques you can use with any image processing software, but Photoshop provides another way to enhance faint images. It’s the “multiply” blend mode. Take an image, duplicate the layer, and set the new layer’s blend mode to “Multiply.” You’ll see a marked increase in contrast and density.

I did this to the photograph from Figure 5-57 and got Figure 5-59b. In some ways it’s not quite as good as Figure 5-59a. Multiplication didn’t produce as much enhancement of the highlight detail. On the other hand, there is less color distortion in the multiplied version, especially in the midtones and shadows.

For extreme highlight recovery, multiply more than two layers together. I took Figure 5-59b, duplicated the base layer again, and blended it in multiply mode. That produced Figure 5-59c. Now there’s a decent range of densities, but color distortions, noise, and defects are similarly amplified. It will take a lot of work to clean up this photograph.

Here’s a very useful variant on the last trick. I kept the new layer as a multiply blend, but I desaturated that layer. In effect, that only multiplied the luminosity. That eliminated some of the color distortions at the cost of reduced saturation (Figure 5-59d). Sometimes this would be an acceptable trade-off. Luminosity multiplication is a good technique to remember when you want to exaggerate or enhance the tonal scale without doing the same to the color values.

  The fully restored photograph, with its adjustment layers.