Great Scans

Discover What Helps and What Doesn't

By Ctein Back to


Scanning is serious business. Think of your scanner as a combination enlarger and enlarging lens, the intermediary between your original film (or photographic print) and your finished photographs. For many photographers, a good enlarger and lens were the single most expensive and important purchases they made in their career, and they would pore over the details and published tests to determine which ones would really serve their needs.

Learning how to make high-quality enlargements was not an overnight task. Cleanliness, sharpness, freedom from stray light and contrast-robbing flare, all were concerns of the darkroom printer. Just as with darkroom work, making good scans is a matter of both equipment and methods.

Flatbed or film scanner

For scanning prints, you’ll use a flatbed scanner. Print scanning is relatively undemanding. The density range and image detail of a print rarely exceed the density range a moderately priced scanner can capture.

Scanning film is dicier. A dedicated film scanner almost always captures a much better density range and more detail than a flatbed scanner. Yes, a good medium-format film scanner will set you back around $2,000, but how much did that really good enlarger and enlarging lens cost you (in current dollars)? You will get many years of use out of a good film scanner; it’s a long-term purchase that pays for itself with better prints and faster printing.

Large-format film scanners can handle 4×5-inch and even 8×10 formats. They produce superior results, but can you afford one? They’re outside my budget. For large- format film, I’m forced to rely on a flatbed scanner; $1,000 or less buys a pretty good one.

Is it sharp?

Flatbed scanner manufacturers’ resolution figures never tell you how much detail the scanner captures. Strangely, few reviews even test this.

The optical or physical resolution of a scanner is a measure of how many pixels the scanner captures per inch.

For instance, my last-generation Epson 4990 scanner is spec’d at 4800×9600 ppi. That means it captures 4800 pixels per inch across the width of the platen and 9600 pixels per inch in the direction of travel with the scanner head. The way it achieves twice the pitch lengthwise is by microstepping the scan head in half-pixel increments. That’s a real and legitimate technique for obtaining higher resolutions and is often used in advanced scientific equipment. Ignore the software-interpolated numbers; there is no additional detail captured, only a lot more pixels generated.

The true resolution of flatbed scanners is usually much less than the maximum pixel count. Obtaining real resolutions of 2400 or 4800 ppi requires extremely tight focus tolerances. A fraction of a millimeter error in the position of the film plane is all it takes to blur fine detail. Unlike film scanners, flatbed scanners rarely have focusing optics. With no way to adjust focus, you’re dependent upon the flatbed scanner being manufactured to near-perfect mechanical tolerances. What do you think the chances are of that?

I compared the Epson scanner to my Minolta Dimage Multi Pro film scanner. Both claim a top resolution of 4800 ppi, ignoring microstepping. That corresponds to approximately 95 line pairs per millimeter (lp/mm). Fortunately, I have targets that well exceed that.

I scanned the reflection target on the Epson scanner at various pitches all the way up to 4800 ppi. The best widthwise resolution I got was just barely 40 lp/mm at extremely low contrast, although it sharpened up well. The best lengthwise resolution was 32 lp/mm with low contrast. There was nothing gained from scanning at higher than 2400 ppi.

Transmission scanning was a little sharper. The maximum widthwise resolution was 50 lp/mm with good contrast, which would justify going to 3200 ppi. Lengthwise resolution was still down at 32 lp/mm with low contrast (Figure 1).

The Minolta scanner was almost twice as sharp as the Epson, even though both claimed the same pixel pitch. At 4800 ppi, its widthwise resolution was 90-plus lp/mm with low-to-fair contrast in the center and 80 lp/mm at the edges with a little smearing. The lengthwise resolution was 65–80 lp/mm with good contrast.

I’ve not tested most of the film or flatbed scanners on the market; you no doubt will find some that perform better or worse than this pair. But in general, a film scanner is much better than a flatbed for scanning film, and you should never, ever trust the manufacturer’s quoted figures for how much detail a scanner will give you.

Unfortunately, I don’t know of any source for inexpensive high-resolution targets for running your own tests. If any readers know of targets that hit 100 lp/mm or better (and don’t cost hundreds of dollars), please drop me an e-mail.

When you’re scanning film, always scan the film at the highest resolution your scanner meaningfully supports (e.g., 3200 ppi on my Epson). Film grain can never appear smaller than one pixel in size. Because film grain acts a lot like random noise, it doesn‘t blur out very quickly when resolution drops. Consequently, a low-resolution scan isn’t grainless; instead it has large, mushy, soft grain, looking a lot like an out-of-focus print (Figure 2). Of course, once you have the photograph in your computer, you can use noise-reduction tools (reviewed last issue), but the less you have to use them, the better quality and the more fine detail and tonality you’ll retain in your photograph.

Cleanliness is next to…

All flatbed scanners eventually collect a haze on the underside of the glass platen. This is a big problem when making ref lection scans (it has much less effect on transmission-film scans). Just like the scum that forms on the inside of your automobile windshield, the haze on the platen reduces contrast and degrades shadow detail. Pay attention to how the glass looks when the lamp inside the scanner is on—a scummy haze will be obvious. If you get the feeling that you’re losing maximum density range in your scans, take a close look at that glass; odds are that it’s collected a film that you didn’t notice.

External glass surfaces, of course, should be cleaned at every opportunity. If you can see any dust or smudges on your scanner glass, on the platen, or on the underside of the film-illuminating lid, you’re seeing too much. Believe me; your scanner is more sensitive to these things than your eyes.

Some scanners are easier to disassemble than others. Fortunately, these days there are lots of online instructions for how to take apart your scanner. If that’s something you’re comfortable with, you can keep the glass crystal clear at no expense beyond your periodic maintenance time. Film scanners, on the other hand are pretty much sealed black boxes. Fortunately, I’ve never seen evidence that my Minolta scanner needs cleaning. When it does, I’ll just have to f ind a competent repair shop.

There’s only so much I can do myself to maintain a complicated piece of equipment.

Many quality scanners come with something called Digital ICE. It does a good job of eliminating dust and scratches when it’s built into the hardware. Some flatbed scanners include a software-only version of Digital ICE and it doesn’t work very well. Hardware-based Digital ICE performs an infrared scan of film. Color film dyes (except Kodachrome’s) don’t absorb much infrared; dust and scratches do. That gives the software the information needed to subtract dust and scratches from the scan, and it’s why Digital ICE doesn’t work with Kodachrome or with black-and-white films.

Some flatbed scanners have real infrared Digital ICE; most don’t. How do you f ind out what a flatbed scanner really has? Download the full manual for the scanner from the manufacturer’s Web site. If the instructions for Digital ICE warn that the software won’t work with black-and- white or Kodachrome film, you’ve got the real deal.

What does a good scan look like?

In my article on scanning back in the September/October 2007 issue of this magazine, I spent a lot of time discussing what constitutes a good scan. Here I’ll keep it short and simple: a good scan is one that uses 16-bit channel depth (that’s 16 bits per pixel in grayscale, 48 bits in color) and that uses most of the range in the histogram. Don’t try to get 100% coverage; if you do, you’ll likely f ind that you have some small areas that are going to pure black or white in the finished scan and that means you’re throwing away a bit of highlight or shadow detail. If the histogram looks like it’s about 90% full, with minimum values around 15 and maximum values around 240, you’re right where you should be.

Almost all color photographs have density ranges that span the full range of values from near black to near white in all three channels. If you use the levels control built into the scanner software to individually adjust the color channels until they all span a similar 90% range, you’ll be surprised to find how often you are close to a correct color balance for the photograph.

Until pretty recently, I didn’t know that it can make a big difference what color space you make your scans in. I owe Mikkel Aaland, author of the Lightroom Adventure books, for that tip. Film, especially slide film, contains a very, very wide range of colors (negative f ilms don’t have this problem as much because, effectively, their data is compressed). Most photographers these days work in the Adobe RGB color space because it’s a pretty good match to the display capabilities of a high-quality monitor and the color range of a good printer.

The thing is, slide f ilms routinely include colors that fall well outside of Adobe RGB. If you scan such a slide into Adobe RGB color space, you get clipping: one or more of the color channels will have areas that go to pure black or pure white, with values of zero or 255. The original slide will have tone and color gradation in those areas that won’t make it into the scan. This loss of quality can become very evident when you start to make any adjustments to tone or color.

Good scanner software has an option for choosing the output color space. If it doesn’t, it’s time to go buy some better scanner software. Look for ProPhoto RGB, a very large color space. Use a 16-bit workflow with this space or you’ll have problems with contouring and banding. When you open the scan up in Photoshop you’ll get a warning box asking you which color space you want the photo opened up in (unless ProPhoto RGB is your default working space). Be sure to tell Photoshop to open it in ProPhoto RGB.

As a demonstration of the difference color space makes, I scanned a brightly colored Kodachrome slide into Adobe RGB and Wide Gamut RGB color spaces. My scanner software is too old to know about ProPhoto RGB; Wide Gamut is a good fallback.

As a demonstration of the effect this has on color and tone values, I lightened the shadows in both scans to bring out the details (Figure 3). Notice that there is tone and color gradation in the red and magenta areas in the ProPhoto RGB scan, and the cyans and blues are clean and pure in color. In the Adobe RGB scan, reds come out pink, most of the variations in tone and color in the deep magentas and reds are gone, and the color in the blues and cyans is distinctly off.

The histograms of the original scans (Figure 4) show the problem with Adobe RGB; large portions of the slide have a solid zero for the green value (maximum magenta). There are also some areas where the blue (maximum yellow) also clips. Grayscale images of the green channels in both slides show how unnatural this is (Figure 5); the solid magenta (black) areas in the Adobe RGB scan are very obviously not normal.

Expanding the shadow range in the Adobe RGB scan didn’t lift the clipped values in the green channel because zero times anything is still zero. In the ProPhoto RGB scan, the deep magenta values increased proportionately along with the other color channel values.

Evil, evil flare

Figure 3 shows another scanning problem: stray scattered light. That can cause trouble even with black-and-white negatives, but it isn’t as big a problem with color negatives because their density ranges are lower. When making the scans for this article, I intentionally misaligned the slide so some clear sprocket hole would show through. You can see bands of light pouring into the image area as strong white light degrades nearby shadow detail.

Easy to fix! Make sure you mask off the areas outside of the film image where white light can get through. It doesn’t matter what you use to do this; black construction paper or strips of black plastic work fine. Just so it blocks the light.

It’s better when wet

Several years ago (November/December 2007), I wrote about the joys of oil-immersion scanning. OK, truth— it wasn’t at all enjoyable. The astonishingly improved scan quality made it worth the trouble, but it was a huge amount of trouble. As that article was going to press, I discovered ScanScience ( and included a sidebar about their products. Since then, I’ve converted entirely to the ScanScience way and abandoned my makeshift oil methods.

The most amazing thing about fluid-immersion (a.k.a., wet) scanning is that grain actually gets finer and the scans are cleaner (Figure 6). That has to do with surface characteristics of film that I won’t get into here, but the difference between ordinary dry scanning and wet scanning is like going from 35mm to medium-format film or going from ISO 100 film to ISO 400.

Good film scanners include very effective grain-reduction software, Digital GEM, but it robs a little bit of sharpness and fine detail. My ScanScience scans have less grain than I got before and they are more detailed and flaw-free.

ScanScience kits fit most scanners or they can make them to order for your film or flatbed scanner. The price depends on the scanner, the format, and what’s included in the kit; they run from under $50 (Canadian) to $280. Check their catalog for details.

It’s simple to do. You’ll need the equipment shown in Figure 7 (ScanScience also sells kits that include gloves, hanging clips, and a squeegee). The canned air, PEC Pads, and PEC- 12 are for cleaning off the film and equipment before and after scanning. At the top is a bottle of the Lumina immersion fluid. At the bottom, on the right, is a thin, optically flat glass plate and next to it a similarly sized sheet of transparent flexible plastic.

After you’ve cleaned everything thoroughly and blown off any residual dust, spray one side of the film and glass plate with a light coating of Lumina (Figure 8). Bend the film into a U-shape (Figure 9) and touch the center of it to the glass plate to make a bead of liquid. Slowly lower the ends of the film and you should get a bubble-free layer of fluid between the film and the glass plate. If not, try it again or use a squeegee to squeeze out the bubbles.

Next, spray the exposed surface of film and one side of the plastic overlay sheet with Lumina. Lower the overlay sheet onto the film the same way you lowered the film onto the glass plate. Pick up the whole “film sandwich” and put it in your film carrier or on your flatbed platen with the plastic overlay facing the scan head. Scan away.

I never make a dry-film scan anymore. It throws away too much quality. Everyone out there who is serious about getting the best possible scans needs to be doing wet scanning, and ScanScience makes it affordable on almost any scanner.

I threw a lot at you in this article. I’m confident, though, that if you follow my suggestions, you will be pleased and amazed by how good your scans look. Even if you can’t afford the best scanning equipment, improving your technique will do a lot to improve your scans.

About the Author

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.