Scanning an Image or Document

Activity

• Bring a photograph from home, scan it and file it.
• Scan a text document, file it and email it to yourself.

Converting" Photos: Digital Scanners

A digital camera is fine if you're taking new pictures, but what about the hundreds of pictures filling all those shoeboxes in the closet? Or, what if you like the sense of permanence of film-plus-print that conventional photography offers, not to mention the wide range of film types, and even interchangeable lenses available with some cameras? Finally, film captures much more detail in a picture than do current digital cameras: If you want to enlarge your photos much, you'll need film originals to do so.

The way to get conventional film or printed photos into your camera is through a scanner. Scanners get their name from the way they convert the film or print into digital data: A high-resolution line sensor (all it's light-sensitive cells are arranged in a line) moves or scans across the photo medium, measuring all the color and brightness values one line at a time. Because they have the luxury of doing the job one line at a time, even relatively inexpensive scanners capture much more detail than do most digital cameras.

Just as with digital cameras, the pixel resolution of a scanner is an important characteristic when it comes to determining how accurately it can reproduce a photograph. Scanners are a bit different case than digital cameras though, when it comes to specifying resolution: The sensor in a digital camera always sees the world through the same "window", and so always captures the same number of pixels. By contrast, virtually every image a scanner captures will be a different size, based on the dimensions of the original being scanned. Rather than stating the maximum number of pixels scanners can capture, it is much more common to refer to their resolution in terms of "pixels per inch" (ppi) or "dots per inch" (dpi). Many scanners can capture more pixels per inch along the direction the sensor head is moving than along the length of the sensor itself. Thus, you'll frequently find scanners with resolutions of "300 x 600 dpi" or "600 x 1200" dpi.

So how many dpi do you need? The answer depends a great deal on what you want to do with the device: If you only need to scan from conventional photographic prints, 600 dpi is more than enough, and 300 dpi will be plenty for most applications. This is because photographic prints themselves have a fairly limited resolution, generally accepted to be something on the order of 200-400 dpi. On the other hand, film is quite a different beast, and resolutions of 2000 dpi or greater are required if you want to achieve the maximum possible enlargement.

The huge difference in resolution requirements for film vs. prints is why the scanner market is generally split into two segments: "Flatbed" or "document" scanners are designed for scanning reflective media (photo prints, printed pages, etc.), and "film" or "slide" scanners are made to handle film. Aside from media handling, the biggest difference between these two classes of scanners is the resolution they can achieve, with film scanners typically offering much higher resolution over a smaller area.

Another important parameter, particularly for film/slide scanners is something called "bit depth." In the case of slides especially, portions of the image can be extremely dense optically, letting very little light through. For the scanner, this means it must be able to measure very small amounts of light accurately, while at the same time not losing detail in the brighter regions as well.

Computers keep track of numbers with "bits," short for "binary digits." A bit can have values of only one or zero, but by stringing enough of them together, any size number can be represented. The trick for a scanner is to be able to measure with enough "bits" of accuracy to accurately distinguish between of brightness in the deep shadows of an image, and yet have enough range to handle the brightest highlights as well. You'll find scanners referred to as 24, 30, or even 36-bit units. These figures refer to the total number of bits the scanner can capture for all of its "color channels" taken together. (Huh? "Color channels?" - This just refers to the fact that scanners determine the color and brightness of an image by measuring the amounts red, green, and blue light transmitted or reflected by it separately. They thus are said to have separate red, green, and blue "color channels.") The three common bit depths just mentioned correspond to 8, 10, or 12 bits of accuracy for each color channel separately. Eight bits corresponds to 256 different levels of brightness, 10 bits correspond to 1024 levels, and 12 bits to 4096 levels. The more bits, the more accurately the scanner can measure the reflected or transmitted light, and the finer steps it can recognize. (As noted earlier, having fine measurement steps is particularly important in shadow areas.)

While many film scanners measure 8 bits per channel, this really isn't adequate for darker images. The HP PhotoSmart measures 10 bits per channel, and produces good results even on moderately dark slides. (Color negatives aren't as dense as slides, so 10 bits is more than enough for most negatives you're likely to encounter.) As to resolution, the PhotoSmart scanner scans prints at a resolution of 300 dpi, and negatives and slides at 2400(!) dpi. In each case, the available resolution comes close to the maximum available from the medium itself. (Again, see our review of the PhotoSmart Scanner for more information.)

We need to discuss one final issue about scanners; their interface to the computer. On Windows PCs you can connect peripherals to the built-in "printer port," or via an add-on "SCSI" interface card. Newer computer models now include Universal Serial Bus or "USB" ports as well. The importance of the interface type for a scanner has to do with how quickly each type of connection can move data. The parallel port is the most common, and in the past has been the easiest interface to use (although the new USB ports may be about to usurp this position), but is also by far the slowest of the three. SCSI ports are the fastest of the three types, but generally require adding an interface card to your computer. USB ports are somewhere in the middle in terms of speed (although quite a bit faster than parallel ports), but USB-based peripherals are only just now (October 1998) beginning to appear on the market.

The reason the computer connection is so important is that scanners have to move a lot of data in the process of scanning a picture. If you only have one or two prints to scan, you can probably afford to wait, but much more than that and you're likely to become impatient. Full-resolution film scans could take a painfully long time to complete with parallel-port-connected scanners. While the prospect of adding a card to their CPU is enough to send most computer users running in the other direction, this needn't be the case: The benefit of a SCSI connection shouldn't be under-estimated: The PhotoSmart Scanner is quite responsive, thanks in large part to its high-speed connection to the host computer.

"Fixing" Pictures

As was mentioned at the outset, there's literally a world of things you can do with your pictures, once you've got them into your computer. What we want to focus on here though, are some of the basics of getting good-looking pictures. There's lots of information and support available for creating photo projects (see any of the many consumer-level imaging programs, including Microsoft's excellent PictureIt) but very little on how to adjust your photos to correct common problems of exposure or color balance. Fortunately, the basics of making good-looking prints are pretty simple, and once learned, can be applied to any subsequent projects you undertake.

What goes wrong?

Think back to various photos you've gotten back from the photofinisher that were less than you wanted them to be. What were the problems with them? If you consider your various "photo failures" for a moment, you'll probably find that they fall into a few basic categories: Exposure problems (overall too bright or too dark), contrast problems (bright areas are too bright relative to dark ones (common in flash photos), or the whole picture lacks contrast and is "flat"), overall color-cast problems (common in indoor pictures taken with daylight-balanced film), or problems with dull, lackluster color. ALL of these problems can be corrected on the computer relatively easily. What the computer won't do though, is turn you into a better photographer, or fix blurry images caused by poor focus or camera shake.

The big secret!

There's actually a single "secret formula" for successful computer photo prints that works most of the time: "Let your black be black, and your white be white." (Amen.)

What?! Is that all? Actually, it's a lot of it. A significant majority of all bad-photo problems are the result of not properly utilizing the tonal range of the paper and ink. If the darkest parts of your pictures don't go all the way to the maximum density your printer is capable of, your images will look washed out and pale. Likewise, if the lightest parts of your pictures don't go all the way to the whitest white available, the photos will look dark and "muddy." What's surprising to most people is the extent to which problems of dull, "dirty" colors are also the fault of poor tonal adjustment. We keep throwing around the term "tonal" here, so it might be good to briefly define what we mean when we're talking about tonal range or tonal balance: In photographic terms, "tone" simply refers to the lightness or darkness of an image or part of an image. As just noted, effective use of tonal range is a big deal when it comes to getting good prints, largely because the available tonal range of photo prints is so much less than that of real scenes. Think about it: The lightest part of a photo print can't be any brighter than the light that's reflecting off the underlying paper. Compare that maximum brightness to the glare of full sunlight on a cloudless day! Much of the art and magic of photography lies in intelligently compressing the incredible range of brightness our eyes respond to in natural scenes into the range that can be reproduced on paper.

In computer terms, what this boils down to is making sure that the blackest black in our images translates to red/green/blue values of 0, 0, 0, and that the brightest white translates to 255, 255, 255. (255 is the highest "brightness" number that can be stored in most computer files for each of the three RGB "primary" colors.)

In most PC imaging programs, you'll find adjustments for "brightness" and "contrast," that help you stretch or compress the tonal range as best suits each photo. These two controls sometimes work differently in different programs, but the concept is the same: The "contrast" control affects how the computer spreads out the range of brightness values from darkest to lightest, while the "brightness" control moves the picture's total range of tonal values up or down the scale.

The saturation slider is one you'll need to experiment with a bit to learn how much to increase or decrease it for best results: It's definitely a control for which a little bit goes a long way! A common mistake of beginners is to focus on the bright colors in an image, and not realize what's happening to the colors that really should appear as pastels. A good example of this is Caucasian skin tones: A neophyte image-adjuster may boost the saturation in an image to get rich, vibrant greens in foliage, but not notice that the subject's face has also turned a ruddy shade of red.

You'll probably notice saturation and contrast interacting quite a bit, which is a natural consequence of the underlying color theory: Low-contrast images tend to have less-saturated colors, while higher-contrast ones will naturally be more saturated. Often, some back-and-forth will be necessary, moving between the saturation adjustment and the tonal controls. If you get the tonal balance right in the first place though, saturation adjustments should be minimal.

Sharpness

Image "sharpening" is a subject that really deserves a whole separate article, but a few basics here will help clear up the confusion that exists around the subject.

The first thing to note, and very forcefully at that, is that no adjustment of "sharpness" controls in an imaging application can correct for poor focus when the picture was originally taken! If the picture is out of focus, it will stay out of focus, regardless of what you do.

"Sharpening" in imaging applications should only be used to compensate for the softening that naturally happens when an image is scanned or printed. (Or for that matter, captured with a digital camera.) In the camera or scanner, the pixels naturally average-out any picture detail within their boundaries, amounting to a blurring of the photo. Likewise, when a photo is printed, the printing process tends to reduce picture detail again. (More on this later.)

To correct for this input/output blurring, software like the PhotoSmart Photo Finishing software includes a "sharpening" functions. This works by slightly increasing the contrast anywhere there's an abrupt change from light to dark or vice versa: Right along the dark side of an object's edge, the sharpening software makes the image a little bit darker. Likewise, it makes just the edge of the lighter side a bit lighter. The more-abrupt contrast change tricks our eyes into seeing a "sharper" edge, even though no new information has been added to the picture.


As with so many things, a little sharpening is a good thing, but more isn't necessarily better. You'll want to experiment with the sharpening control, to see what settings work best with your own pictures. Two bits of guidance: First, try less, rather than more - a little goes a long way. Second, increase the amount on images that you've enlarged more, since the blurring effect of the initial scan or digital photo will affect a larger area. In the PhotoSmart software, the sharpening control starts out all the way to the left, at the "minus" end of the scale. This is a little confusing, as it seems like this would be decreasing the image sharpness, but in fact it just means that no sharpening is being applied at all.

Sharpening is easy to do in the computer, but very difficult or impossible to do conventionally. With the right amount of sharpening, your digital prints can actually look sharper than conventional photo enlargements. In fact, with a good-quality photo printer, and the right (small) amount of sharpening applied to your images, people will almost always guess wrong when shown a conventional photo next to the computer print and
asked to tell which is which!

back to top