Use your browser's "Back" button to return to the previous page, or the links at the top and bottom of this page to navigate to related information. If you have difficulty fitting the text on this page onto your printer output, simply resize your browser window to a narrower width and print again.

Back to Full Finding the Right Digital Camera Article
Up to Imaging Resource Cameras Page
Up to Imaging Resource Scanners Page
Up to Imaging Resource home page

Finding the Right Digital Camera

The overall goal of the Imaging Resource is to help people chose and use digital imaging technology as effectively as possible. As simple as they are to use, choosing the right DPS camera involves a number of decisions. Key issues are computing platform, image quality, on-board image capacity, exposure versatility, feature set, and included software. In this article, we'll lead you through some of the issues to consider in making a purchase decision, and at the same time suggest some things to look for in the test images we've collected on our web site.

We're interested in your feedback and experience: We invite you to elaborate on this article by leaving comments via the database link at the bottom of the page. You can also give feedback in our "Forum" discussion-group section.

Now, on to the issues:

Computing Platform
This is an obvious first choice, but is becoming less of an issue than it once was. Most current cameras can be interfaced to both Mac and Windows platforms, although some include more software for one platform than the other. A few still work only on one or the other. (Contrary to earlier trends, these days, if a camera only works on a single platform, it's more likely that platform will be Windows.)

Image Quality
Image quality is a complex subject involving exposure accuracy, color purity, optics, and image compression techniques. We'll cover this whole area in greater detail later. For now, the main thing to keep in mind is that there's really no substitute for direct experience: Look at the sample images on this site that correspond to your expected usage you may be surprised by the differences between cameras!

On-Board Image Capacity
Depending on your application, the amount of on-board image storage could be pretty important. There is generally a tradeoff between image capacity and image quality however, so don't get too excited by a large image capacity until you check to see how much actual memory the camera contains compared to the resolution of the images. One camera may promise storage for twice as many images as another, yet have no more actual storage space. The consequence would likely be a loss of image quality. On-board camera memory capacities (as of early 1998) range from 2 to 8 megabytes, while image storage can vary from 10 to more than a hundred images. Again, look at the test images on this site to decide what the usable storage capacity of each device would be for your application.

Many cameras allow you to make your own decision about the memory/cost tradeoff, by allowing you to store images on plug-in cards. This approach allows each user to add as much or as little memory to the camera as they feel they need. Once upon a time, most cameras used a single type of memory card (so-called linear PCMCIA), but those days are long past. Today, there's a plethora of memory types in use, although the market seems to be settling down to a choice of either CompactFlash or SmartMedia. (Look for a FAQ on removable memory to appear soon on the Imaging Resource website.)

Exposure Versatility
One notable disadvantage of digital cameras is their inability to handle radically different lighting conditions just by using a different type of film: With a digital camera, the camera is the film, so whatever you buy is what you'll have to live with. This means you'll want to be sure your camera's exposure system (shutter, variable lens opening, supplementary flash) can handle a wide enough range of conditions.

In this category, look at a camera's "equivalent ISO number," which is a measure of how sensitive the camera's sensor is to light. A higher number means the camera can successfully capture images with less light. Most DPS cameras have equivalent ISO ratings in the 80-200 range, although some go well beyond these limits. Beyond the ISO rating, look for a wide range of shutter speeds and lens apertures. Typical numbers here are shutter speeds from 1/30 to 1/1000 of a second, and lens openings from f2.8 to f16, although the present generation of cameras has shutter speeds ranging from 1/2 to 1/10,000 of a second(!) The wider the range spanned by both these sets of numbers the better.

Also look at the capability of the built-in flash if one exists. Key factors here are the distance range over which the flash will be useful, and the number of "modes" it has. First-generation cameras had fairly limited flash power, evidenced by maximum working distances of 10 feet or less, but many current models reach 16 feet or more. Flash "modes" refer to different ways the camera's light sensors can control the flash circuitry. In the most basic mode, the camera "stops down" (closes) its lens aperture and runs the flash at full power so most of the light hitting the subject is provided by the flash. Some cameras offer a "fill" mode for their flash. In this mode, the camera sets the shutter speed and lens opening based on the amount of light coming from the scene overall. In "fill" mode, only enough flash illumination is used to "fill in" the shadows and bring the illumination up to the minimum required. The result is often a more natural, evenly-lit subject. Many cameras also offer special "red-eye reduction" flash modes, in which the flash blinks one or more times before the actual exposure, to make the subjects' eyes "stop down", reducing the internal reflection that produces red-eye.

Feature Set
Besides the basic parameters listed above, you'll find a variety of features on the cameras that may have varying degrees of usefulness in your application:

Close-Up and Wide-Angle
The ability to take close-up pictures is critical to many applications. Many DPS cameras have a "fixed-focus" lens, in which the focal point is set so everything from about 4 feet on out is in focus. This is handy if you need to photograph subjects in that range, but pretty useless if you have to take a picture of something only a few inches wide. Some DPS units have optional attachments to improve their close-up performance, while others offer special "macro" focusing modes that let you get much closer to the subject than a standard lens would. (Look at the macro test shots on the Comparometer(tm) (http://www.imaging-resource.com/IMCOMP/COMPS01.HTM) to see how well various cameras do in this regard.)

Wide-angle photography is another important ability if you need it. (This could be particularly important for real estate or insurance photography, where shooting conditions may be cramped.) Again, some cameras have aftermarket accessories available to expand their field of view, so you will want to take this into account when purchasing. Several cameras feature zoom lenses, with a range of focal lengths running from a mild wide-angle (roughly equivalent to a 35mm lens on a 35mm camera) out to a mild telephoto (slightly more than the equivalent of a 100mm lens on a 35mm camera). While you invariably pay more for a camera with zoom capability, we've found the feature very useful in real-life shooting situations.

Battery Life
We're not sure if this is a "feature," but it's important, and this seemed a logical place to include it. You'll find a wide variation between cameras in how long their batteries last. Some manufacturers specify battery life, and others don't. Unfortunately, we haven't developed a (reasonably easy) consistent way of measuring power drain for DPS cameras. Be wary of manufacturers claims for battery life as well, as they are sometimes overstated, or may be specified with costly battery types (such as lithium cells). An important question is whether the camera can run off rechargeable batteries: If you plan to take a lot of pictures, the ability to use rechargeable cells could save a lot of money in the long run. (Note that the mere presence of an external power adapter may not mean the camera can use rechargeables, and probably does not mean the unit will recharge batteries while it is plugged in.)

External "booster" battery packs are available for some cameras, increasing their working time in the field manyfold. Keep this in mind if you plan to spend long periods on the road, far from power outlets. If a "booster" pack isn't offered, can you change or replace batteries in the field?

External Power Adapter
While we're talking about power, how about an external power adapter? You'll find DPS cameras generally consume much more power while communicating with the computer via their serial ports. If you have an option to run from a power adapter while downloading images, your batteries will last a lot longer. It's important to note here though, to be certain to use the correct power adapter with you camera the wrong one could easily fry a thousand-dollar investment!

Self-Timer
A self-timer is simply a shutter delay which allows you to trigger the camera, then run around to be included in the picture before the shutter snaps. This is an easy addition from the standpoint of the technology required, but often omitted. If you plan to be part of any group photos, see that your camera has this feature!

Included Software

Image Download
The software included with the camera can be pretty important. In fact, this is an area where you'll find some of the greatest variation between cameras. In our reviews, we'll try to give you some idea of how the different units fare in this respect. Things to consider include how convenient it is to unload images from the camera. Can an entire camera-full of images be pulled onto your hard drive quickly for later processing? Or, is the camera held captive while each image is individually processed? Alternatively, can "thumbnail" images be viewed prior to downloading the full-size files? - This can save a lot of time in culling the few best shots from a large group.

Image Manipulation
Getting the images into the computer is just the beginning: Does the included software allow you to manipulate the images after you've captured them? How easily can you compensate for poor exposures, color casts, or misaligned images? How about minor retouching to remove blemishes, errant reflections, etc.?

File Formats
Particularly important is the software's file-export capability: Can the furnished software easily provide images in the file format you need? Your requirements may be different if you plan to capture images for multimedia presentations rather than printed output. Most camera software is fairly competent in this respect, with some manufacturers increasing their present level of capability. Virtually all packages provide for export of standard RGB TIFF or JPEG files, which most document layout and word-processing applications will support as well. If you need formats beyond the standard TIFF or JPEG, look into what the camera software supports more carefully.

Cataloging
One of the first things you'll learn once you start using a digital camera is how fast images pile up! (If you haven't already, you'll doubtless find yourself buying a much larger hard drive.) Having adequate storage space is only half the battle though: Finding images is another matter entirely! In recognition of this fact, some cameras include at least rudimentary image cataloging capability in their software package. (On the Mac, just the ability to create a "thumbnail" preview to display as the file icon is a big plus. A word of warning though: Hundreds of file icons with thumbnail images attached can really slow down access to folders on your hard drive. Turning off the "preview" option and using a cataloging program instead can really help system performance.)

Some manufacturer-supplied camera software includes rudimentary cataloging capability, but in our experience, none of these packages offer the level of capability you'd need to manage more than a few dozen files. Accordingly, if you need to track large numbers of images, you should really look at one of the many image-database programs on the market. Some manufacturers have begun to bundle more powerful third-party image cataloging products with their cameras, a very welcome addition.

Links to External Applications
One of the nicest characteristics of "desktop" applications is how easily you can move data or design elements from one application to another. The Macintosh has excelled in this respect for years, and the Windows platform mad rapid strides beginning with Windows 95. There are varying degrees of integration available though. Some software requires you to save a file to disk in one application before it can be imported into another. Other packages let you place an element from one application directly into a document of another. Sometimes, you can simply "drag and drop" an image from the camera software into a page layout program. Look at the camera software with your specific application in mind, and see what's required to move the images from the camera to the programs where you'll finally use them.

Image Quality
Image quality is one of the most important characteristics to consider in choosing a camera, and is impossible to evaluate from manufacturers' spec sheets. This underscores the importance of the test images on this site: Ultimately, the only way to tell if a given camera will produce acceptable results in your application is to look at and try out images of similar subjects.The reason image quality is so hard to get a handle on is twofold. First, there is currently no standard, objective scale by which to measure color accuracy in digital cameras. Second, all DPS devices use image compression technology to cram images into their limited memory. Depending on the algorithm used and the amount of compression applied, image quality can vary widely, even between devices using the same CCD sensor!

Let's look at some of the elements that contribute to image quality:

Color quality
Color quality is a complex and generally misunderstood topic. While it is probably less an issue for typical Point-and-shoot applications than for high-end studio cameras, the large differences we found between cameras suggest a detailed discussion of color quality would be useful.

Color quality is actually made up of two related but different parameters color purity and tonal balance. People tend to assume color errors can be corrected easily in an image-editing program such as Photoshop, and therefore tend to discount their importance. This is generally true of tonal errors, but errors due to color purity are virtually impossible to fix.

Gray balance and color accuracy (Tonal errors)
As mentioned above, tonal errors in a digital image are fairly easily compensated for, often with very gratifying results. Of these, gray balance is both the most dramatic and the most easily corrected. While a full treatment of gray balance is beyond this particular discussion, we can nonetheless cover a few key points, and see the effect relatively simple adjustments can have. For a brief lesson on how to perform gray balance corrections in Photoshop, look for a forthcoming tutorial titled "Black, White, and Gray: The Keys to Color" in the FAQ area of the Imaging Resource website.

As the name suggests, gray balance refers to balancing the red, green and blue channels across the entire tonal range, bringing gray tones in the image into neutral color balance. Obviously, if the blue channel of a camera responded more than the red or green channels, pictures would have an overall blue cast. While gray-balance problems are sometimes as simple as this, generally they are more complex.

In a digital camera, we care about color across the entire range of tonality, which means we need to concern ourselves with the gray balance in the shadows, the highlights and all points in between. In practice, it is neither practical nor necessary to make gray-balance adjustments point by point across the whole tonal range. Usually, adjusting the balance in the highlights, shadows and midtones brings dramatic results with relatively little effort.

In Photoshop, these adjustments can be made with the Image/ Adjust/ Levels function, and correction curves can be saved for fairly automated application to groups of images captured with similar lighting. The color above and at right show sample results. The image above is "as captured" by one of the cameras we've tested. The image on the right is derived from the same file, but has had a rudimentary gray-balance adjustment performed on it. Note how much "cleaner" and brighter the colors are, almost as if a layer of grime had been wiped from the image. While gray-balance adjustments can produce a dramatic improvement in color quality, they are by no means a panacea. Although proper gray balance can substantially brighten colors, it will not correct color impurities, resulting from poor-quality color filters. You also need to keep in mind that time spent adjusting gray balance is time lost to more productive work. A camera that looks like a bargain may prove otherwise if you have to spend 20 minutes color-correcting every image it produces.

Evaluating color purity and color accuracy
Important as it is, color accuracy is a slippery parameter to quantify. Rather than trying for some sort of absolute standard for color accuracy (which would likely be difficult to interpret anyway), we've opted simply to provide consistently exposed test images that would help you make your own subjective evaluations. The "Dave Box" target includes a Macbeth color chart as a well-established and reasonably consistent color reference, readily available at most well-stocked camera stores.

Dynamic range
Dynamic range measures how wide a range of subject luminance ( e.g., brightness) a sensor can accurately reproduce. Most digital cameras can capture a wider range of brightness values than the printed page can reproduce. The key issue is how good a job the hardware and software does of compressing the full range of scene brightness into a range that output devices can reproduce. This capability, whether in a camera or a high-end scanner, is called tone compression, and it is an important characteristic of high-quality input devices.

Tone compression is as much an art as a science, and different devices use different input-to-output curves to accomplish it. Generally, you want to avoid losing either the highlights or the shadows, yet still maintain reasonable contrast in the midtones. Visually, photographers are accustomed to looking for tonal problems in very light and very dark subjects. Traditionally, people shoot white porcelain objects to study subtle highlight detail and dark camera bodies or electronic equipment to look at shadow detail. This is a valid approach, but subtle differences are often difficult to detect on the final printed output. On screen, without an assist from Photoshop or some other image-editing program, the problem is even greater. Especially in the shadow areas, the tonal response of CRTs is very poor.

The way to really see what's going on is to use the Photoshop Levels control (or the brightness/contrast adjustments in consumer-grade imaging programs) to stretch the tonal range of the image in the areas you're interested in. For highlights, this is accomplished by moving the black-point slider well up into the midtones. This forces everything darker than the midtones on down to black, stretching the remaining tones over a wider range and emphasizing tonal differences in the brighter portions of the image. For shadow areas, reverse the procedure, moving the white-point slider down into the midtones. This forces everything from the midtones on up to white to go all the way to white, stretching the shadow tones across the full range and emphasizing subtle tonal differences there as well. The pictures below show this process applied to an image with deep shadow detail. In the first we see the the original image, and the display from a Photoshop Levels adjustment window. Note that all the image data are clumped on the far left, in the deep shadow. In the image itself (directly above), there's no apparent detail to be seen. In the next frame, we've moved the highlight slider far to the left until it's just touching the right-hand edge of the histogram "lump." See how this brightens the screen display, pulling up detail where none was to be seen. Finally, the last panel shows how the brightness levels of the image have been "stretched" to cover the full tonal range. The image is the same as in the middle panel; the change has just been made permanent in the file. Note that if the gray balance of the camera was inaccurate, these procedures would also show color casts in either the highlights or the shadows that wouldn't be obvious to the unaided eye. (In other words, don't be surprised if your shadows end up looking yellow or red or whatever after you've tinkered with them to this extent.)

The starting point: There's a range of tone there, but all clumped at the extreme shadow end of the range.
By moving the "highlight" slider down to lightest tone on the graph, we see the hiddin detail.
When the operation is complete, the available brightness values are spread across the full range of the display.

What do you look for in evaluating a camera's tonal range? Whether in the highlights or shadows, you want a camera to reproduce tonal variations smoothly, without obvious "tonal breaks," "posterizing" or "quantization" (three different terms all meaning the same thing). The most common defects occur when highlights wash out to white or shadows plug up, going black in regions where there is still some tonal variation in the subject that you'd like to retain. These faults are sometimes hard to see, but Photoshop will smoke them out every time.

For the DPS cameras, the "deep shadow" test of charcoal briquettes in a black box was almost too severe a challenge. Most of the devices we've tested could only just barely discern the presence of the briquettes, let alone show any significant detail. Nonetheless, you'll find clear differences in performance if you play with the camera files in an image-manipulation program.

Camera limitations
In discussing techniques for finding camera limitations that aren't visible to the naked eye, a natural question is, "Who cares?" After all, if you can't see it, why worry? The answer is that sooner or later, you'll need to "push" an image in some way, perhaps to open up the shadows, boost midtone contrast or hold back the highlights a little. When you do this, any tonal imperfections will be magnified, along with whatever detail you're trying to bring out. If you've never shot a photo that wasn't perfectly exposed, then congratulations, and just skip this entire section. If you're subject to the same slings and arrows as the rest of us mortals, though, you owe it to yourself to look carefully at what a digital camera will do before you invest your hard-earned money. With a film camera, you can always use a different type of film to deal with poor exposure, a contrasty subject, etc. With a digital camera, what you have is all you'll get.

Sensor noise
The characteristic that most limits the overall tonal range of a digital camera is noise in the sensor array. Without getting overly technical, a noisy sensor means you'll see "grain" or "snow" in solid grays or colors. This effect is particularly pronounced in shadow areas. If you download and play with images from this site, you can use the procedure described earlier to boost the shadows into an easily visible range. The darker steps of the Kodak gray scale will give you the most consistent reference to work with, but for subjective evaluation, the charcoal briquettes in the black box are perhaps the most useful. Sensor noise usually appears as "snow" in the image, much as you would see on a television set experiencing poor reception. Depending on the construction of the camera system, the noise may appear as randomly distributed monochrome or colored flecks. (Certain sensor irregularities may result in some patterning to the noise flecks, but usually the distribution will be fairly random.) You're likely to have gotten a good look at sensor noise in the previous experiment, while looking at shadow detail. In the process of pulling the shadow tones up into the visible range, you also would have greatly increased the visible effect of any sensor noise that might have been present. The illustration here shows a shadow detail sample from a studio camera with high noise and a marked color cast in the deep shadows. In DPS cameras, noise effects are largely masked by the JPEG image-compression process. Noise is nonetheless a primary limitation, even if we don't see the effects directly.

Sensor and optical artifacts
As we discussed earlier, each image captured by a digital camera is actually three separate images (red, green and blue). In order for your picture to look right, all three of these images must be kept in perfect registration. If the registration between the tricolor images is off at any point, the result will be a color "artifact." (Artifact here is just another word for something in the image that came from the camera, rather than the scene.) In practice, most DPS cameras use "striped arrays," so the color registration is both fixed, and inherently offset. This gives rise to a particular type of color artifacts.

"Striped" sensor arrays
Ideally, a digital camera would arrange to have separate red-, green- and blue-sensitive sensor elements staring at each pixel-size area of the scene. In most cases, this is prohibitively expensive, even for high-end studio cameras. A popular, low-cost way of approximating this capability in a digital camera is to "stripe" the sensor array with microscopic color filters. The result, as shown here, is generally an array of RGB color triplets, much as you can see if you look closely at a color TV picture tube (or computer display CRT) when it is operating. This approach trades off resolution for single-shot capture capability, but carries with it other penalties as well.

Array striping artifacts
The most significant limitation of striped sensor arrays is that they are prone to generating color moiré and other artifacts when viewing small, high-contrast objects. (This situation frequently arises in product photography, where fine black type on labels can cause these kinds of problems.) In the Davebox image, you may be able to see such artifacts around type, or as a colored moiré on the resolution target with the repeating patterns of vertical lines. In print applications, as long as the file isn't magnified too much, these artifacts are somewhat hidden by the half-tone printing process. If the image is magnified at all, though, these artifacts can be quite evident. In multimedia applications, these effects are almost always visible. In the resolution test images appearing on the Imaging Resource website, you'll find several places where closely spaced parallel black and white lines produce color artifacts when photographed by various cameras. These false colors are caused by an interference pattern between the spacing of the lines in the test target and the spacing of the red-green-blue filters on the surface of each camera's imaging array. Differences in compression techniques and optical systems between cameras make them more or less prone to producing such artifacts. While this pattern represents a particularly severe test, a camera that has trouble with it is also likely to have problems with things like venetian blinds, etc.

Overexposure "blooming"
A final limitation of CCD sensors is their reaction to severe illumination overloads. This is the reason for the inclusion of the shiny pot lid in the Dave Box target, which reflects light sources back into the camera lens. In the face of extremely high light overloads, some CCDs will "leak" charge from the overexposed elements into adjacent cells. This phenomenon is called blooming, and various methods are employed to prevent it. It most frequently shows itself as a colored fringe around specular (shiny) highlights. Frequently, the sensor will bloom differently in each of the red, green or blue channels, producing the colored fringes where one channel has bloomed more than the others. The impact of this for your particular work will depend on its nature: If you intend to do a lot of photography of chromed auto parts, blooming could be a big problem. On the other hand, it would be a complete nonissue in photos of bath towels. In any event, little if any blooming was evident in the DPS cameras tested.

Resolution(!)
Resolution is one of the most misunderstood, misrepresented and confusing parameters in the entire field of digital photography. It is also one of the hardest characteristics to specify in a precise, objective manner, particularly for the digital Point-and-shoot cameras. One of the most important things to understand about resolution in digital cameras is that pixels are not resolution! ALL of the entry level Point-and-shoot cameras use large amounts of image compression to squeeze a reasonable number of images into their limited RAM. The complexity and variety of these compression schemes makes it nearly impossible to arrive at any meaningful correlation of pixel count with the actual ability of a camera to resolve detail. Once again, it's really important to look at the test images to determine how each camera actually performs.

It was to resolve just this conundrum that the "WG-18" resolution test target was developed. Developed by an international committee of imaging scientists, it contains a multitude of test patterns designed to reveal exactly how well a given camera can resolve fine detail. A full treatment of the target is well beyond the scope of this article, but even the unitiated can look at images of it taken with two different cameras and rapidly ascertain how well or poorly they perform in relation to each other. As always, let your own eyes be the judge, and use the Comparometer(tm) to perform your own side-by-side comparisons.

Image Compression: 10 pounds into a 5-pound bag
At first glance image compression looks like magic. In fact, at second glance, it's pretty easy to convince yourself it is magic! How else would you describe a technology that can cram sixteen 1-megabyte images into a single megabyte of actual storage? Of course, nothing comes for free, not even with the spiffiest of high technology: While you can get away with pretty substantial amounts of image compression without people noticing, the level used in many DPS cameras is well beyond that point. (Although, as memory has become cheaper, manufacturers are finding it to their advantage to trade more memory usage for less compression and therefore better picture quality.) Nonetheless, different devices succeed to varying degrees, depending on the details of their particular compression techniques. The key to usable image compression is to throw away "unimportant" information in the images, and to take advantage of local areas of similarity within each image. For instance, you don't need 24 bits of information to tell you what color a particular piece of an image is, if it happens to be about the same as the pieces on either side of it: Just record the differences! If the differences are small, the information you'll need to store will be small as well. Also, the eye tends to be much more sensitive to certain kinds of detail in an image than to others. If we can find a way to throw out only the sort of detail our eyes aren't very sensitive to, we can reduce the size of the file without our eyes noticing what we've done.

This is the essence of all image compression schemes, but as you may suspect, the concept is easier to describe than to implement. While standard techniques exist, there is a lot of latitude as to how manufacturers can tweak them to produce the best results for their particular camera. Recent advances in compression technology go beyond standard techniques, and some manufacturers have taken advantage of them to capture finer detail with little increase in memory usage.

When looking at the images on this site, the effects of image compression generally can be seen as a "blockiness" in areas of fine detail and high contrast. Look around the edges of objects contrasting strongly with their backgrounds: You'll see errors introduced by the compression process in the form of square blotches at the corners, and "stairstepping" down diagonal edges. (The tree leaves shot against the sky in the "House" shot are a particularly severe test of compression techniques.) Note too, that it is important to look at areas with much "flatter" contrast as well: Sometimes, manufacturers tweak the compression methods in ways which throw out too much information in areas with subtle contrast differences. Look for example at the subtle shading in the shingles on the roof of the house in the "House" test shot: Cameras using more compression tend to lose the shading and reduce the roof to blocky chunks of flat color.

Digital Point & Shoots: Today into Tomorrow
Digital Point & Shoot cameras have made dramatic strides in a very short period. At the time of this writing (early 1998), the higher-end units were rapidly approaching parity in image quality with conventional film-based point & shoot cameras. As this trend continues, the day of "filmless" photography will truly arrive.

Reader Comments!
What are your experiences in evaluating digital cameras?, Leave a note, or see what others have contributed. (No comments yet - add your own!)