Part 1: Budget, SLR vs. point-and-shoot, and resolution
Part 2: Exposure controls (program modes, shutter speed, aperture, white balance, and ISO)
Part 3: Light meters, lenses, image stabilization, autofocus, and flash
Part 4: Viewfinders, shutter lag, start-up time, histogram, exposure lock, and autobracketing
Part 5: Output type, noise, color, sharpness, JPEG vs. Raw.

Over the previous four installments of this column, I presented you with a barrage of questions and options related to choosing a digital camera. If you've been following along, you've asked yourself what type of shooting you do, what features you want, what you need in terms of size and weight. After answering all of these questions, you should have narrowed down the field of possible candidates to just a handful of cameras. You're now at the very last stage: the final choice.

The good news is that making the final choice may not be too hard, because, in the end, one question should outweigh all others when selecting a camera: How good are the images it produces? The bad news is that, if you're already favoring a particular camera for its features and design, you may have to nix it and go with a secondary camera that yields better image quality.

Image quality is, of course, a subjective topic -- one person's hideous noise might be another's beautiful texture. Nevertheless, with a little bit of attention, you can identify problems that might vex you later. Plus, identifying these problems now can help in your image-editing efforts when you begin shooting with the camera.

Print or Monitor?
Before you can begin comparing image quality, you've gotta have some images to evaluate. Looking at them in the camera's LCD isn't enough because color and contrast are rarely accurate on a camera LCD.

To evaluate image quality, you need to get images out of the camera and back to your computer. The easiest way to do this is to take a memory card to the camera store and fill it with images. If you're evaluating cameras that use different memory card formats, you'll need to invest in several cards, but media is cheap right now.

Laptop owners have the option of taking their computer to the store and loading images in directly from the camera.

If neither of these options work for you, several camera review sites regularly post full-resolution output from review units. These images usually cover a much greater range of lighting and shooting conditions than you'll find in a typical camera store.

You'll perform your first evaluations on a computer monitor, and there's much to learn from these assessments. However, it's also possible to get hung up on issues that may not be a problem in your final output. For example, when you can zoom in to look at the individual pixels of an 8-megapixel image, you're likely to find some troublesome artifacts. Whether these artifacts will be visible in an 8-x-10-inch glossy is another question.

So, when assessing an image, think about your final output. If you mostly shoot for on-screen viewing, consider the typical dimensions and file size, and do a round of evaluations at that size. If you predominantly shoot images for print, output your samples using your typical printing technology, media, and print sizes. Many troubles and artifacts visible onscreen vanish when resized and output using a particular printing technology.

Noise Levels
When comparing output from different cameras, the issue you're most likely to notice first is noise. Noise is roughly analogous to grain in a piece of film and, like grain, it's not necessarily a bad thing, as it can add texture and atmosphere to an image.

However, sometimes digital camera noise has a more intrusive character than film grain, manifesting as speckly colored patterns rather than an underlying texture. If you decide you want texture and atmosphere, you can easily add noise later in an image editor, so it's better to get a camera with very low noise.

Noise is almost always more prominent in the shadow parts of an image, and this is the first place to look when evaluating an image.

There are two types of noise: chrominance noise and luminance noise. Luminance noise is the preferable of the two, as it looks more like film grain. Where luminance noise is simply a variation in the brightness of the pixels in the image, chrominance noise is a variation in the color -- sometimes an extreme variation. If you see brightly colored blue, red, or even purple pixels, then your camera is exhibiting chrominance noise (Figure 1).

Figure 1: This image has some noise issues.

Because noise gets worse as you increase the camera's ISO settings, do noise tests across the camera's ISO range. Because ISO is a valuable exposure parameter, you'll want to carefully consider noise levels at higher ISOs. If possible, take some long-exposure images, as noise also increases with exposure time.

Color Troubles
Color reproduction is probably the most subjective of all image quality considerations. What you think of as beautiful, saturated color, someone else might perceive to be garish, velvet-Elvis-painting color. So ignore any "experts" and just go with the color you like. As with noise, though, it's still good to know whether a particular camera has certain color tendencies. Color issues to be aware of include the following:

• Color casts. Some cameras produce images with a particular cast -- an overall color tone that makes the image appear as if it were shot through a colored filter. The cast may affect only a particular part of the image -- shadows might appear too blue, for example.

• Bad white balance. A camera calibrates its idea of color according to the current light. This process, called white balancing, can sometimes go awry, resulting in images with out-of-whack color. A bad white balance is similar to a color cast, except that white balance problems usually affect the entire color range, not just the shadows or highlights. Removing a bad white balance can be difficult when shooting JPEG files. However, if you shoot in the Raw format, you can correct white balance later.

  Most cameras offer a choice of automatic and manual white balances. Because you'll most often use automatic white balance, closely scrutinize images shot in auto white balance mode.

• Chromatic aberration. All cameras are subject to a weird type of artifact referred to as "purple fringing" (see Figure 2). Sometimes, this chromatic aberration is caused by poor lens optics; at other times it's the result of pixels on the camera's image sensor overflowing with light and corrupting neighboring pixels. Chromatic aberrations usually occur only when shooting high-contrast subjects in bright light -- you'll often see it appear along the edges of leaves against a bright sky -- and even then usually only at extreme focal lengths. Chromatic aberration that's readily apparent on-screen is often invisible in print, and it's usually not too difficult to remove, if need be.
  
  
  Figure 2: Some cameras are particularly susceptible to chromatic aberration troubles.

• Inaccurate color. A camera can sometimes yield colors that are dead wrong but that still look great -- maybe even better than the original. What's more, a camera might change colors consistently. Some cameras, for example, always pump up the saturation. If you're a stickler for accuracy, evaluate color with this in mind. If you simply want consistently attractive images that require little editing, accuracy may not be crucial.

Good Is in the Details
The amount of sharpness and detail in an image is the result of several camera factors: a camera's resolution, a lens's quality, and a camera's built-in sharpening routines.

The tricky thing about assessing detail is that you may not know that detail is missing. Figure 3 below shows the same scene shot with the same camera but using two different lenses. Though neither image looks bad, there's a marked difference in detail.

Figure 3: Although there's nothing wrong with the upper image, once you see the scene shot with a better lens, it becomes obvious how much extra detail is there to be photographed.

The easiest way to assess detail when comparing cameras is to examine identical images side-by-side. For this kind of testing, go to the camera store and shoot the same scene with all of your candidate cameras. Even if it's simply a shot of the cash register, that's enough to give you an idea of differences in detail and sharpness.

Because in-camera sharpening can greatly affect the quality of an image -- sometimes for the better, sometimes for the worse -- shoot images with varying degrees of sharpening applied. Look for bright and dark halos around high-contrast areas of each image.

Some cameras oversharpen their images by default. If that's the case with a model you like, make sure it lets you reduce sharpening. Other cameras -- especially digital SLRs -- apply very little sharpening to their images. When you first pull images out of these cameras, they may look very soft. This is an intentional part of the camera's design. You can't remove sharpening once it's been applied, so SLR manufacturers usually opt to leave you in control of the sharpening process.

Bad Exposure
You'll use your camera's automatic metering for almost all of your images, so get a sense of how well it exposes in difficult lighting situations. This is a great time to turn to camera review Web sites. Pay particular attention to how well the camera meters in low light, in complex mixed lighting situations (lots of bright and dark in the scene), and when shooting with flash, as well as a model's overall consistency of metering.

Some cameras exhibit lens distortions when shooting at the extreme wide or telephoto ends of their focal ranges. Though these problems can be corrected in post-production, it's worth keeping an eye out for them.

JPEG vs. Raw
By default, most digital cameras store their images as JPEG files. JPEG strikes an excellent balance of quality and compression. However, because JPEG is a lossy compression scheme, images can be visibly degraded by the compression process. What's more, while most cameras capture 10 to 12 bits of color per pixel, JPEG files can hold only 8 bits of data per pixel. When a camera stores an image as a JPEG file, it has to throw out a fair amount of color data. That's why some newer cameras also include the ability to save images in an uncompressed format, such as Raw.

For the greatest image quality and image-processing flexibility, I recommend a camera that saves files in the Raw format. If you plan on shooting in JPEG (either all or just part of the time) then you should examine your test images for JPEG artifacts -- regular, square blocky patterns throughout an image -- when shooting at the camera's highest quality level.

Now Go Buy that Camera!
Most modern digital cameras yield very good images. There are differences, and some results might be more to your taste than others, but you'll probably see acceptable results from most cameras.

When considering image-quality issues, remember to factor in price. The model you're considering might not capture images as good as what your neighbor gets with his $2,500 digital SLR, but if you're only aiming to spend $500, you should expect to take a hit in image quality. For this reason, compare similar cameras, and don't be hung up on how your candidate cameras compare to models you can't afford.