|
||||||
|
|||||||
|
Digital Camera Reviews |
|||||||||
|
Finding the Right Digital Camera Computing Platform Image Quality Onboard Image Capacity Reusable Memory Cards Exposure Versatility The first step in evaluating a digicam's exposure system is to look at its
"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 digital point-and-shoot (DPS) cameras have equivalent ISO
ratings in the 80-200 range, although some go well beyond these limits. In
addition to the ISO rating, look for a wide range of shutter speeds and lens
apertures. Typical numbers for shutter speeds are from 1/1,000 to 1/30 second,
and for lens openings from f/2.8 to f/16. (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 these two sets of numbers the better. Flash Capability The Feature Set Camera Lens Wide-angle photography is another important capability. It can be particularly useful for indoor family activities, where shooting conditions may be cramped, or for real estate or insurance photography. Again, some cameras have aftermarket accessories available to expand their field of view, so you will want to take this into account when making your purchase decision. Many digital 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) 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, and well worth the investment. Self-Timer Battery Life Another important question to ask about a potential digicam purchase 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 many fold. 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, make sure you change or replace batteries in the field. Digital Camera Reviews
External Power Adapter Included Software Image Download Image Manipulation File Formats Cataloging 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 Image Quality 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! So, let's look at some of the
elements that contribute to image quality: Color quality 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 & Color Accuracy (Tonal Errors)
When reviewing the color accuracy of a digital camera, it is important to pay
attention to the colors across the entire range of tonality, which means you
must look at 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 at each step across the whole tonal range. Usually, adjusting the
balance in the highlights, shadows, and midtones brings dramatic results with
relatively little effort.
While gray-balance adjustments can produce a dramatic improvement in color
quality, they are by no means a panacea. Proper gray balance can substantially
brighten colors, but 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 & Accuracy Dynamic Range Tonal 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 original image, and the display from a Photoshop Levels adjustment window. Note that all of the image data is clumped on the far left, in the deep shadow. In the image itself, 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 after you've tinkered with them to this extent.)
Digital Camera Reviews Finding "hidden" images in deep shadows
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 Sensor Noise If you performed the earlier experiment, you're likely to have had a good
look at sensor noise while looking at shadow detail. In the process of pulling
the shadow tones up into the visible range, you would have also greatly
increased the visible effect of any sensor noise that might have been present.
The illustration below 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 & Optical Artifacts Striped Sensor Arrays 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" Resolution(!) It was to resolve just this conundrum that the "WG-18" resolution test target was 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 uninitiated can look at images of it taken with two different cameras and rapidly ascertain how well or how poorly they perform in relation to each other.
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 detail
our eyes aren't very sensitive to, we can reduce the size of the file without
our eyes noticing what we've done.
When looking at sample images on the Imaging Resource website, 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 that contrast
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. Note too, that it is important to look at areas with much
"flatter" contrast as well. Sometimes, manufacturers tweak the compression
methods in ways that throw out too much information in areas with subtle
contrast differences. These cameras tend to lose the shading and reduce even
toned areas to blocky chunks of flat color. Digital Point & Shoots: Today into Tomorrow
|
Home ▪ Buy Digital Cameras ▪ Camera Bags ▪ Digital Camera Battery ▪ DVD Camcorders ▪ Mini Digital Camera ▪ Digital Photography
As
the name suggests, gray balance refers to bringing gray tones in an image into
neutral color balance. This is done by balancing the individual red, green, and
blue channels across the entire tonal range. Of course, if the digital camera's
blue channel responds more than the red or green channels, the pictures would
have an overall blue cast. While gray-balance problems are sometimes as simple
as this, generally they are more complex.
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 charts 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. 
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 the Imaging Resource 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.) 
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.
The
most significant limitation of striped sensor arrays is that they are prone to
generating color moirés and other artifacts when viewing small, high-contrast
objects. (This situation frequently arises in product photography, where fine
black type on labels can cause problems.) In the Davebox images posted on the
Imaging Resource website, 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. 
A
final limitation of CCD sensors is their reaction to severe illumination
overloads. This is the reason for including a shiny pot lid in the Dave Box
targets on the Imaging Resource website, because it 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 photograph a lot of chromed auto parts,
blooming could be a big problem. On the other hand, it would be a complete
non-issue in photos of bath towels. 
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 memory. The
complexity and variety of these compression schemes makes it nearly impossible
to arrive at any meaningful correlation of pixel count with the camera's ability
to resolve detail. Once again, it's really important to look at test images to
determine how each camera actually performs.
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.