Digital Photography For What It's Worth

Exposure strategies for digital cameras with priority and fully manual exposure controls

On this page—

• Who's in Charge Here?

• Stops, Shutter Speed and Aperture

• Reciprocity and Exposure Values

• Exposure Compensation Controls

• Exposure Strategies

• Guiding Constraints

• Recording Mode—Critical Pre-Exposure Digital Decisions

  • ISO and Noise

  • Color Mode
  • Resolution
  • Compression and Color Interpolation
  • In-Camera Sharpening
  • White Balance

• Resolving Power

• Depth of Field (DOF)

• Hyperfocal Technique

• Motion Management

• Dynamic Range

• Tonality

• The Zone System

• Manual Exposure Rules of Thumb

• Advanced Digital Exposure Considerations

• Editor's Note

• Editorial: The Proper Role For Rules

• References and Links

See also the Hyperfocal Checklist

Last updated July 27, 2004

Who's in Charge Here?

This article's all about seizing control of exposure. If you don't, the camera will, and the result won't always be what you had in mind. It may or may not be worth taking control of exposure at the company picnic, but when the stakes or aims are high, you can improve your odds substantially by stepping in.

Inner Vision

Your camera's built-in metering system will seldom have trouble coming up with technically "correct" exposures. In some situations, that's all you'll need, but don't expect the meter to offer exposure settings optimized for your photographic goals in a particular scene. Those goals are almost always driven by things the camera knows absolutely nothing about—things I'll lump together under the term inner vision:

• what you see in the scene, in the broadest sense of seeing

• what you deem to be the scene's most important elements

• the emotional response the scene evokes in you

• what you'd like other humans to see and feel in your photograph

Capturing that inner vision is the ultimate goal of any serious photograph, but the gulf separating what the human brain-eye system sees and feels in the scene and what even the most sophisticated camera senses in the frame is wide and deep. By taking control of exposure and composition, you actively manage that gap to align your vision and the camera's capabilities as best you can. Reducing the gap may make a huge difference, even if you can't fully close it.

Here we'll tackle exposure and related issues with a digital slant. For the composition piece, which is no different on the digital than on the film side, I heartily recommend Brian Peterson's book Learning to See Creatively—How to Compose Great Photographs and practice, practice, practice.

Issues to Face

Your camera knows nothing of the motion and depth of field challenges you face. Left to its own devices, it will render whatever it's metered in a medium tone, regardless of the tonality you actually see or want. And it may well put what it knows of the performance of its own lens ahead of considerations far more import to you. To cut the best deal across all these critical fronts—tonality, depth of field, motion management and resolving power—you're going to have to step in. 

If your camera's built-in TTL metering system offers exposure locking and reasonably tight spot metering, as do most current higher-end digital cameras, its exposure suggestions can serve as a solid starting point for all your exposure judgements. Your inner vision and the rules of photography will guide you from there. 

 Only you can see the picture you've envisioned for the scene before you, and only you can make the technical trade-offs needed to capture it with the equipment at hand.

If you're already familiar with concepts like stops, shutter speed and aperture, click at left to skip ahead to exposure strategies now. To review exposure basics, read on.

Stops, Shutter Speed and Aperture

For anyone who needs it, the next few sections offer a leg up on the definitions and science behind basic photographic terms like stop, shutter speed, aperture, f-number, reciprocity, exposure value (EV) and exposure compensation (EC). Without a solid understanding of these basic building blocks, the rest of the article won't make much sense. 

Stops

In photographic parlance, a stop or full stop is simply a factor of 2 change in the amount of light fed to the image receiver in the course of image capture. Increasing exposure by a stop doubles the light input, and decreasing exposure by a stop halves it.

 Think of a stop as a unit of exposure or light input divorced from the details of how it's achieved. 

On film cameras, stops are still real click-stops on mechanical shutter speed dials and lens aperture rings, but electronic controls have largely robbed the term "stop" of its mechanical meaning on the digital side. 

But even disembodied stops remain useful because they correspond perceptually to more or less evenly spaced gradations of light and dark. Stops give us an easy way to talk and think about the brain-eye system's final linear response to exponential variations in light input. (The auditory and visual systems both deal in logarithms of input amplitudes.) Film and digital camera designers work hard to emulate this log-linear response so that photographers can rely on their natural sense of tonality.

Since stops correspond to powers of two in light input, they're relatively easy to figure in your head. They also nicely scale the effective dynamic ranges afforded by most film and digital cameras in a manageable number of log-linear steps—usually 3-8 stops.

 When photographers speak of "opening up a stop" or "stopping down", they're usually talking about adjusting exposure by the most appropriate means, not just via aperture changes. That's the way I'll use such terms below.

Shutter Speed

 For a given scene and lens, the amount of light fed to the image receiver is completely determined by the product of exposure time and aperture area.

Because of the way the exposure duration is actually controlled inside film cameras, the term shutter speed is often substituted for exposure time, but to call for an exposure of 1/250 sec is clearly to specify a duration, not a speed. Strictly speaking, the concept of shutter speed doesn't even apply to the subset of digital cameras with electronic rather than mechanical "shutters", but the terminology is firmly entrenched and will likely remain so.

Since exposure varies linearly with exposure time, a stop in speed is simply a doubling or halving of exposure. The shutter speed dials on most 35 mm SLRs come with click-stops at 1, 1/2, 1/4, 1/8, 1/15, 1/30, 1/60, 1/125, 1/250, 1/500, and perhaps even 1/1000 sec and beyond. Many digital cameras emulate this familiar sequence of full stops, and some higher-end digitals like the Oly Camedias insert partial stops as well.

Aperture and f-stops

Opening the lens aperture a full stop by definition doubles the amount of light allowed to reach the image receiver, be it film or CCD. Stops of aperture are often called f-stops.

Doubling the light input via aperture requires a doubling of the aperture area. For the roughly circular iris apertures still in common use—even in digital cameras—that means opening the iris diameter by a factor of sqrt(2) or ~1.4.

F-numbers

Aperture diameters have come to be expressed as fractions of lens focal length (f), a notation that tremendously simplifies exposure considerations by divorcing aperture designations from the particulars of lens focal length. 

For a digital lens with an actual focal length of 14 mm, f/2.8 denotes a 5 mm aperture, where the 2.8 is known as the f-number. On a 35 mm SLR with its zoom lens set at 140 mm, f/2.8 means a 50 mm iris opening (that's one reason fast telephoto lenses are so large). The physical iris openings differ greatly in these two examples, but at f/2.8, both lenses make the same contribution to exposure. F-numbers corresponding to full stops are roughly whole-number powers of the square root of 2. 

In the table below, the f-numbers corresponding to full f-stops appear in bold; the intervening apertures represent the so-called "1/3" and "2/3" stops. For now, the seemingly arbitrary "stop numbers" shown are useful for figuring the number of stops (exposure doublings) between any two aperture settings, but they'll later reappear as exposure values. Note their negative logarithmic relationship to the relative aperture area.

Technical Note: The conventional shutter speeds and f-numbers found in the table and text above deviate slightly from the numbers that would follow from a strict adherence to the physics underlying the table. Whether these minor discrepancies reflect practical conveniences, rounding errors, conventions or something else, I don't know, but no one seems to care. They're now firmly entrenched in photographic practice.

Reciprocity and Exposure Values

One would like to think that the same amount of light delivered to the image receiver would result in the same exposure, regardless of the rate of delivery. In other words, blasting the receiver with a certain total dose of light in a short time should have the same effect as dribbling in the same dose over a longer duration. This concept is known as reciprocity, and fortunately, it holds up under most circumstances. 

Reciprocity means that you can safely rely on a perfectly reciprocal relationship between aperture and exposure time: If you open up one full stop in aperture to double the aperture area while halving the exposure time, the resulting film density or CCD charge remains the same. The exposure value table below nicely demonstrates the reciprocity relationship.

Using Reciprocity

 Once you decide to take control of exposure, reciprocity becomes one of your most important tools. It allows you to work your way from a technically correct but often artistically challenged exposure determined by a light meter to an equivalent exposure carefully matched to your photographic intent.

By reciprocity, f/2 @ 1/500 sec, f/5.6 @ 1/60 sec and f/11 @ 1/15 sec are all equivalent exposures at EV = 11, but only the first would be suitable for stop-action shots at a basketball game. The first two settings could be handheld, but the third would probably fall prey to camera shake in anyone's hands.

Reciprocity Failure

Unfortunately, for film at least, reciprocity tends to break down at very long (multisecond) and very short exposure times such that greater than expected increases in exposure time become necessary to compensate for a given decrease in aperture—hence the term reciprocity failure.

To what extent reciprocity failure might apply to digital cameras, I'm not sure. CCDs are said to be very linear devices, and their exposure-charge curves probably don't have much of a toe. Clipping of the shoulder of the exposure-charge curve at high exposures due to blooming amounts to a reciprocity failure of sorts. So does the draining off of excess photoelectrons to mitigate blooming in many CCDs.

Exposure Values (EVs)

Exposure values provide a convenient way to quantify available light intensity and therefore exposure. In the Additive Photographic Exposure System EV table below, exposure value (EV) is defined as the sum of the respective stop numbers corresponding to the aperture and exposure time of interest, with one unit of EV corresponding to one stop of exposure.

 Think of EV as a measure of available light intensity as the camera's meter sees it, divorced from the details of how the camera might go about acquiring it.

A bright scene metered at EV 12 reflects more light to the camera than a darker scene metered at EV 8 by a factor of 16, or 4 stops. Conversely, a camera metering a scene at EV 8  is 4 stops more sensitive (requires 16 times less light input for proper exposure) than a different (or differently adjusted) camera metering the same scene at EV 4. The difference in sensitivity in the latter example might well reflect a difference in ISO settings.

As available light instensity and therefore metered EV increase, the exposure called for by the meter (i.e., the amount of light to be admitted by the camera) must decrease in order to maintain proper stimulation of the image receiver. In other words, as available light intensity increases, the camera must either stop down the aperture or decrease exposure time or both to avoid overexposure. 

Note that this standard definition of EV runs counter to the way exposure compensation (EC) controls are typically marked. When you increase EC by +1.0, you're forcing the camera to admit twice as much light as the meter suggested. But that's what the meter would have called for if the scene had somehow darkened by EV -1.0. EV and EC are measured in the same units (stops) but run in opposite directions.

Luckily, none of that makes much difference in common practice. What really counts most of the time is what happens at constant EV—the one corresponding to the correct exposure determined by your meter:

 In the absence of reciprocity failure, all aperture and shutter speed combinations yielding the same EV produce equivalent exposures. 

This very powerful result allows you to optimize your technique for DOF, resolving power, motion control, tonality or whatever's most important at the scene, without compromising exposure. That's what reciprocity is all about.  

Using the Additive EV Chart

To get the EV for any given exposure from the chart below, simply add the stop numbers corresponding to the desired aperture and exposure time. Thus, the EV for f/2.0 @ 1/8 sec = 2 (from the aperture column) + 3 (from the time column) = 5. 

* Technical note: The stop numbers in the table above are actually base-2 logarithms of the reciprocal of exposure time and the square of the f-number, respectively. The addition of stop numbers reflects the fact that  

where a is the aperture f-number (again, the 2.8 in f/2.8), t is the exposure time in seconds, and log₂ is the logarithm to the base 2. The coefficients and signs in the last version of the equation are simply built into the table for convenience. (For those rusty on their base-2 logarithms, log₂ (1/x) = -log₂ x, log₂ 1 = 0, log₂ 2 = 1, log₂ 4 = 2, log₂ 128 = 7, log₂ 1024 = 10, and so on.) The conventional exposure times and f-numbers listed approximate whole-number powers of sqrt(2).

EV Example

Suppose you select f/2.8 in aperture-priority mode and your camera meters a shutter speed of 1/250 sec. From the table, EV = 3 + 8 = 11 for f/2.8 @ 1/250 sec. Now you can move to any other EV = 11 combination that fits your needs—for example, f/2 @ 1/500 sec for better motion resistance, f/4 @ 1/125 for better resolving power or f/5.6 @ 1/60 sec for greater depth of field—and still get the same exposure.

Priority Metering

Note that this constant-EV calculation is precisely what priority metering does for you automatically: The camera meters the scene to determine the proper EV. The firmware then works to maintain that EV as you take control of either aperture or shutter speed to optimize your technique.

The biggest risk with priority metering lies in the fact that you can easily and unknowingly take exposure beyond the camera's ability to follow—for example, by setting a fast shutter speed in shutter-priority mode in low light requiring an f/1.4 aperture when f/2 is the widest aperture the camera can deliver. When my C-2020Z's main LCD is on, the parameter I'm controlling turns red when the camera can't hack a suitable setting for the parameter it's controlling. When the LCD's off, I can proceed unaware that the camera's fallen off the wagon.

 Only vigilant exposure monitoring can keep you from straying outside your camera's exposure envelope in the priority modes. 

To guide manual exposures using the firmware's constant-EV calculator, I sometimes duck into a priority mode temporarily to get the camera to work out equivalent exposures for me. Once I see a combination close to what I'm after, I return to manual mode, dial it in, make the necessary adjustments and shoot.

I find it well worth the battery hit to keep my LCD and exposure display on whenever I'm using priority or manual exposure.

Exposure Compensation Controls

Exposure compensation or exposure correction (EC) controls provide an easy way to bias an exposure by 2-3 full stops up or down from the camera's metered aperture and shutter speed, usually in 1/2- to 1/3-stop increments. EC is particularly useful for manual bracketing and for overriding the camera's exposure theory in priority modes, where EC adjusts only the exposure setting left to the camera's control. In auto or program mode, EC again allows intentional under- or overexpose relative to the firmware's exposure strategy, but I have yet to figure out how the bias gets apportioned between aperture and shutter speed in auto mode. My Oly C-2020Z doesn't support EC in manual mode, probably because it doesn't make much sense in that context.

What good is EC? For starters, many digital cameras behave like color slide film—the best images are often slightly underexposed, particularly when bright scene elements are present. EC is the fastest and simplest way to underexpose. In bright sunlight, my C-2020Z tends to do its best work at EC -0.3 or -0.7. With EC and a little effort, you can easily feel out your own camera's exposure sweet spots, but count on variation with photographic conditions, as dpFWIW contributor Tom Lackamp details in his take on digital exposure below.

In landscape and close-up work, depth of field requirements typically dictate a specific aperture, but what if tonality requires an EV different from the one your camera deems appropriate? If the desired EV is less than 2-3 stops from the meter's EV, EC makes it simple to go there in aperture-priority mode without altering the aperture and without resorting to full manual exposure.

Using EC

If you're unfamiliar with EC, the fastest way to learn is to play around with your EC control and watch its effect on exposure settings. (Most cameras display exposure settings on their rear LCDs if nowhere else, but you may need to half-press the shutter release to update them.)

Technically speaking, the EC controls in most cameras are calibrated in negative EV units, presumably to avoid confusing the preponderance of owners tempted to use them but unaware of the formal definition of EV. On every EC-enabled camera I've ever seen, digital or otherwise, for each positive unit of added EV, exposure doubles, and for each negative unit, exposure halves. That works for me, but as you run down the standard EV table above, just the opposite obtains—for every positive unit of added EV, exposure drops by a half.

Perhaps an example will make this less confusing. Say your camera's in aperture-priority mode at f/4 and the meter sees an EV of 12, which calls for a shutter speed of 1/250 sec for a proper exposure. Since you've fixed the aperture, if you set EC = +1, you'll get f/4 @ 1/125 (double the time, EV = 4 + 7 = 11), and if you set EC = -1, you'll get f/4 @ 1/500 (half the time, EV = 6 + 7 = 13). 

Still confused about EC vs. EV? Well, at least now you'll feel justifiably so.

Exposure Strategies

Now that we've examined the physical basis of exposure and the means for controlling it, let's talk strategy.

First of all, there's nothing wrong with using the automatic or program exposure mode in your digital camera—provided it gives you the image you're after. Automatic metering systems in today's cameras are very adept at coming up with reasonable exposures based on available light, various camera settings and known properties of the camera's lens or sensor. But blind acceptance of automatic exposures amounts to playing dice on the artistic, inner vision side of the equation, with the house odds stacked against ending up with an optimum exposure. Of course, an EV different from the meter's is out of the question without intervention.

The trick, of course, is to learn when and how to depart from the camera's inclinations, and that's where tonality,  motion control, reciprocity, exposure compensation, and manual exposure come in. The very effective semi-automatic aperture- and shutter-priority exposure modes available in many digital cameras play the reciprocity game for you: You seize control of aperture or shutter speed, and the camera varies the other automatically to maintain the metered EV within the camera's exposure envelope. Often, this is all the control you need to get the shot in your mind's eye, but higher-end digital cameras with exposure compensation and fully manual exposure allow you break the bonds of camera-imposed exposures to reach the tonality and motion control you had in mind.

A Stepwise Approach to Exposure

Some master photographers can divine exposures with amazing accuracy without the aid of a meter. Others prefer to rely on calibrated external meters and manual exposure control. For the rest of us, it's perfectly reasonable to use the camera's automatic metering as a starting point—provided one can handle the inevitable exposure trade-offs ahead.

Here's the overview from 35,000 feet:

• Decide how you want to handle basic pre-exposure issues like resolution, compression level and in-camera sharpening, preferably through testing.

• Starting with the camera's metering, use reciprocity and filters as needed to achieve just the right combination of ISO, aperture, shutter speed to make the picture in your mind's eye using the guiding constraints detailed below. A priority mode will often suffice.

• As needed, escalate to exposure compensation (in a priority mode) or even manual mode to override the camera's meter to achieve proper tonality.

And here's the blow-by-blow approach:

1. First, set ISO, choosing the lowest setting feasible for the task at hand. Action shots are one of the best reasons to venture beyond minimum ISO. It's generally best to avoid camera settings that allow the camera to change ISO.

2. If you already know that aperture will be your critical setting—say, in landscape or close-up work—dial in your aperture in aperture-priority mode and work from there. If shutter speed is key—as it would be in action shots, for instance—start by setting shutter speed in shutter-priority mode.

3. Next, check your exposure display to confirm that the setting left to the camera's control is consistent with your photographic goals and the camera's abilities. If not, try using the exposure compensation (EC) control to bend the camera to your needs. If EC's leeway (typically 2-3 stops on either side of the camera's opinion) isn't enough to reach your target exposure in a priority mode, resort to manual mode if available.

4. As long as time's on your side, testing and bracketing are dirt cheap with digital cameras. Don't be afraid to go out on a limb—it's the best way to learn, and it's never been safer.

Guiding Constraints

As we've already seen, there are many ways to skin the exposure cat. Rather than work off a recipe, learn to make exposure trade-offs to suit your own tastes and further your own photographic goals. To get there, you'll need to 

• Know exactly what's at stake with your exposure choices.

• Experiment like crazy to develop a feel for exposure and the trade-offs involved.

Luckily, the feel will come surprisingly quickly with the instant feedback and freedom to screw up digital photography alone affords.

Narrowing the Choices

Exposure decisions can be exceedingly complex, with many variables to juggle. To navigate this jungle of seemingly endless choices, you need a path leading from the scene and equipment at hand to the photo you want to capture—the one that conveys the order and emotion unique to your vision. 

With the photographic goal firmly in mind to help you prioritize your choices, the exposure constraints below will guide your path.

To make things a bit more concrete, the discussions below are cast in terms of the camera I know best—the highly malleable 2.1 megapixel Oly C-2020Z. However, the same considerations readily transfer to any digital camera with similar controls and features, including the Oly C-30x0Z.

Monitoring Exposure

 Your camera can teach you a lot if you check its exposure display early and often. If you shoot without checking the consequences of your settings, you may not get what you bargained for. Take a moment to frame a test shot and see the camera's take on what you're about to do, particularly when pushing the exposure envelope.

In all metering modes on the C-2020Z, a half-press of the shutter release with the main (rear) LCD turned on will display the setting(s) you're controlling in green and any settings left to the camera's discretion in white. If the setting(s) you're in charge of turn red, you've brought the camera to its knees—for example, by setting a fast shutter speed in shutter priority mode in low light requiring an f/1.4 aperture when f/2 is the widest the camera can deliver. In all modes but auto, the C-2020Z goes on to underline the offending red setting(s) and beneath that shows an up or down arrow corresponding to the exposure control button you need to use to return to the camera's recommended exposure.

In aperture- and shutter-priority modes, this simple check will show you what the camera's coming up with for the setting left to its control—e.g., shutter speed in aperture-priority mode. If the camera's choice in a priority mode proves inconsistent with your goals for the shot at hand, you'll have to 

• rethink the setting you're controlling,
• override the camera with an exposure compensation (EC) adjustment, or
• take your chances with manual exposure.

In manual mode, checking the exposure display will show you whether the camera's looking or more or less light or is content with the EV corresponding to your choices.

Even program mode displays its choices with that half-press of the shutter button, but I seem to catch on faster when forced to commit to at least part of the exposure decision.

Recording Mode—Critical Pre-Exposure Digital Decisions

Before you even get to an actual exposure, you'll need to set up an appropriate recording mode for your digital camera. Involved are no less than 6 fundamental and largely independent camera settings, each representing an important decision with very real consequences. For the most part, these decisions will be unfamiliar from your film experience, but they're critical nonetheless. The table below summarizes the recording mode settings and the issues at stake.

I think of these 6 settings together as my recording mode. The only one tied directly to exposure is ISO, of course, but the others deserve mention in this context because...

Inappropriate recording mode choices can easily and irreversibly introduce noise and artifacts into your images and can severely limit your post-processing and printing options down the line.

Before taking your first serious digital photograph, it would be a good idea to work through the 6 absolutely irreversible recording mode issues—ISO, color mode, resolution, compression and color interpolation and sharpening—preferably by testing to find what meets your needs. Let the most demanding credible end-uses for your images be your guides. (Remember, digital cameras make testing easier and cheaper than ever before.) Fortunately, you can often fix white balance problems in post-processing—especially when the scene contains something that really is white.

Click at left to skip to resolving power now. To review the elements of recording mode one by one, read on.

Color Mode

This one's simple. 

Record in full color unless there's an overwhelming reason to do otherwise.

Recording exclusively in B&W (grayscale), sepia, blackboard and other special in-camera recording modes severely limits all your downstream options with those images. If you have the time, the skill and the right tools, these effects can always be achieved in post-processing, usually with much better results. That's especially true in B&W work.

If you lack the post-processing resources or otherwise prefer to record in, say, B&W or sepia mode, I strongly recommend taking and archiving a full color exposure as well. Whatever you do,

Don't forget to restore full color recording after using other modes.

Digital ISO Settings

The ISO setting on a digital camera determines its overall light sensitivity, just as the ISO rating on a film canister informs the photographer of the sensitivity of the film inside. The ability to change light sensitivity at will without physically swapping out the image receiver is one of the more important benefits of digital recording. Of course, digital and film light sensitivities arise from vastly different physical processes, but digital camera engineers work hard to align their ISO settings with established film ISO ratings to preserve the applicability of tried-and-true exposure expectations like the sunny f/16 rule. After all, forcing experienced photographers to relearn exposure on moving to the digital side is no way to sell digital cameras.

The manual for my Oly C-2020Z states,

The sensitivity [ISO] scale is based on the one used for picture film, but the numbers are for reference only.

In other words, exact equivalence with film speed (ISO) isn't guaranteed, but for any given scene at any given ISO, the C-2020Z and film will require about the same light input or EV.

How close the equivalence comes probably varies from camera to camera (mine acts more like ISO 80 when set to ISO 100, as discussed below), but the potential for discrepancy is by no means unique to the digital side. The true ISO of a given camera-film combination doesn't always match the nominal ISO of the film used. For precision work with external light meters, professionals usually find it necessary to calibrate the true ISO of each different camera-film combination they use. Digital cameras are no different.

Noise

The big difference between digital ISO and film ISO relates to image noise, which is only partly analogous to film grain. To be sure, image noise increases predictably with increasing ISO, just a film grain does, and noise can sometimes look like grain, but once it becomes visible, noise is much more an image flaw than grain. Furthermore, noise can be manipulated in ways that grain can't.

To simplify the discussion that follows, I've extended the concept of noise beyond the realm of random time-varying phenomena to include more deterministic but equally undesirable image artifacts like those due to dark current. (No one wants it in their images, but some technically-minded visitors don't consider dark current artifacts true noise. I won't get into that debate here.)

In digital photography, the noise of practical interest comes in two flavors—random and fixed-pattern—differing somewhat in cause and remedy. Both types tend to be most problematic in low-light situations.

Random Noise

Random noise (RN) arises from 

• primarily thermal fluctuations in the electronic components that handle the analog CCD output signal

• quantum (statistical) fluctuations in the numbers of photons reaching CCD sensels from the scene 

RN varies unpredictably, both in time and across the image frame. Since RN is uncorrelated, it can be reduced effectively by image averaging, a technique well known to CCD astronomers, professional and amateur alike. For every N identical exposures averaged together, the RN-related signal-to-noise ratio increases by a factor of sqrt(N).

RN is aggravated by underexposure, by high CCD temperatures and by physically small sensels. It produces a speckled pattern in affected images, often most conspicuously in the shadows, as seen in the example at right. (The best appreciate the noise, view the full size image at 400% magnification.)

To minimize RN, avoid underexposure, keep ISO at the minimum feasible setting and keep your camera cool. If you plan to shoot outside on a cold night, let the camera equilibrate with the cool air beforehand. (Cooling your camera below ambient temperature will invite condensation on the lens and elsewhere, possibly damaging the camera.) Some go so far as to avoid LCD use to keep their LCDs from heating the CCDs nearby, but I have yet to see a compelling case for this practice. If you decide to take several redundant exposures for averaging in post-processing, record them a lossless (TIFF or RAW) format and use a sturdy tripod and a remote shutter release to eliminate any possible camera shake.

Fixed-Pattern Noise

Fixed-pattern noise (FPN) varies in time but is rooted in inhomogeneities among the CCD sensels and thus exhibits a fixed pattern across the image frame.

Dark current noise (DCN), the most commonly visible form of FPN, results when stray electrons leak into sensels from the surrounding substrate in the absence of incoming light. Some sensels leak faster than others, and the longer the exposure and the hotter the CCD, the more dark current electrons a given sensel will accumulate. In low light, dark current electrons may even outnumber the photoelectrons liberated by gathered photons in the leakiest sensels. At constant temperature and exposure time, each sensel's dark current electron load varies randomly about a mean. A dark frame exposed with no light input (e.g., with a lens cap on) will show the instantaneous dark current pattern for a specific temperature and exposure time. An average of several time- and temperature-matched dark frames will approach the mean dark current pattern. Within limits, you can reduce DCN by subtracting dark currents from an image, preferably with an average of several dark frames, as CCD astronomers often do, but you will add some RN in the process. 

Fixed-Pattern Noise in a Cold 4-Second Exposure
DCN is aggravated by high CCD temperatures, by long exposures and by CCD aging. It produces an array of scattered abnormally bright ("hot" or "warm") pixels in affected images, again most conspicuously in the shadows, as seen at right.  This 4-second image of the crescent moon and Mercury (top right corner) and its matching dark frame were recorded back-to-back with the camera thermally equilibrated with the cold night air. The dark frame won't do for subtraction (try it and see) because it was recorded in a lossy (JPEG) format.
	4 sec Scene
	Dark Frame

To minimize DCN, avoid exposures over 1/2 sec and keep your camera cool as explained above. If you decide to take several redundant dark frames for averaging and subtraction in post-processing, record them in a lossless (TIFF or RAW) format on location, taking care to match the CCD temperature and exposure time of the target image. Be sure to use a sturdy tripod and a remote shutter release to eliminate any possible camera shake.

Gray Frames

Another source of FPN less important in digital photography is variability in sensitivity (the proportionality between captured photon count and output voltage) among sensels. This fixed inhomogeneity can be corrected by shooting a known uniform target several times to obtain an average gray field. Dividing the image at hand by the gray field normalizes sensitivity across the CCD.

All-Out Noise Abatement

The full court press against noise in post-processing involves the averaging of several redundant images, division of that average by a gray frame and finally, the subtraction of the average of several dark frames, all recorded in a lossless format.

ISO, Amplification, Analog-to-Digital Conversion and Noise

Digital cameras adjust their light sensitivity (ISO) in two ways:

• By varying the amplification applied to the CCD's analog output signal before analog-to-digital (A/D) conversion

• By remapping ~12 bits worth of analog CCD output onto 8 bits of digital output in the camera's A/D converter (ADC)

Hybrid approaches are probably common here, but either way, the noise inevitably present in the CCD's analog output gets amplified, electronically or mathematically, right along with the signal, and the pre-ADC amplifier adds further noise of its own. Increased sensitivity comes at the price of increased noise, period. 

For every doubling of ISO, the light input required for a proper exposure drops by a full stop (a factor of 2), while image signal-to-noise ratio (SNR) drops by a factor of 1/sqrt(2) = 71%. Conversely, the required light input increases by 2 stops and SNR doubles in going from ISO 400 to ISO 100 at constant image brightness. The table below shows the quantitative relationship between ISO and time-varying random noise at constant image brightness. 

To make matters worse, random noise and fixed-pattern dark current noise both increase with increasing ISO in CCD-based imaging devices. Long multi-second exposures tend to run afoul of dark current noise, even at minimum ISO, and higher ISO settings only compound the problem. See the dpFWIW article Low-light work with the C-20x0Z for details on long exposures with digital cameras.

Using ISO Wisely

Controlling exposure by manipulating ISO in lieu of aperture or shutter speed is generally a bad idea because higher ISO settings beget more image noise, as we've just seen. But ISO adjustments are often inescapable in stop-action work, where short exposures and low available light are a common mix. 

The increased noise isn't always subtle, but if a higher ISO makes a must-have shot possible, the noise may well be a price worth paying. Image noise tends to be more apparent in the shadows (which may be expendable) and at higher final magnifications, like the ones needed for 8x10 prints (which may not be needed). With forethought and extra exposures specifically designed for the purpose, you can significantly mitigate noise in post-processing in relatively static scenes like the night sky sans moon, but such subtraction and averaging techniques aren't applicable to the short exposures needed for stop-action work, where need to bump ISO issue most often arises.

The safest ISO policy is this: 

Bump ISO only as a last resort and never more than absolutely necessary.

If you dial in a high ISO for a specific purpose, make sure you set it back to your camera's minimum before you forget.

Note that some large-sensor cameras like the Nikon D1x and the Canon D30 and D60 deliver very acceptable noise levels at ISO 400 and beyond, but be prepared to shell out big money for that kind of low-light performance.

Beware Auto-ISO

BTW, keep an eye on ISO in low-light situations in auto-exposure mode. Some cameras like my old C-2000Z take liberties with ISO under such conditions, even though a specific ISO has been set in the menus. Since the C-2000Z otherwise respected my ISO setting, I nearly always used it in aperture- or shutter-priority modes to lock in ISO 100. Thankfully, the C-2020Z and later Oly digitals offer more predictable ISO control options alongside an auto-ISO setting.

Action photography aside, avoid auto-ISO settings like the plague.

Image Resolution

Pixel count is by far the single most important determinant of image quality in digital photographs, particularly when it comes to printing, and resolution (e.g., 1600x1200 vs. 800x600) determines pixel count. The higher the resolution, the more pixels you'll capture and the better the image you'll have. A 1600x1200 image contains 4 times the pixels of an 800x600 version and will be commensurately sharper and less "pixelly" at any given print or display size. A 1600x1200 JPEG compressed at 4:1 will contain about as many bytes as its uncompressed 800x600 counter part, but the 1600x1200's quality will almost always be conspicuously superior to the 800x600's at such mild compression levels. In a very real sense, reduced resolution is the crudest possible form of "compression", with results to match.

That makes the resolution decision pretty easy:

 Record at your camera's maximum resolution whenever feasible.

If you still need convincing, read Higher resolution or lower compression JPEGs? by dpFWIW contributor and physicist Rick Matthews. Rick's illustrations say it all. 

To avoid the hassle of downsizing images after the fact, some users record images destined for Web use or e-mails to friends at lower resolutions. But what happens when you end up wanting to print an unexpectedly good one at 8x10? Cropping, downsampling and other pixel-wasting operations are best left for post-processing with a specific end-use in mind, and many editors now offer batch processing and macros to automate such tasks. (If you're unfamiliar with these manipulations and the issues they entail, be sure to read John Houghton's excellent image sizing primer.)

Bottom line: At exposure time, grab all the pixels you can because...

Inappropriate resolution choices can severely limit your post-processing and printing options down the line.

Memory Allocation

Memory card capacity is by far the most important limiting factor here, but  recording at full resolution is your single most effective allocation of that valuable resource. Since a 4-to-1 drop in resolution (e.g., from 1600x1200 to 800x600) is usually much more detrimental than a 4:1 JPEG compression, judicious use of compression can more than offset the memory hit that comes with increased resolution—typically with far less impact on image quality and post-processing and printing options. 

Shot-to-Shot Latency

Many factors affect the time lag between one shot and the next in a digital camera. File write time, the time required to store an image on a memory card once the camera's processed it, is usually the rate-limiting step in getting to the steady green light for the next shot. Write time increases roughly linearly with pixel count and therefore with resolution. In-camera processing times increase with higher resolutions as well. Even though compression itself adds to the processing time, judicious use of compression can more than offset the time increased resolution adds to your camera's shot-to-shot latency. In theory, in-camera sharpening also adds to latency, but I haven't noticed the difference if it does.  

Image Compression and Color Interpolation

Digital image files are inherently large. The true-color (24-bit) images produced by most cameras take up 3 bytes per pixel. With EXIF header information, that comes to 5.5MB for a 1600x1200 image from a lowly 2 megapixel camera!

One way to reduce image files to more manageable sizes is to apply compression. With smaller files, you can fit a larger number of images or images of higher resolution on a single memory card. Another way is to record only raw CCD data in-camera, leaving color interpolation and expansion into RGB triples for post-processing. Cameras offering RAW recording also give you the option of performing color interpolation and white balance adjustments in post-processing on a computer with a lot more resources than the one in the camera.

Of course, compression and RAW recording aren't exposure issues per se, but like other recording mode settings, they're certainly worth addressing before embarking on serious work with a digital camera.

JPEG Compression

Nearly all digital cameras offer image recording with the lossy but highly effective JPEG (Joint Photographic Experts Group) compression scheme. The JPEG standard was explicitly designed to compress photographs and other continuous-tone images in accordance with the brain-eye system's well-known ability to detect smaller changes in intensity (luminance) than in color (chrominance). In other words, JPEG compression favors the retention of luminance over chrominance data.

JPEG does this very well, but there's a catch: As with any lossy compression scheme, the greater the compression, the smaller the final file size but the greater the data loss, and the more apparent the compression damage when the image is re-expanded. The compression damage takes many forms, all lumped under the term JPEG artifacts. 

By design, JPEG compression is poorly suited to line drawings and text. GIF and PNG are the compression schemes of choice for such images.

Choosing an appropriate JPEG compression level amounts to making the best trade-off between image quality and final image file size based on your most stringent credible end-use for the images at hand. Only you can make that call based on testing with your own equipment.

The coming JPEG2000 update of the current JPEG standard promises to do a much better job with wavelet technology, but few cameras now in use will be able to take advantage of it. 

You can learn more about JPEG compression from this excellent JPEG FAQ page. 

JPEG Artifacts and Compression Levels

JPEG artifacts are unwelcome image features arising directly from damage done to the original image data during compression and re-expansion. The smaller the final compressed file size, the greater the damage will be and the more conspicuous the JPEG artifacts will become.

John Houghton kindly supplied this heavily compressed illustration of common JPEG artifacts. Note the ripples along the margins of the red square, the blocky transitions in the blue gradient background and the ringing around the text and the red square.

Now that you've come face to face with JPEG artifacts, you're probably wondering if the "maximum resolution, compressing as needed" strategy touted above is really all that smart. Once again, I invite you to see for yourself at Higher resolution or lower compression JPEGs? by dpFWIW contributor and physicist Rick Matthews.  

Worst Case Scenarios

Very sharp, high-contrast boundaries like those in line drawings and text show obvious rippling and ringing at almost any JPEG compression level. Such boundaries represent the very worst case JPEG scenario, but luckily, they aren't all that common in photographs. (That's why they weren't a priority for JPEG designers.)

In routine photographs, JPEG artifacts are most likely to show up as subtle bandings within large low-contrast areas like the sky or as speckles or ripples along color boundaries, particularly diagonal ones. Images with rapid color changes over very short distances also suffer to some extent under JPEG compression. 

The less image detail, the higher the image quality for a given JPEG compression ratio. In highly detailed scenes with lots of color changes from pixel to pixel, the space actually saved with a given compression level will be greatly reduced. 

See For Yourself

To see the damage for yourself, copy some challenging images from your collection based on the guidelines above, compress the living daylights out of the copies, and then zoom in on the resulting images. Then back off on the compression level to see where the most noxious artifacts first appeared.

JPEG Options

Most digital cameras allow you to set the level of JPEG compression applied, and many now allow you turn it off completely. As with resolution, image quality, post-processing and printing options, memory card capacity and shot-to-shot latency all hang in the balance. Your willingness to change cards frequently enters the equation when compression is disabled, even with large cards.

The compression options in the table below are typical of higher-end cameras.  

RAW and TIFF Recording

To avoid JPEG artifacts in your captured images, you'll have to turn to an output file format using either lossless compression or no compression at all. That's where RAW and TIFF recording come in, but be prepared to pay a hefty price in 

• drastically reduced memory card capacity in terms of images per card

• drastically increased shot-to-shot latencies via extended file write times

• prolonged download times from the camera, and 

• with RAW files, extra and prolonged initial post-processing steps.

For some photographers, or for some shots, the TIFF or RAW cost/benefit ratio may make perfect sense, but the fact of the matter is that most digital photographers record JPEGs most of the time.

RAW Recording

RAW output files contain digitized but otherwise raw sensor data without color interpolation, white balance adjustment or sharpening. In many ways, they represent a "digital negative" and are sometimes referred to as such. Professionals strongly favor RAW recording because it maximizes both post-processing flexibility and final image quality for reasons explained below. Unfortunately, you won't be able to view or edit RAW images straight from the camera with any old image viewer or editor. RAW formats are still quite camera-specific. To convert RAW images to something you can actually work with, you'll have to rely on the proprietary software supplied with your camera. Worse yet, RAW conversions are often quite slow, even on fast computers. 

What are the RAW advantages? For starters, RAW recording allows you to "externalize" certain basic processing tasks better performed outside the camera. In-camera color interpolation, white balance and sharpening algorithms have to make significant compromises in order to keep shot-to-shot latencies reasonable while working within camera limitations revolving around CPU speed, available RAM and allowable firmware footprint. If you'd rather not be subject to such compromises, RAW recording may be the ticket—if your camera offers it. 

External color interpolation, white balance and sharpening algorithms running on a desktop or notebook computer can afford to be much more sophisticated (read "more CPU- and memory-intensive").  External algorithms are also free to take their own sweet time since no one's champing at the bit to squeeze off the next shot (although they may well be waiting to get on with their editing). Another RAW advantage is the ability to adjust white balance after the fact. Although RAW files are typically quite a bit larger than the least compressed JPEGs offered, they're usually significantly smaller than the corresponding TIFFs. To make their RAW files even smaller, some cameras, like the Nikon D1x, offer (presumably lossless) RAW compression.

Less widely appreciated is the greater dynamic range afforded by RAW recording. RAW sensor data is typically digitized at 10-12 bits per sensel, but JPEG and TIFF images must be requantized to 8 bits per primary color channel during color interpolation. Reduced bit-depth and associated quantization errors mean more blown-out highlights and black-hole shadows, among other artifacts. By post-processing at the higher RAW bit-depth and then saving to an 8 bit per channel file format, significant improvements in final image quality can be realized.

TIFF Recording

Most cameras offering uncompressed or losslessly compressed recording use TIFF (Tagged Image File Format) files for that purpose. TIFFs aren't subject to compression artifacts, but they're very large to gargantuan, depending on the resolution chosen. The uncompressed TIFFs output by my 2MP camera run 4-14 times larger than the corresponding JPEGs! Some cameras record compressed TIFFs, which achieve at most a 2:1 to 3:1 reduction in file size using the lossless LZW compression scheme. Unfortunately, compressed TIFFs aren't as standardized as their uncompressed counterparts and may not be recognized as TIFFs, even by software claiming to support compressed TIFFs.

Think of an uncompressed TIFF as the image the camera had right before JPEG compression would have been applied. Color interpolation from the Bayer pattern sensor data has already been performed in-camera. The image has already been requantized to 8 bits per primary color channel. Any white balance settings in effect at exposure time have already been applied. And if in-camera sharpening is enabled, the image will have been sharpened as well. 

JPEG, TIFF or RAW?

The best way to determine the recording format best for you is to test against your own most stringent credible end-uses. If you can live with JPEG recording—and most users can—your digital photography will be greatly simplified on many fronts. Someday cameras may offer losslessly compressed recording in PNG (Portable Network Graphics) format. Until then, TIFF and RAW remain the formats of choice for applications requiring lossless recording.

Dialing In JPEG Compression Levels

Don't assume that the image quality gained with less compressed recording automatically justifies the potentially huge memory hit involved on a routine basis. Many C-20x0Z users, myself included, find the moderately compressed 1600x1200 HQ recording mode perfectly acceptable for all but the most demanding applications—even for 8x10 prints.

As dangerous as that strategy may be, I generally reserve my least compressed (SHQ) JPEG option for selected shots. Uncompressed (SHQ TIFF) recording has its uses—for example, in dark field subtraction work—but for me, TIFFs take up way too much memory card space for routine work.  A professional working in a studio with a stack of large memory cards and the means to unload them quickly into a nearby computer might feel differently, of course, but that's not my MO.

In-Camera Image Sharpening

Before leaving the realm of irreversible pre-exposure decisions, be aware that many experienced digital photographers strongly recommend against in-camera image sharpening—at least for shots you'll likely be post-processing. Everyone agrees that sharpening is best saved for the very last step in any post-processing sequence, so why apply it up front in-camera?

That said, the sharpening algorithm in my C-2020Z hasn't been an obvious debit with regard to either post-processing or image quality in most instances. I nevertheless disable in-camera sharpening on a routine basis because I don't have a good track record when it comes to picking important shots ahead of time. If disabling sharpening has shortened my shot-to-shot latency, I haven't noticed it.

On the C-2020Z, you disable in-camera sharpening by selecting the "soft" setting in the sharpening menu. The "normal" setting turns sharpening on. The C-3030Z added a "hard" option with extra sharpening. The terminology is unfortunate, but at least the control is there.

You'll find more on sharpening in post-processing elsewhere on this site.

White Balance

Even the most common "white" light sources—the sun, indoor lighting and flash—differ substantially in color composition. These differences are well demonstrated in this Light and Color tutorial. Here at the Earth's surface, for instance, the quality of sunlight varies tremendously with the weather and the time of day: Outdoor ambient light is bluer under overcast skies than it is on clear, sunny days and much redder in the early morning and late afternoon than around noon. Artificial light is even more variable. Incandescent light is quite red, while flash is rather blue. Fluorescent lights generally tend toward the green but vary significantly with the type of lamp — daylight, cool white, neutral white, etc. In fact, the fluorescent color casts are problematic enough to warrant several fluorescent WB presets in my C-5050Z, as shown in the table below.

When viewing a scene directly with our own eyes, we tend consider any light with roughly equal primary color intensities as "white" or at least neutral, and we unconsciously correct for subtle biases in the balance of primaries based on expectations accumulated over a lifetime of visual experience. But cameras aren't that smart. The light coming off a scene inevitably carries the color bias of the source. You'll probably be unaware of it at the time, but without help, any camera will faithfully record that bias, and welcome or not, you'll be seeing it in your photographs. (Why the brain-eye readily applies all manner of corrections to our visions of live scenes but not to photographs of them, no one knows, but that's the harsh reality photographers face.) 

Since any source-related color bias will be most conspicuous in objects that should have been white by human standards, photographers have come to think of this issue as one of white balance, but the real issue is not so much a matter of whiteness as of neutrality.

When I took the photo at right under incandescent light, I was completely oblivious to the now obvious reddish cast of my source. Had I taken it using my camera's tungsten white balance setting, the cast would have been largely if not completely eliminated.

White Balance Controls

Film photographers intent on WB control typically have to contend with time-consuming and potentially costly film and/or filter changes, and their options and wiggle room tend to be rather limited. Luckily, digital cameras have largely done away with all that. With digital white balance (WB) controls, compensating for unwanted source-related color biases at recording time can be as quick and easy as navigating a menu. In fact, fingertip WB control is one the truly great conveniences in digital photography. In-camera WB adjustments are made by the firmware when the raw CCD data undergoes color interpolation. If you use RAW recording, you'll have to do your own white balancing in post-processing; otherwise, the camera will do it for you, with or without your help.

Gee, It Looked White to Me

The highly effective auto-WB feature built into your own brain-eye system makes it difficult to appreciate the camera's WB struggle. When you behold clean snow or any other surface that your brain expects to be pure white, white is what you see, at least at first glance, regardless of the light source.

But cameras record the light they see without making judgements about how things should look. Without some intervention, snow in late afternoon sun photographs with a distinct reddish cast. At times, that may be just the look you're after, but more often than not, pure white is the goal for snow because that's what anyone standing there would have seen. And that's where you and the camera's WB settings come in.

Note: Exposing snow and other bright surfaces correctly is an entirely different matter, as discussed below.

WB Settings

The automatic TTL WB systems found in most current digital cameras work amazingly well — well enough in fact to be the default setting for most users — but they're not infallible. Fortunately, many higher-end cameras also offer manual WB settings allowing you to inform the camera about the dominant illumination at hand in tricky situations likely to trip up auto-WB. 

One such situation arises in external flash (EF) work with the internal flash turned off (EF-IF for short). Disabling the internal flash defeats auto WB in low-light conditions, at least with my C-2020Z. Manual WB (sunny or overcast) is the only sure way to keep EF-IF shots from coming out too blue.

Note: If you know of any other conditions likely to make auto-WB fail, please drop me an e-line at dpFWIW@cliffshade.com.

WB implementations rapidly have become more and more sophisticated. My 1999-vintage C-2020Z, a star in its time, offered fully automatic TTL WB plus 4 manual WB pre-sets covering the most commonly encountered lighting situations. By late-2002, my C-5050Z had auto WB plus 9 presets plus 4 storable custom WB settings and a manual WB mode, as shown in the table below; even its factory presets are adjustable.

* Table Note: The C-5050Z has no generic fluroescent setting. It's much better than the C-2020Z at correcting fluorescent color casts, but it forces you to learn much more than you ever wanted to know about fluorescent lamps. 

Note that the named pre-sets all refer to the source of illumination in the scene, with no mention of the scene elements reflecting the light to the camera. That's your cue to select WB pre-sets based on the light source, not on the subject matter. (Don't get locked into the pre-set names, however — the tungsten preset might be profitably applied to any overly red light source.) Now think of the expense and carrying capacity that would tied up in a collection of optical filters capable of neutralizing all the sources listed in the table above.

Manual WB — Show Me Something Neutral, And More

Newer digital cameras like my Oly C-5050Z offer highly accurate "one-touch" WB-by-example features that tap your brain power for the white or neutral recognition piece of the WB equation. A white or gray card carried in your camera bag provides the neutral standard under this "show me something neutral" WB scheme. Since "white" papers actually vary substantially in color, photographic gray cards are safer for this purpose because they're guaranteed to be neutral. 

Manual WB has other valuable uses. You can warm a shot by showing your manual WB something blue, or cool the scene by showing it something red beforehand. Many digital infrared (IR) enthusiasts use manual WB to manage the sometimes garish false color schemes their cameras would otherwise assign to IR images captured under color recording. (Remember, the concepts of "color" and "neutral" are completely undefined outside the visible spectrum.)

WB Fallbacks

Digital cameras also provide some valuable fallbacks on the WB front:

• You can always check your results on the LCD at the scene. You may not pick up subtle WB problems, but it's better than nothing.

• When in doubt, you can bracket for WB, time and memory permitting.

• You can usually fix WB problems in post-processing—especially when the scene contains something that really is white. Many photo editors offer very simple "show me something white" correction schemes.

Getting WB right has never been easier.

Sunlight Variations

Most of the variations observed in the color of sunlight at the Earth's surface stem from spectral variations in atmospheric scattering. In clear air, short (UV-A and blue) wavelengths  suffer up to 16 times more scattering than longer (red and near IR) wavelengths because scattering efficiency by air molecules (N₂, O₂, CO₂, etc.) varies inversely with the fourth power of wavelength.

Direct sunlight here on Earth is yellower (redder and greener) than the light leaving the sun because the atmosphere scatters a good bit of the blue away before the light ever reaches you and your subject. (Yes, some of the scattered blue will eventually scatter back to the surface, but there's still a net loss of blue.)

Overcast light is bluer than direct sunlight because visible wavelengths are all scattered equally well by particles the size of condensed water droplets in clouds—hence clouds in shades of gray at midday. Scattered blue light falling onto the cloud tops from above gets tossed back in the mix, now with an even chance of getting to the ground. There's still a net loss of blue, but under overcast skies, it's smaller relative to losses at other wavelengths.

Outdoors, shaded areas are illuminated predominantly by skylight, which is quite blue on sunny days and closer to neutral on overcast days. Shaded spots under leafy trees can also pick up extra green light reflected from or transmitted through overhead leaves.

In the morning and evening, sunlight takes a much longer path through the atmosphere, losing even more blue and some green to scattering along the way. That's why early and late day sun is redder, and why sunrises and sunsets feature the warm longwave (red, orange and yellow) colors we so admire.

Near IR wavelengths suffer very little scattering, even under hazy conditions—hence the incomparable clarity of IR photos. Aerial surveillance photos are commonly made with near IR pass filters to take advantage of this fact.

Testing Recording Mode Choices

Only you can decide how best to play the JPEG artifact vs. file size trade-off and the in-camera color interpolation and sharpening games. The proper balances depend critically on your equipment and quality needs. Fortunately, digital cameras make the required testing a snap.

A simple but useful recording mode test requires 2 sample scenes—one in bright sunlight, preferably with large areas of low contrast, and the other in open shade with richly detailed shadows. Shooting from a tripod using a remote if available, capture each scene with identical exposure settings across the recording modes to be compared. Print the images at constant final subject size to see what's acceptable to you. Pay close attention to

• Overall sharpness and lack of pixellation at realistic print viewing distances.
• Areas where colors merge—JPEG artifacts will be most apparent here.

Let the most demanding credible end-uses for your images be your guides.

Resolving Power

Determining and describing the optical quality of a camera lens turns out to be a very complex and highly technical enterprise involving issues of distortion, aberration (primarily chromatic) and resolving power—the ability to distinguish small image features (like line pairs) at small separations. Resolving power depends primarily on 

• the maximum sharpness or resolution of the lens (here measured in distinguishable line-pairs per mm), and 

• the image contrast it delivers at scales at and above maximum resolution. 

To complicate matters further, these 2 attributes are largely uncoupled—for instance, high marks on sharpness don't guarantee stellar contrast performance. A top-notch lens needs to perform well on both fronts, but that'll cost you.

Technical note:  The term "resolution" means different things when applied to lenses and CCDs. CCD resolution refers to the sensel count (say, 1600x1200) in the chip's active image-forming area. For optimum camera performance, lens and CCD resolutions must be carefully matched.

The issue of resolving power in digital cameras is further complicated by the discrete nature of CCD sensels, but resolving power remains a useful concept when applied to specific lens-CCD combinations. Until digital cameras sporting interchangeable lenses become commonplace, resolving power will remain a property of the digital camera as a whole.

Alas, the topic of lens-CCD quality is so complex that widely accepted, easy-to-interpret measures are unlikely to become readily available to prospective digital camera buyers, at least in the foreseeable future. Some of the major digital camera review sites have taken to posting test images based on a bewildering assortment of standard test patterns, but even the experts seem to have a hard time agreeing on how these should be interpreted. Some practical gauge of combined camera-lens resolution is sorely needed.

This morass tremendously complicates the process of selecting a digital camera for serious photography, but once you've cast your lot, you can rely on some fairly simple lens-related quality principles to guide your technique. 

Resolving Power Basics

Aperture is the key to resolving power management. Large apertures allow unwanted prism effects to creep in from the edges of the lens,  while small openings promote diffraction blurring at the iris. The trick is to find the aperture corresponding to the resolving power sweet spot for your lens and use it whenever you can.

Prism effects

It's an inescapable fact of lens design and manufacture: All lenses perform better at less than wide-open aperture because optical quality  inevitably falls off toward lens edges. Chromatic aberration (the focusing of different wavelengths coming from the same subject point onto different points on the image receiver) is one of the most commonly encountered "prism" or "edge" or "off-axis" lens effects, especially in less expensive lenses.

Diffraction blurring 

Diffraction blurring occurs when incoming light diffracts (bends) around the edge of the iris instead of passing cleanly through. For a lens of focal length L, the higher the f-number N, the smaller the physical aperture (L/N) becomes. The closer the physical aperture gets to the wavelengths of visible light (400-700 nm), the greater the bending at the iris and the greater the blurring at the image plane will be. Thankfully, diffraction blurring isn't as damaging to image quality as blurring due to poor focusing, but it's clearly noticeable in most digital cameras at f/8 and smaller apertures. In fact, few digital cameras offer apertures smaller than f/8 for that very reason. 

At any given f-number, digital cameras are more prone to diffraction blurring than 35 mm cameras. The reason boils down to image receiver size. The 8-10 mm diagonals typical of consumer-grade digital camera CCDs are small compared to the 43 mm diagonal of the 35 mm film frame, and small digital diagonals require lenses with very short focal lengths — typically in the 5-20 mm range. At a wide-angle zoom setting of 5 mm, a digital f/8 exposure calls for a tiny physical aperture of 5/8 = 0.625 mm. The same shot with a 35 mm camera would involve a physical aperture of around 24/8 = 3 mm. 

That's why current consumer-grade digital camera lenses are said to be diffraction-limited. Expect this to change only when physically large CCDs approaching the size of the 35 mm film frame become affordable, but don't hold your breath.

Resolving Power Sweet Spot

OK then, how does resolving power bear on exposure strategy?

Practically speaking, resolving power peaks at the aperture(s) striking the best balance between diffraction blurring and off-axis effects. Like many a 35 mm camera lens, the f/2.0 lens on my 2.1MP C-20x0Z reaches maximum resolving power at ~2 full stops down from wide open — i.e., around f/4 at wide angle and f/5.6 at full zoom. For the slightly faster f/1.8 lens on my 5.2MP C-5050Z, informal testing suggests a resolving power sweet spot around f/2.8 to f/4.

Informal Resolving Power Test For the 5.2MP C-5050Z
f/1.8 @ 1/1000 sec	f/2.0 @ 1/1000 sec
f/2.8 @ 1/800 sec	f/4.0 @ 1/400
f/5.6 @ 1/200 sec	f/8.0 @ 1/100 sec

Table notes: These handheld 388x238 crops came from otherwise unedited 2560x1920 SHQ JPEG images taken in landscape program mode at full wide-angle zoom (FL = 7.1 mm) with in-camera sharpening disabled at -5. Differences in detail capture are best appreciated when viewed at 200-400%. Thanks to Tom Lanckamp for the idea of using chain link fence as a resolving power test target. Next time, I'll use a tripod and remote control to eliminate any blurring due to camera shake.

Even at relatively wide f/2.8 to f/4 sweet-spot apertures, I usually end up with DOF to burn — near that of a 35 mm camera at f/22! Better yet, the attendant fast shutter speeds are good insurance against camera shake, which remains my No. 2 photographic nemesis — right after my woeful lack of creativity. Since diffraction blurring gets particularly nasty at f/8 and smaller apertures at current CCD sizes, I try not to go there now, but I sure wish I'd known that when I first went digital.

Optimum Aperture for Resolving Power

Chances are, the aperture sweet spot for your lens is also ~2 full stops down from wide open. For the sharpest images your camera can muster, target that aperture when you can and work to avoid f/8 and smaller apertures. If bright ambient light keeps you away from your aperture sweet spot, consider mounting a neutral density (ND) filter or a polarizer.

Not surprisingly, Program mode usually pursues a very similar strategy in both the C-2020Z and the C-5050Z, but both cameras seems to settle on f/2.8 more often than f/4. (Hmmm, maybe they know something I don't.)

Yes, Digital Lens Quality Matters

Don't fall into the trap of thinking that you can scrimp on a digital camera with a cheap lens just because the CCD's pixels are a lot coarser than the silver halide grains in film. Many digital photographers have come to realize that image quality can easily become lens-limited, starting somewhere between 1.5 and 2.1 megapixels (MP). Lens quality clear counts at 3.3MP and above.

The designers behind the emerging crop of 2MP cameras with very long, very sharp electronically stabilized zoom lenses like the new Oly C-2100UZ and the Sony Cybershot DSC-F505 may well be zeroing in on a sweet spot where field flexibility, image sharpness, low-light performance and file size all come together quite nicely. These new-generation 2MP cameras promise to give 3MP offerings with lesser lenses a run for their money.

Depth of Field (DOF)

Technically, focus is perfect only in a single plane, which presumably coincides with some part of the subject. In practice, however, focus appears acceptably sharp to the brain-eye system for some distance in front of and behind the plane of true focus. This range of acceptable focus is called depth of field (DOF). Relative to the camera, the nearest plane of acceptable focus is called the near limit of DOF and the farthest plane, the far limit. These boundaries are seldom equidistant from the plane of true focus. In hyperfocal technique, for instance, the near limit may be at 3 feet, the plane of true focus at 6 feet, and the far limit at infinity.

It's important to distinguish DOF from the locations of its near and far limits. DOF is the distance between the near and far limits. An important rule of thumb pertaining to non-close-up work states that if the subject occupies a constant portion of the frame, DOF will also remain constant, even though the near and far limits shift substantially as one goes from a close-in wide-angle shot to a telephoto shot from afar.

DOF is further complicated by the fact that DOF and its near and far limits all vary somewhat independently with aperture, magnification, focal length and, of course, with the definition of "acceptable focus". The last is usually specified as the diameter of the largest acceptable circle of confusion, which can be thought of as the image of an imperfectly focused point image projected onto the true focal plane. 

Working with DOF

Achieving a DOF encompassing just the elements that need to be in focus is always an important goal in photography, digital and film alike. To match what the eye sees, landscapes generally require a very large DOF including both near and distant scene elements. In this arena, digital cameras deliver DOF film photographers can only dream of. In close-ups, even digital cameras struggle to provide the DOF to cover a single flower, let alone the background, but film cameras struggle much more. In portraiture, however, the worm turns. DOF is often purposely reduced to help separate in-focus subjects from their less important blurred backgrounds. For film photographers, limiting DOF is easy, but digital photographers find it a struggle. These consistent differences in film vs. digital DOFs flow directly from differences in the sizes of the image receivers typically involved:  Most digital sensors are much smaller than 35 mm, medium format and large format film frames.

The Road Ahead

The remainder of this lengthy and rather complicated section will cover

• Determinants of DOF— including aperture, magnification, focal length, a wrap-up summary and several handy rules of thumb

• Limiting DOF—both at the scene and in post-processing, with a glance at other means of subject-background separation

• Magnification Rules—wherein the single most important factor determining DOF gets special treatment

• Relating Digital to 35 mm Camera DOF—leveraging your film experience

• Hyperfocal Technique—when pretty much everything needs to be in focus

We'll tackle these topics one by one in the order listed.

Determinants of DOF

For close-up work (when camera-subject distance is small relative to the hyperfocal distance), DOF varies directly with aperture f-number and even more strongly and inversely with magnification, which depends in part on focal length. With more distant subjects, focal length comes directly into play as well.

Focal length plays a critical role in DOF considerations, so we'd better get our signals straight. Often, we'll need the actual focal length (f) of the lens or zoom setting at hand. For digital cameras, actual focal lengths typically fall in the 5-20 mm range. When taking advantage of relationships long ago worked out for 35 mm cameras, however, we'll sometimes need to express focal length as the equivalent focal length (f₃₅, EFL) of a 35 mm camera. Because the 35 mm frame is 2-5 times larger than most digital camera image receivers (CCDs, CMOS sensors, etc.), the values of f and f₃₅ usually differ substantially for a given camera. For the 3x zoom lens in my Oly C-5050Z, f (actual focal length) = 7.1 - 21.3 mm, while f₃₅ = EFL = 35 - 105 mm. Using the wrong version in a formula involving focal length is likely to generate a large error. 

Aperture

At any camera-subject distance, the narrower the aperture (the larger the f-number), the greater the DOF. For close-ups, DOF doubles for every 2 stops of decreased apertu