You’ve photographed everything you can get in your viewfinder. You’ve experiments a little with effects. Then you try to photograph friends in a dark room, only to find they are either over or under exposed. You’ve tried shots at night, only to find out you have a black or muddy image. You’ve tried boosting these images in brightness and contrast on your computer only to find you get a lighter grey or muddy image. What can you do in these low light situations? Welcome to the never-ending problem of getting low-light images! A problem that plagued film photographers for decades (and still does), and now arises in the digital photography field as well.
In this article, we’ll show you a few of the things you can do to get better pictures in low-light situations with your digital cameras. Not all cameras will be able to apply all these techniques, but most will. Experiment, and you’ll find a whole new area of photography to play with!
Cameras need light to fall on either photographic emulsion or CCD pixels. Without the light, the image can’t be formed. If there’s not enough light on an emulsion, the grain can’t react chemically to the incoming photons and don’t change their state. With CCDs there’s not enough current triggered on the CCD pixels to register a light change from black. In either case, we need to find ways to either make the film or CCD more sensitive, or boost the amount of light being reflected from the subject into the camera.
The latter is the easiest to take care of by using a flash. Flashes provide brief intensive illumination that allows the CCD pixels to see the image for a short amount of time in almost normal light, producing a normally exposed image. Flash is often the best way to handle low-light situations, but the amount of light needed and the control of that light has to be carefully controlled.
The alternative to flash is to change the sensitivity of the CCD. With film photography, we move to a more sensitive emulsion, reacting more quickly to small increments of light. The relative sensitivity to light is measured by the ISO rating (it used to be called an ASA rating) of the emulsion with higher numbers more sensitive. So, a film with an ISO of 400 will be more sensitive to light than a film with an ISO rating of 200 (approximately twice as sensitive, in fact). We can’t really adjust the sensitivity of our CCDs to light, though, without replacing the chips themselves. Many digital cameras allow you to alter the ISO rating of the chip using controls on the panel, but you’re not really doing anything more than telling the camera’s computer to adjust the finished image either up or down in exposure. So, ISO rating changes on most digital camera are a computational exercise for the camera, although some digital cameras do adjust the reading sensitivity of the CCD by altering current thresholds. The latter occurs only in the more expensive cameras, though.
For those who work with low-light photography a lot, there are digital cameras using CCDs that are more sensitive to light than the usual CCDs in our digital cameras. These CCDs react to smaller amounts of light than usual and allow low-light situations to be photographed easily. However, these same CCDs tend to overexpose in normal or bright light situations. Switchable CCD panels in high-end digital cameras are likely to appear, but will be very expensive.
Other than changing the CCD, you can use software to compensate for low light levels. All photo manipulation software such as Adobe PhotoShop includes Brightness and Contrast adjustments. Using this adjustment, you can lighten and improve the contrast on an underexposed image. However, brightness and contrast adjustments will not bring out more detail than the CCD managed to capture, because a dark blob is still a slightly lighter blob even when brightness is boosted. In order to bring out detail in an underexposed, filters must be employed. These image enhancement filters look at neighboring pixels and try to determine what each pixel’s brightness should be based on differences between pixels, not the actual value of each pixel. This technique is widely used in satellite photography, as well as low-light video cameras. Image enhancement can bring some pictures to life, but there are still limits to what most filters can do.
ISO (which actually is the International Organization for Standardization and not the more commonly -- and wrongly -- stated International Standards Organization) ratings indicate the “speed” or sensitivity of a film emulsion to light. The scale established is roughly linear, so that an ISO 100 film is twice as fast (twice as sensitive to light) as an ISO 50 film, and an ISO 800 film is eight times more sensitive than an ISO 100 film. There is some failure in this linear relationship at higher ISO numbers, but the approximation is useful.
As ISO ratings increase, the amount of light needed from a subject decreases, so an ISO 200 film will need, in theory, only half the light of an ISO 100 film. This isn’t true, in fact, but there is a full stop difference between the two. Some of the faster films, such as those rated at ISO 1600, can take properly exposed photographs by candlelight (and are often used for photographing stars).
ISO equivalents for a CCD in a digital camera are an approximation by the manufacturer of the CCD’s relative sensitivity to light compared to film. Most digital cameras have an effective ISO rating of 100, roughly requiring the same amount of light as an ISO 100 film to achieve identical exposures. Digital cameras usually allow ISO ratings to be changed over a small range (such as ISO 50 to ISO 400), although the CCD doesn’t change at all. The camera’s sensitivity to the CCD current and the computational adjustment of the image are adjusted to look like a change in ISO rating for the CCD. Although you can use the ISO adjustment, you’ll find that as you increase the ISO rating of your camera the amount of noise on your pictures increases, too.
A few cameras, such as Kodak’s DCS620 and Nikon’s D1 allow much wider ISO adjustments through manipulation of the CCD currents themselves. By using several sophisticated techniques borrowed from satellite and astronomical CCD processing, ISO sensitivity can be adjusted enough to allow low-light photography from these digital cameras without flash.
Noise and its effects
Noise is a general term applied to any corruption of the image or defects of the light-gathering process performed by a CCD. There are several types of noise we have to worry about with CCD-based cameras, but the two main types are inherent noise and induced noise.
Inherent noise is a direct effect of the CCD pixel’s sensitivity and variations of light gathering power. Some pixels will be defective, some hypersensitive, and some very insensitive to light. Most are average, but a few deviations are found in every CCD, regardless of price, size, and manufacturer. The noise induced from these pixels can lead to whiter or darker spots on your image: the light spots are where hypersensitivity occurs, and the darker spots where pixels are insensitive or dead.
Induced noise is a byproduct of the electronic processing that is performed in your camera when it reads the CCD. As the amount of current drops in the CCD, induced by light falling on each pixel, the amount of noise increases. Think of it this way: when you are carrying on a conversation with a slowly running stream next to you, you can easily hear the other person. When the stream becomes a waterfall, you have difficulty hearing the other end of the conversation from the background. That’s what happens when you adjust the ISO sensitivity of your camera. As mentioned elsewhere, increasing the ISO setting means you are sampling less light from each pixel and treating it as though it was more light than it really is. The threshold between background noise and incoming signal is closer than normal, so the amount of noise increases.
Induced noise manifests itself in several ways in a finished picture, but the most obvious is an overall decrease in contrast, many flecks of pixels that don’t match the image, and occasional white or grey spots. These are produced because the electronic circuitry is having trouble telling whether the pixel’s output is real, or background effect, and the image gets muddied because of this. As you boost the ISO rating, the effect becomes much more pronounced.
How bad is the effect? If you only have a small ISO range to adjust (most cameras adjust from 100 to 200, for example) the effect is minimal but noticeable under enlargement. If you can adjust the ISO setting over a wider range, as some high-end cameras allow, the noise becomes more pronounced as the ISO rating reaches 400 or higher. By the time you push to ISO 800 or 1600, the images can be very difficult to see clearly without digital signal processing.
Unfortunately, if you have to push the ISO rating on your camera in order to get the picture, there’s not much you can do in-camera to minimize the noise. The less contrast and more grey tones in the subject you are shooting, the more problems you will have. Contrasty, multicolored images come through pushed ISO settings a little better because the electronics can differentiate the image from background. To really reduce noise, you need to use external image processing software such as PhotoShop (for mild cases) to advanced digital signal processing packages like MaxIM (for advanced problems).
Techniques to minimize low-light problems
How can you take low-light photographs that don’t look like chocolate pudding? In marginal conditions, where a CCD can still produce a decent photo if everything is just right, there are a few things you can do.
First, don’t push the ISO rating more than you have to. As the ISO rating goes up, noise increases dramatically. If you have to go for a longer exposure, use a tripod but keep the ISO setting down.
Second, if you can adjust the aperture on your camera, open up as wide as possible. The object, of course, is to get as much light into the camera as possible, and opening the lens wide lets you do this. You may have to switch to manual on some lenses to accomplish this. The faster the lens (the lower the f-number) the more light you’ll get. Check out any sports event: the professionals are using incredibly fast (and expensive) lenses with f-stops starting around f1.4. Compared to an average lens with an f-stop at f3 or f4, there’s several stops of light difference. Instead of shooting with ISO 1600 film, the fast lens user can shoot with ISO 200 instead and minimize the noise.
Motion can kill your low-light shot. Motion exists both from your subject and from the camera itself. You can minimize one source of noise easily: brace the camera well. Even the slightest camera vibration at pushed ISO ratings will cause blur, and increase the noise directly. A tripod for long exposures is standard practice, but even for shorter exposures in low-light situations, a tripod can save a shot. It’s possible to shoot fairly short duration shots (approximately 1/30 sec) and still save the image in low light conditions when you use a tripod. For anything longer, you really should use one anyway. If your subject is in motion, try to time the shot so they are at their slowest, or stopped. This isn’t always possible, but can be managed in most cases. If the subject really can’t stop, try making the photograph creative instead, using blurred motion or delayed shutter effects to give the illusion of motion in the finished image.
Most digital cameras have a white balance adjustment. Most users have no idea what it is or does, or what white balance is. Simply put, white balance is the way a camera compensates for different color and lighting conditions. We see colors correctly under fluorescent and incandescent light, as well as under direct sublight. That’s because our brain performs a correction to let us see “red” even when the real color under poor light conditions is more bluish or purplish. White balance is the way a camera corrects for this effect, using algorithms.
Digital cameras almost always use a best guess at white balance by reading the entire image at the moment you take the shot, and balance the colors and light for the overall image. If your image is dominated by one color, such as blue, the overall cast of the finished image will be wrong because the camera adjusts to average incorrectly. This also happens with low light conditions where the colors are read incorrectly by the CCD pixels. How dramatic is the difference in light intensity? Quite a bit. On an average, sunny day, a white card gives off a color temperature of about 6000K (Kelvin). Under fluorescent lights, the temperature is as low as 4000K, while incandescent lights drop to 2000K. That means that there’s a considerable variation in the way a white card looks to the camera under all three conditions, and the average reading will be wrong for all but the sunny image.
Some sophisticated digital cameras allow you to set a white balance reading by taking an image of a white card, and then apply that setting against all other images taken in those conditions. Most cameras don’t have this ability, and just guess. Those that do have a white balance adjustment let you adjust the balance across a reasonable range, much the same way you can adjust the brightness and contrast of a television picture. If you do have a white balance adjustment, experiment with it prior to taking valuable low-light pictures. Shoot an image in low light using a number of different white balance settings, and then view them on your computer. Don’t trust the small LCD screen on the camera to adjust white balance properly, because it is not portraying the colors correctly, either.
What can you do if the image was taken with the wrong white balance? Actually, it’s fairly easy to correct in software like PhotoShop. You can use several of the sliders for brightness, contrast, as well as the colors to adjust the image. You won’t get quite as good an image as you would have if you had the camera set right in the first place, but it can come very close.
What is CCD Blooming and Flat-fielding?
Lowlight photography with digital cameras inevitably involves the blooming effect of the CCD’s pixels. To understand blooming, you need to know a little more about the CCDs used in your digital camera.
CCD’s use hundreds of thousands or millions of pixels, individual light-sensitive elements that read the amount of light falling on them and generate a current proportional to the light intensity. Individual pixels are not all the same in that each pixel’s response to the same amount of light varies a little, often with differences up to 5% or more. Across an entire CCD this variation can be up to 10% (and with low-end digital cameras, even more). This means that an exposure that is right for one pixel or group of pixels may be slightly over or under exposed at another group. Even more problems can arise because of distortion effects in the lens of the camera, as well as the coverglass that protects the CCD chip in some cameras. The non-uniformity of pixel sensitivity doesn’t matter much for snapshot photographs, but for higher resolution cameras and those used for professional purposes, this variation has to be accounted for. The term “flat-fielding” refers to the correction applied by the computer to account for this variation in sensitivity.
Pixels that receive too much light or are far too sensitive to small amounts of light register at the maximum current, show a bright white spot on the CCD image, and are called “hot” pixels. Conversely, pixels that do not react properly to incoming light and show black spots are “cold” pixels. Some pixels are totally insensitive to light or react improperly, and are “dead” or “blemished” pixels. Even the most carefully manufactured CCDs in high-end cameras have their share of dead and blemished pixels, most of which are compensated for in the camera’s electronic image processing software, or simply ignored altogether.
Blooming is an effect whereby an over-exposed (hot) pixel leaks some current into neighboring pixels, causing them to become hotter as well. Instead of a single pixel or small group of pixels showing a white image, a larger group around the hot pixels shows white or shades of white as well. This can affect the overall exposure of areas on the CCD and result in images that are incorrect or lead the camera’s computer to underexpose areas of the image. The only way to handle blooming is to ensure the camera does not overexposure the image and lead to blooming areas.
The obvious way to handle low light situations is to use flash to help illuminate the subject. Most digital cameras have a small flash unit built in, and the majority have a connector for external electronic flash units as well. Most flash systems are computerized to the extent that you simply need to turn the flash unit on and let the computer decide how much flash to apply. Sensors mounted either on the camera body, inside the camera, or on the external flash can sense the amount of light to generate and the best shutter speed to use. Although built-in flash units are adequate for many low-light situations, keep in mind that these units do not generate a lot of light and can typically illuminate only relatively close subjects. Trying to light a subject fifty feet away with the built-in flash is just not possible!
In order to use flash effectively, consider the photo you would take without flash. The exposure would be long because of the low light, often longer than a second and usually not shorter than 1/15 second. At these lengths of time, blur is inevitable and often drastic underexposure is still possible. Flash units supply a short-duration (between 1/500 and 1/2000sec) burst to light the subject during the time the photo is taken, essentially freezing the subject. You can’t usually see the effect of the flash on the subject and it takes very quick electronics to control the amount of light reflected back from the subject to control the exposure and avoid over- or under-exposure. If you are trying to light a group, make sure the flash covers the entire width or height of the area; the addition of a diffuser or fresnel lens can help broaden the flash’s beam at the cost of lower output.
For better control of flash pictures, consider using synchronized flash. Some digital cameras allow you to control the exposure time used when a flash is employed. Most cameras have synch speeds between 1/60 and 1/250sec, defined as the minimum amount of time the shutter or CCD is active in order to receive the light from the flash properly. If you go faster than the camera’s synch speed (which is often easy to do with manual exposure settings), you may not get the effect of the flash on the subject in time.
Many higher-end digital cameras allow you to not only select standard features like red-eye reduction (a pre-flash is generated to reduce the subject’s iris size), but also select special effects such as rear-curtain synchronization. With rear-curtain synchronization, the flash goes off at the end of the exposure, providing for a blurred motion effect with the subject frozen at the end of the motion trail. This is very effective for long exposures of several seconds, and gives the impression of motion frozen in time.
Image enhancement filters
Some photographic software has image enhancement filters built in but for real image enhancement functions you will probably need to purchase or download special software. A search of the Web for “image enhancement filters” will produce tens of thousands of hits, most specific to a technique or application (such as satellite images or astrophotography).
Most image enhancement filters work by analysing an image overall, then breaking it down into smaller chunks, studying neighboring pixels and looking for differences. Histograms and complex algorithms are used to determine what each pixel’s relative brightness and color should be, often resulting in a hidden image emerging.
Image enhancement filters are often a hit-or-miss process, especially when there is a low pixel count as with many digital cameras will less than a million pixels. Even there, though, image enhancement filters do allow some detail to be brought out in low-light pictures, and averaging and interpolation of pixels can actually result in sharper and brighter final images with more pixels than started with.