Balancing ISO and digital noise for sharper low-light photos
Life is full of compromises. One example in photography involves shutter speed and aperture: To achieve the right exposure, slower shutter speeds must be offset by smaller apertures. A similar balance exists between ISO and digital noise.
Making sense of ISO
ISO is a measure of your camera’s sensitivity to light. Aside from some extremely simple point-and-shoot models (like the iPhone), most cameras include an adjustable setting for ISO. Your camera may have an ISO button on its body, or the setting may be hidden away in a menu on the LCD display.
A low ISO number (ISO generally starts at 100) indicates that the camera’s sensor is relatively insensitive to light. Every doubling of the ISO (such as from 100 to 200, or from 200 to 400), doubles the camera’s light sensitivity. It’s comparable to a “one-stop” change in exposure. Suppose you were going to take a photo in dim light, and the camera was set to 1/30 second and f3.5. You might not be able to open the aperture any wider—f3.5 may be as far as it will go—and 1/30 second is too slow to take a sharp photo at that aperture setting. But if you double the ISO from 100 to 200, you can take the same photo at 1/60 second; and at ISO 400, you can take the same shot at 1/125 second.
Best of all, this technique works in most of your camera’s exposure modes, including Program; so you don’t have to work with your camera set to manual to take advantage of ISO. If you have the camera set to Program and you get a slow-shutter-speed warning—generally a little shaky hand in the display—you can manually crank up the ISO to take a sharper picture. Most cameras don’t let you change any settings in full Auto mode, however, which is a great reason to shun Auto and use Program instead.
The downside of ISO
So why not set your camera to a high ISO and leave it there all the time, so you can always shoot with a very fast shutter speed? The problem is digital noise. When you make your camera’s sensor more sensitive to light, you tend to get more errors—random pixels of unwanted color. Regardless of how low you set the ISO, you’ll inevitably see some noise in a photo. But the higher the ISO, the worse the noise gets. Compare the same scene (below) shot at ISOs of 100 and 1600. When you see the entire image at Web resolution, they may look pretty similar:
But zoom in to see the details, and it should be easy to tell which one is which:
The threat of noise shouldn’t scare you away from raising the ISO setting on your camera when you need to; when you’re done, though, you should make a point of lowering the ISO to the camera’s lowest setting. Most photos with even a moderate amount of noise look just fine on a computer screen or when printed at small size; noise becomes noticeable only when you make large prints or crop a photo to show smaller details.
In any event, the average camera’s noise performance has improved dramatically in the past few years, enabling you to shoot at previously unheard-of, crazy-high ISOs and still get very usable photos. Some cameras today go as high as ISO 12800—seven stops faster than ISO 100, and strictly the stuff of science fiction as recently as a decade ago.
Not only will you get more noise in high-ISO photos, but you’ll get more noise in low-light photos, such as those taken indoors or with long exposures at night.
If you’re shooting at higher ISOs and you save your photos in JPG format (rather than in your camera’s Raw mode), make sure that your camera’s noise reduction feature is turned on. Your camera runs a basic noise-reduction filter on your photos to get rid of the worst of the noise when it saves them initially. Some cameras have a more thorough noise-reduction filter specifically for higher ISOs; check your camera’s user guide to see whether yours does.
Software after the fact
If you’re willing to do a little extra work, I highly recommend running a noise reduction filter on your computer. You can find noise removal filters for your favorite photo editor, for example, or you can try a stand-alone program. Noiseware Community Edition, for example, is a free noise-removal program for Windows (the full version by Imagenomic for both Mac and Windows costs $80). ND Noise, a Java-based application, is available cross-platform. If you’re serious about fighting image noise, also check out Nik’s Dfine or Neat Image for Mac and Windows. Many of these programs come as stand-alone packages and as plug-ins for software such as Adobe Photoshop, Lightroom, and Apple Aperture.