|
Look at these photos—100 percent details from images of an identical subject shot with a 10MP digital compact (note: I found similar results when shooting with 8 and 12MP cameras):
 |
ISO 100: Nice and sharp. You can see lots of fine details—just look at the hair and eyelashes—and dark areas (look at the pupil) are clearly rendered. You can make a nice, big print of this shot! |
 |
ISO 200: Still OK: Acceptable grain, but evidence of some deterioration can be seen in the darker areas. See the difference in the pupil? Still, it's printable, even at 8x10. |
 |
ISO 400: Going south: Grain becomes obvious, especially in darker areas, and we now start to see loss of sharpness. I wouldn't enlarge this image, but you might get away with a standard 4x6-inch print. |
 |
ISO 800: Getting worse: Grain is unacceptable throughout the image, causing blotched skin tones, loss of detail, and lots of mottling in the darker areas. |
 |
ISO 1600: Yuck! (that's a scientific term): Many compact cameras boast the ability to capture ISO 1600--but this is the kind of unacceptable image quality you can expect. |
My recommendation? Limit your camera's ISO setting to 200 or lower—but that recommendation is limited to the particular camera I tested, which is a compact digital camera with a small sensor. Here's how you can determine the upper limits of good image quality/high ISO for your camera by running test shots.
1. Put your camera on a tripod, disable anti-shake.
2. Use consistent lighting, either outside in open shade (adjust white balance accordingly) or non-flash studio lighting if you have it.
3. Choose a subject that won't budge and has a variety of colors and brightness. An X-Rite ColorChecker is ideal if you don't have a specially-designed-for-photo-tests mannequin, which is what I used.
4. Shoot at every ISO your camera has, in JPEG or RAW depending on what you're using (both if you shoot both), adjusting aperture and shutter speed to keep exposure on consistent.
5. Load each shot onto your computer and examine at 100% magnification, comparing each shot. To be absolutely sure, print each shot at the maximum size you expect to use and compare them at typical viewing distance (which is usually 12-18 inches or so). If using an online processing service such as AdoramaPIX, be sure to label each shot on the back, because otherwise they'll look identical.
Finally, what is acceptable graininess for one photographer will be too noisy for the next guy. A photojournalist might tolerate more noise than an architecture or landscape photographer. Use your judgement!
|
