|Shot with Canon Powershot A630 ISO 800 |
Example of noise evident in leaves.
Noise is an artifact in digital photography that occurs when there is insufficient light to produce a good image. While it is commonly equated to film grain in film photography this is not an accurate description of the effect. The equivalent of film grain in digital photography is pixel size. Just as more expensive digital cameras have less pixilation higher quality films had less grain, this was not a function of the amount of light, but of the quality of the exposed surface area in the camera. Todays digital cameras have so many pixels packed into them that they have far surpassed even the highest qualities of fine grain 35mm film.
Then what is the equivalent of noise and how does low light cause it? The best equivalent to noise comes not from photography, but from television, analog television to be precise. Snow on an old TV is like noise is in a digital camera. It is the result of the signal not being strong enough. In TV, as the equipment amplifies that signal it also amplifies the carrier wave, in photography, the lower the signal (light) the more the carrier wave (noise) gets amplified.
In Digital photography each pixel you see is actually 4 pixels: one red, one green, one blue, and one for tone (light to dark). The pixel rendered in a JPEG is a representation of these four values, in RAW format all four values are stored and can be manipulated individually. When there is insufficient light for a photo the ISO is increased, this is the equivalent of amplifying an analog signal. This has the effect of amplifying the base signal of these pixels, which is what gets rendered as noise. Just as the current carried in one strand of wire will induce a current in wire adjacent to it, the pixels in the camera sensor interfere with the pixels next to them. This is why sensor size is so important and why cell phone cameras with sensors about the size of your pinky finger nail have much more noise and much poorer quality images (particularly in low light) then full frame DSLRs. The closer together the pixels are the more interference they create in the adjacent pixels. This is also why noise appears in shadows first, this is the area of the photo with the weakest signal.
Now that you know what noise is, there is really only one way to prevent it, adding more light. This has the effect of making the signal stronger so that it overrides the noise that is always present. Newer cameras are less sensitive to noise and are better at filtering it out of an image, but the only sure way to prevent it is to have enough light for proper exposure. This is not always practical though, in these cases you need to know how high you can take the ISO before the amount of noise is to great for the image. This will very some from image to image as photos with more dark shadows will show more noise then will pictures with a lot of mosaics of color. A general rule, the busier the subject is the less noise will be noticeable. In the example above, the noise is much more evident in the plant leaves then in other areas of the image. The close up shows just how prevalent the noise is. As always consider the final usage of the image to determine what end quality is needed.