The unpleasantness of noise can affect more than your ears - it's a problem that frequently appears in photography, too. Digital image noise is the appearance of random changes to the colour of pixels. Unfortunately, even small doses are so effective at ruining an image that you'll end up reaching for the delete button.
Like many of the nuisances facing photographers, avoiding the problem is the best approach, but noisy images are easy to miss on a camera's tiny LCD screen. To make identifying the problem a little more confusing, some LCDs can make it seem that the image has noise, when in fact it's simply the compact screen size that's creating a false impression.
You may think the true test is back at your computer, but that's not always the best indicator either. Some graphics programs are better than others at displaying images. Plus, there's the issue of monitor resolution and number of displayed colours - some settings are better than others, particularly if you're using an LCD monitor. A simple change of graphics program or video settings can help identify the true culprit of spotty images, but it goes to show that you shouldn't automatically blame your camera.
What is digital camera noise?
At the heart of your camera is a sensor (called a CCD, or charge-coupled device) made of millions of photodiodes (light sensitive dots). These absorb light particles and convert them into electrical charges. The more light that falls onto each photodiode, the greater the electrical charge produced. If no light is received, the pixel should have no charge and therefore register "zero". In other words, the pixel should be black (this is a simplification of the process - in reality, it's far more involved). Under certain conditions, the occasional pixel may gain or lose some of this charge. This may cause a black pixel to appear very dark green, blue or red. It doesn't sound like much of a problem, but as mentioned above, it doesn't take very many noisy pixels to ruin a picture.
In normal shooting conditions, noise should not be an issue for new digital cameras (though some cameras are worse than others). Noise becomes evident when cameras are struggling under poor light. Cameras try to compensate for low light by amplifying the signals coming out of the sensor (also known as increasing gain). This causes minor differences between pixels to become exaggerated. A black dot may get turned to another colour due to the amplification. In addition, the camera may further process the image to inflate the problem.
You can see the effect of noise when taking long exposures at night or adjusting the ISO setting. The ISO value represents the sensitivity of the pixels to light. The higher the value, the greater the sensitivity. Unfortunately, the price you pay for this sensitivity is more noise.
Two ways cameras try to beat this problem involve the use of noise process filters and subtraction techniques. The subtraction method works by snapping a second picture of pure darkness (closed lens) at the time the original image was taken. This is used to identify noisy pixels and remove the errors from the final image. The second approach is to use special hardware features that either stop noise at the sensor or knock out the noise with specialised built-in signal processors.
To minimise the impact of noise, you should always shoot at your camera's highest quality settings - use the highest resolution and best quality setting for the file type. If possible, try decreasing the shutter speed or add light to the surroundings.
The quality setting generally refers to the JPEG compression. A highly compressed file is smaller and frequently called "low quality", small or coarse. Terms such as fine, superfine, large and highest quality all describe the better settings. The price you pay for a smaller JPEG file is the introduction of artefacts (random splotches). If your image has some noise before being highly compressed, the resulting photograph can look just plain awful. With memory cards becoming much cheaper, there's no need reason to skimp on quality. Plus, it makes little sense splurging on a top quality camera, only to destroy the benefits by using poor settings.
The effect of noise can vary greatly and, depending on the severity, you may need different techniques to correct the problem. Since noise frequently appears under poor lighting, this can provide a quick solution to the problems: altering contrast. By adjusting the contrast and brightness by a few per cent, it is possible to knock down the noise so it is no longer an issue.
Noise reduction software
If tweaking the contrast didn't help, there are some tools that can help with the job. Many work well, but they all tend to suffer from the same problem - complexity. Generally, you're given sliders and adjustment options that will help you optimise the amount of noise reduction. On the face of it, these settings give you control, but the problem is that it can take a lot of practice to get them just right. In some early attempts, I found it difficult to get the photos to look any better. After a few hours of reading help files and manuals, the results started to improve - which suggests that noise reduction is not a very intuitive process.
The first conclusion I made was that the more noise in the picture, the easier it is to get a noticeable improvement with noise reduction software. For images with smaller amounts of noise, I was getting better results by adjusting the contrast and brightness. In these cases, the noise reduction options tended to change the overall tone of a picture, but hardly improved it.
To find which tool suits your skills and camera, it's best to try a few different programs. Some are stand-alone tools, but noise reduction is starting to appear as a standard feature in graphics software. One example is Paint Shop Pro 9 - it includes a "Digital Camera Noise Removal" filter (from the menu select Adjust-Photo Fix-Digital Camera Noise Removal). A trial version is on the CD, so you can test it for yourself.
If you are looking for a free tool, try DCEnhancer. Generally its results ranged from okay to good; however the lack of a help file left me unsure of how to fully optimise the image. Moving the sliders was simple but, without knowing each one's function, the process was hit and miss. Another limitation is that you couldn't set the JPEG quality for saved images. Then again, you can't argue with its price (freeware - see the CD for a copy).
A more detailed program is Noise Ninja ($US30, shareware, a trial version is on the CD). It comes with an impressive help file that can have you mastering the program in a few minutes. The introductory tutorial is also worth your time and will save a lot of stumbling in the dark.
One last tip: when using noise removal tools, start with a low intensity setting and slowly turn it up. At higher levels, all the tools started blurring the image.