by James McIntosh
Digital sharpening refers to methods used on digital images (those from a digital camera, or from a scanner) to increase the real and apparent sharpness (acutance) of the image. Most digital cameras have built-in sharpening and most scanner software also provide sharpening. Without these, most consumers would be very unhappy with the results. Unfortunately, sharpening too much and too soon can be detrimental to a high quality result.
A lot can happen to the light reflected by a photographic subject before it ends up on a photographic print or an image displayed on a computer monitor. It travels through the air and a lens. It strikes a piece of film, or a digital sensor (after going through an anti-aliasing filter). The photographer scans the film, resizes it, applies tone curves and other manipulation to enhance the image. Many or all of these steps can result is some bits of light ending up where it's not suppose to be. This results in an image that is not as sharp as it could be.
The most common method used to sharpen digital images is called Unsharp Mask. This oddly-named method uses a 'mask' or matrix of values that represent a Gaussian, or 'bell shaped' curve in two dimensions, with the highest values in the center. This mask represents the best model for how images become unsharp, some values from neighboring pixels affect the pixel value. By subtracting this value from the pixel, you remove some amount of unsharpness. Some sharpening software, such as CrispImage, use a more sophisticated mask and employ methods to control halos and noise enhancement.
Most or all sharpening should be done at or near the end of processing, especially resizing. Some advocate applying sharpening in three stages: input, creative and output. If done properly, this approach can work well, but over sharpening before the image is enlarged (input and creative stages) can result in a significantly reduced image quality. If the image is significantly reduced in size (e.g. for web display) any sharpening done before the resizing will have little affect on the end product. Turn sharpening in digital cameras and scanners off, or to the lowest setting, and sharpen when the image has been resized to its final size.
The amount and specific settings for sharpening an image for printing on an inkjet printer will be significantly different than those for sharpening an image for display on a computer monitor. The human eye will perceive the smaller pixels of a high resolution printer differently than those on a monitor. Printers will also blur the image somewhat due to dot growth. In general, an image must be sharpened much more for a print than for monitor display. Some sharpening software have optimizations for sharpening targeted to a specific resolution, which are usually very helpful.
Compression using the JPEG method can leave artifacts especially near edges that sharpening will only worsen. Digital cameras should be set to the lowest compression setting possible (highest quality) to avoid this. If the image is for web display, a fair amount of compression can be applied to the finished image (after sharpening) while maintaining the image quality.
Understanding Sharpness by Miles Hecker.
Understanding image sharpness and MTF by Norman Koren.