Cdouthitt wrote:
The best tool for sharpening, in my humble opinion, is buying the sharpest lenses one can afford and working on a good shooting technique.
Thats a recipe for a really good apple pie, but the question is about orange marmalade! There is no connection between the resolving power of a lens, assisted by good technique, and what is done with a "sharpen" tool in software! None. Two different things.
Software sharpening is
always useful on an image produced by a camera using a Bayer Color Filter Array (almost all DSLR's). The reason is because each image pixel is defined from a matrix of sensor locations. Because a matrix is used the shortest pixel distance that is possible for a tone transition is about 6 pixels, but is more commonly about 12. Ideally it would be 1 pixel from pure black to pure white, but that is impossible with a Bayer CFA.
Sharpening reduces the number of pixels needed for a tone transition.However, there are three common ways to do that, each with slightly different effects. What is called a high pass sharpen tool, or just "Sharpen", looks for multiple consecutive tone transitions and makes them have higher contrast. It has less effect on a single tone transition, but a technique called "Unsharp Mask", or USM, looks for just one transition (and averages multiple transitions to just one tone) adds contrast to single tone transitions. Obviously those two techniques can produce very different results. USM can be adjusted to see each transition in a sequence as if it is just one single transition, and can in effect do almost the same thing as a Sharpen tool. One significant difference is that USM cannot be reversed, while a high pass filter can be adjusted to reverse the effects of a low pass filter.
One other technique, for those who get really serious, is deconvolution. The best example is the Richardson-Lucy technique (used to adjust for the error in the original Hubble telescope). Unlike the other two methods that do not really increase resolving power, and merely add visual acuity with increased contrast, deconvolution can actually increase resolution. It just is not easy to use though...
Another variation of USM and/or Sharpen should be mentioned too. "Smart Sharpen" is a technique to find the significant edges in an image and apply sharpen only where it is most useful. That has the advantage of not sharpening noise, for example.
What's it all mean though, and how do you use that information? Realize that if an image has been reduced in size from the original, USM will probably have more effect. If the image as been enlarged Sharpen will probably have more effect. Also there are different filters that can be used, and a sharper filter works best on images that have been reduced in size, while a smooth filter works best for enlarged images.
Any particular image will benefit from either USM or Sharpen, and using both is probably the best solution.