Uh, you can't evaluate the final sharpening on a computer display. Even if you zoom out to 50-25%, you are looking at a higher rez image at 1/3 to 1.4 the resolution if your display is 100 ppi. All you can do is do the proper capture sharpening add creative sharpening and the final output sharpening should not break the image.
While that's essentially true, it may escape some readers why that is so. The reasons are quite simple and, more importantly, one can learn through experience how to deal with them.Proper Capture sharpening
is something that has not been understood well by many, for a long time, and some still don't. All it's supposed to do is recover from (as much as feasible) the sharpness (resolution and
contrast) losses incurred during the Capture process. Some are still using yesteryear's tools (because nothing better was available back then) with some added features to hide the shortcomings.
What many do not realize is that the Capture process losses can be accurately predicted and quantified because they are predictable physical processes. When we know what caused them (optics, diffraction, sampling), we can use modern techniques (like Deconvolution) to recover from them to a large degree. BTW, the optics not only hurt the original signal's resolution due to residual aberrations, but are also a major cause of contrast loss due to veiling glare (which is not really being addressed by the ancient tools).
Modern Capture sharpening tools are quite outdated in their design, and they look very much like USM sharpening tools with some extras for artifact suppression.
UnSharp Masking (USM) based tools are just that, yesteryear's tools. They mimic a process that was used with analog film, and turned it into a digital tool that produces the same kind of artifacts as there were in film, and then uses edge masking techniques to cover up most of those artifacts. That's ancient, and it's also what Guy Gowan seems to use, although he seems to focus more on the final output than on real Capture sharpening.
The image as displayed on screen, can help to look in detail whether the adjustments do or don't create artifacts, but it cannot give a good impression of the real effect on final output, mainly because the display resolution is too low. Of course, through experience one can develop a feeling of how things will ultimately work out in output. The experience part is often not mentioned by those who slam the usefulness of judging on display sharpening adjustments.Creative sharpening
is a term that inaccurately describes all of the actual processes at this stage, because from it's origin it's partly based on the same ancient USM sharpening techniques. Of course it usually has relatively little to do with sharpening, but more with local contrast adjustments, e.g. Clarity.
Modern implementations of this Creative stage of image manipulation can, thanks to modern Digital Signal Processing (DSP) techniques (e.g. Wavelets and other frequency domain based adjustments), indeed address the manipulation (amplification or reduction) of specific spatial frequency bands in specific areas of the image. But most of the Creative influence on the overall look of an image comes from Local Contrast adjustments, not real sharpening but the impression of sharpness caused by contrast. Contrast suggests subject matter surface detail will be brought out by harsh lighting.
To get some idea about how the result will look in final output, one can very roughly approximate it by zooming out to a true output size on screen representation. The finer detail adjustments cannot be accurately judged on screen though, because the display resolution is too low, and the quick down-sampling not accurate enough. Through experience though, one can develop a feeling for how things will work out in the final output.Output sharpening
is usually also hard to exactly jugde on display, again, because the display resolution is too low, and because a light emitting output device has a different characteristic compared to reflective printed output, and viewing conditions can have quite a different effect on both. Of course when output is generated for screen display, e.g. Web Publishing, it becomes easier to judge the final result for a given display type (although there are a lot of different display qualities available for which one can optimize).
Again, experience plays a large role here in judging how the on screen adjustments will translate to the actual output medium. So while the on screen displayed image is not an accurate preview, one can learn how to translate that preview to how things will actually look, and experience takes time to develop.