Not that I don't appreciate the effort you took, Robert, posting the sharpening sample, but since you didn't connect the graph analysis in Jack's blog articles to the look of the sharpening results I fail to see how the science helps.
Well, I didn't connect them directly, but Jack's graph analysis for sharpening is exactly the same as mine, showing the effect of sharpening (in particular incorrect sharpening) on the edge profile and MTF. So the graph analysis shows that a particular sharpening is sub-optimal (the FM3 one in particular) both from a haloing and resolution point of view, and the photo of the rock shows the same thing, which shows that theory and practice are in line with each other. What the theory shows more clearly than the photo though is that resolution, especially at high frequencies, is reduced by too much sharpening. That makes sense if you think about it ... the sharpening is causing halos which blur the fine detail.
For me it's always been the slider behavior and positioning relationship between Amount, Radius & Detail in ACR that affects image sharpness differently depending on distance the detail was from the lens combined with resolution/sensor size at the time of capture.
For instance a small Amount above ACR's +25 like say +40/Radius-1/Detail-25 sharpening detail lit at 45 degree angle just feet from the lens is all that's needed as opposed to detail farther away lit at 75 degree angle needs a larger Radius and Amount-50/Detail-50. Sometimes I can crank Radius to 2.5 and increase Detail to remove "mosquito" edge artifacts, but it's different image to image. How do you connect science analysis to so many unknowns and inconsistencies as to what's really going on with software?
I notice ACR slider position relationship changes as it acts on various clump size of detail which has not been characterized/profiled in these discussions.
I agree that at the end of the day all post-processing is a subjective thing, and having rules to say that you should do ACR sharpening at x followed by some other sharpening at y is much too simplistic as it doesn't take into account how big the print is, how far away it is being viewed at, whether or not you actually WANT halos around the edges, how good your lens and focusing are and so on.
But I think the 'science' can help remove fuzzy thinking. For example, I took the same shot that I've posted above and applied a 2px gaussian blur to simulate out-of-focus. I then tried various sharpening/deblur corrections:
As you can see, allowing Focus Magic to do it's own thing (it selected a deblur of 6px) has given pretty good result. Sharpening in ACR marginally improves MTF50 and MTF Nyquist, but at the cost of haloes (I did the very best I could in ACR and found that a high amount and fairly high radius (around 3) gave the best result).
So what this tells me is that if I've messed up the focus in a shot but I still need the photo, that I'm wasting my time trying to correct it in LR/ACR. If the only other choice is Focus Magic, then I am probably just as well off using that in one pass (3rd line down), although the one in the 2nd line down is better but much more time-consuming (I applied Focus Magic 3 times with varying amounts of Fade after each pass).
What I will try is another deblur plug-in like InFocus. If it is much better than Focus Magic for this type of defocus then this may give me a reason to buy it. I'll also have a look at piccure+ as this seems to be getting good press in this topic.
Robert
