I have a few pics that I have successfully sharpened somewhat, and some others not so good with a bit of noise as well.
Is there a general method of one before the other?
Good question. Everyone might well be different, but I always use noise reduction first. Often I don't need to though with the low ISOs used here. Sharpen always last here.
The conventional wisdom that I have always followed is that you sharpen at least twice, possibly three times: input sharpening, creative sharpening and output sharpening.
I would guess that noise reduction should be fairly early in this process, around the same time as input sharpening or soon after, but that is just a guess. I do not know whether it even matters mathematically. In raw processors such as Lightroom, the user has no control over the order in which those operations are done.
The reason for 'input sharpening' is to correct for the AA filter. This isn't a perfect filter and reduces high frequency detail more than is desirable. So you boost it a bit to restore it. If you're camera doesn't have an AA filter you don't need to do it - also remember that some de-mosaicking algorithms have quite a bit of inbuilt sharpening, if you sharpen more you can get undesirable artefacts. The reason for 'output sharpening' is to correct for losses caused by resampling to the output resolution. Better resampling algorithms (such as Lanczos) shouldn't produce much sharpening loss, and may not need sharpening. Very crude resampling (such as nearest neighbour) produce nasty artefacts that will only be accentuated by sharpening.
Thanks all, clear as mud, until the tech stuff hit the fan. Just kidding. So now I know what an AA filter is / does but does my camera have one? Being a small sensor I guess it would. So the logical next question is how much input sharpening? How long is a piece of string? What looks best I guess. My little bit OCD nature doesn't like ambiguity. But I don't think I need to worry too much...
Interesting... They don't occur in the order you create them? I have only played with RawTherapee a little bit and I see the list of ops to be applied. But atm I am just using the simple tools in FastStone.
As I understand (and I may be wrong) most sharpening and noise reduction methods work to some degree at cross purposes. I say most because recent NR tools seem to be addressing this short coming. What I have read suggests that if possible you do not want to sharpen noise, so apply some NR before sharpening is desirable. And also if possible limit the sharpening to areas that need it, i.e. edges, and NR to areas that need it, i.e. smooth surfaces like skies. And there maybe more than one step in the workflow where NR or sharpening is required.
If I may draw a comparison with ACR, which I believe uses the same processing (develop) engine as Lr, there is some degree of control. With ACR we have the Detail panel and can apply sharpening and NR (and suddenly now AI NR too!). That is the first line of attack. And then with masking we can apply localize sharpening and NR, albeit the controls and parameters are a bit different from the Detail panel. And the localized sharpening and NR are cumulative, so you can paint in more or less and layer up several applications.
While Adobe are not explicit in their documentation of the Detail panel as to the order in which NR and/or sharpening are applied, I think that there is a clear ordering between the Detail panel and the localized (masking) NR and/or sharpening.
There are two ways of doing any pf these things - the technical, methodical way and the trial and error way. In terms of whether your camera has an AA filter, it's less likely top have one with a small, high resolution sensor, because diffraction's doing the filtering and AA filters are expensive. Most high resolution large sensor cameras also omit the AA filter these days. So, tell us the camera we're talking about and we can see if it has an AA filter.
Then, how much sharpening to apply - the basic 'trial and error' method is to increase until you begin to see sharpening artefacts (bright fringes round high contrast edges) then turn it down a bit until you can't. The method of sharpening matters too. The old 'unsharp mask' method is more prone to artefacts than newer wavelet based methods.
If you want to go the methodical way, then you test your camera/lens to find its MTF curve, then you characterise the spatial frequency response of your sharpening tool, then you work out the degree of sharpening that precisely compensates for the loss of sharpness introduced by the lens. But most of us use trial and error. You pretty soon work out how much to apply with your camera.
Input sharpening is also useful to correct for blur due to the finite sensor sampling aperture, which, these days, is usually near 100% of a square whose edges are the same size as the pixel pitch. Input sharpening also can be used quite effectively to correct for diffraction.
As to the ordering. Lr and ACR don't give you a choice. I think linear sharpening and blur operations give the same results regardless of order, assuming infinite precision and no resampling. Order matters if either operation is nonlinear.
Bryan: I think some of what has been typed here may be too technical for you at this point in your journey.
Danno gave some great advice and it is what I would recommend that you do as well: in the post-processing software of your choice, open the image and de-noise it as one of your first steps. Do it to taste. Then, do whatever else your creative juices want you to do and, then with one of your last steps, go ahead and sharpen the image to taste. Play around and explore settings and see what works for you.
One thing to consider is what RAW processor you use. For example, DxO PhotoLab Elite and Pure RAW using Deep Prime do their magic upon export. You can only see a small preview window of the effect while processing an image.
Lightroom/ACR, Capture One and RawTherapee have a big preview but they also do their processes while exporting file(s).
RawTherapee tells the order of processes here: rawpedia.rawtherapee.com/Toolchain_Pipeline