Can an argument be raised that capture sharpening is best done to the RAW file while inside a parametric editor (LR/ACR) rather than to a converted file (.PSD/.TIFF) while inside PS? Or does it matter?
Hi,
It looks like the current crop of Raw-converters first do a demosaicing of the Raw data, and then do their post-processing (including generic sharpening) on the conversion result.
In that case it would not make matter much, other than that the parametric workflow postpones he actual operations till final export. At export time, the parameters are applied in a predetermined order, possibly with sharpening and noise reduction as one of the earlier operations in the cue, but not necessarily the first (e.g. preceded by lens corrections).
That would mean that it doesn't make much difference where the generic sharpening takes place, but it also depends on the type of sharpening operation and other image altering steps taken. In fact, the sharpening by FocusMagic applied on an almost finished file, can produce better results that lesser quality sharpening operations applied early on.
As for
Capture sharpening in particular, that's a bit different. If done well, it is best done
very early in the processing pipeline, just like noise reduction. It would make sense to even do most of it
before demosaicing, because that would produce more accurate data to demosaic. However, that requires
much more complicated Raw conversion.
I get the impression that that is what happens in Canon's Digital Photo Professional (DPP) raw converter, when the Digital Lens Optimizer (DLO) is used. The Raw data is recalculated to a version without lens aberrations, with lens blur removed, and resaved in the Raw file container as an additional raw dataset, thus almost doubling the file size. That optically corrected raw image data is then demosaiced into a much better quality, and less sharpening is required as a post-processing step.
So, as things are today, I think it doesn't matter much when the generic sharpening is applied, although it would technically be better to do Capture sharpening in particular early in the process, but then one should have better tools than most of us have today. Without these better tools, we might as well do our sharpening later in the process, which would also avoid surprises when we decide that we need larger output (which magnifies cascaded artifact generation). Doing it late in the process also allows to vary it locally, e.g. not on smooth gradients, or more in corners.
Cheers,
Bart