Sounds interesting. May I ask if you could share which tools you are using, and prehaps a bit on a suggested workflow?
The preferred workflow depends on what your goal is, and what the quality of your base material is. The above mentioned book gives a good overview of what can be done with the traditional tools in Lightroom and Photoshop, but I have a different opinon about the default use of Capture sharpening and the method used. One can best do noise reduction and removal of chromatic aberration artifacts at this initial state.
I'd made a distinction between upsampling, e.g. for large format output and matching the native PPI resolution of the printer, and downsampling, e.g. for web display. They both benefit from using different methods/tools.
So you have to ask yourself, what is your base material, which system was used to produce the image file? That will already steer how to proceed further. Assuming a digital camera was used, the first step is to use a good Raw converter. A good Raw conversion will produce a better base image to do postprocessing on. Some Raw converters also allow to boost productivity e.g. by integrating tethered shooting and good postprocessing options, so that would already steer the workflow. A program like Capture One Pro also has pretty good sharpening capabilities, including the possibility to correct for sharpness fall-off towards the corners. Some cameras produce files that are more difficult to sharpen, but most cameras use an AA-filter which introduces some additional softness that could be restored with a process called 'deconvolution' restoration (real sharpening instead of edge contrast / acutance boosting).
I prefer to not sharpen before downsampling, not even capture sharpen (it also saves time to skip a non-productive step). You can find some more background info in
this thread. I show examples of using FocusMagic as a deconvolution restoration/sharpening tool there.
For enlarged output files one can either (capture) sharpen at the original size (after noise reduction if needed), then enlarge/upsample and do a final sharpening to compensate for the enlargement and output medium losses, and tuned to viewing distance. But one can also do all the sharpening at the final output file size. Depending on one's tools there is a risk of introducing artifacts at a smaller size, and then enlarge those artifacts. On the other hand, depending on one's tools, enlarging itself can also introduce artifacts, and sharpening after enlarging will accentuate those artifacts, so what's best depends on image content and tools used.
In general when using Photoshop it is beneficial to use a sharpening layer (or layers) set to Luminosity blending mode with a "blend if" roll off towards things that are already high contrast to avoid clipping and reduce halo risks. Something like this:
[attachment=20371:Non_clip...arpening.png]
It gives lots of control because one can tweak the blend-if parameters and later add masks for selective sharpening and tweak the opacity of the layer.
Cheers,
Bart
