Bart,
Can you provide a formula to calculate the ideal sharpening radius based on the aperture and pixel size? I'm not a D800 owner but I am interested in this for use with other cameras - particularly my Rollei AFi with the Aptus 12 back. Which I think it has 5.2u pitch.
Hi Eric,
There is a simple answer (but it doesn't work as good with wider apertures), and there is a more detailed answer. The latter is better addressed in a separate thread, to keep this thread on topic.
The simple answer is for the scenarios where diffraction so clearly dominates as with these narrow apertures of say f/16 and narrower, when a maximum DOF is required instead of maximum resolution (which is a valid trade-off, but with consequences for large output). When plotting an Airy Disk pattern and a Gaussian Bell pattern together, it is possible to find a kind of best fit between the areas of main contribution of both patterns.
As a rule of thumb I find the following approximation to work quite well for luminance (the weighted average of Red Green and Blue, 564 nm):
Radius = 0.26 x N / SP,
Where N = F-number, e.g. 22, and
SP is sensel pitch in microns, e.g. 4.88 for the D800/D800E.
Do note that this is only dealing with the diffraction pattern, therefore it works best when diffraction so clearly dominates the captured result. One could see this as the smallest practical radius setting. For wider apertures the residual lens aberrations have an increasingly larger additional influence, and wide open also defocus effects gain in importance. Due to the complexities of the combination of different Point Spead functions/shapes, I've made a tool that allows to empirically find the best radius for any lens/aperture/camera/Rawconverter combination. I'll be introducing this tool later today in a
separate thread.
How does this work when a capture level sharpening is applied and then another pass with detail slider in LR later?
That depends on one's workflow. When we want to strictly separate the Capture/Creative/Output sharpening steps, then Lightroom only allows to address Capture sharpening on the original Raw conversion, and then do an additional run of sharpening by using a local Sharpness brush, or by another Development run on an earlier Tiff conversion result. Or, as a compromise, one could try and mix the Capture sharpening and the Creative sharpening into one. In that case, the smaller Capture sharpening radius sets a lower boundary for the range of radii to consider.
Life does become easier if we concentrate on Capture sharpening, because that is where deconvolution offers a major contribution, also in avoiding halo which will only get worse when doing the Creative sharpening. It is also where noise reduction may interfere with boosting the effect of the detail slider. By concentrating on getting that Capture sharpening stage right, the rest (e.g. the local Sharpening brush, and Clarity) becomes so much more predictable. But more on that later.
Cheers,
Bart
