I know some of capture sharpening is dependant on the image, but I would be interested in what sharpening values people find they are mostly using when doing landscapes with a Pentax 645D?
I hear various comments which range from "it has no AA filter, so don't use any" to "with the Pentax 645D you need aggressive capture sharpening"
Hi Phil,
The capture process is an inherently blurry process. So one should attempt to restore the original sharpness that was there to begin with. For instance, the lens has residual lens aberrations, and diffraction, and both add unsharpness/blur/streaking/etc. The camera/back adds more unsharpness/blur with an IR filter and sensels (with or without microlenses) that have a certain aperture shape, and sampling density (which sets a hard limit on maximum resolution). Then there is additional unsharpness due to the Bayer CFA demosaicing process. In addition there may be an overall camera shake and mirror/shutter vibration.
To restore the original input signal as much as possible, one should use deconvolution Capture sharpening, a mathematical process which will restore resolution by taking out the blur component, as much as possible without creating artifacts of it own. However, because all sensors also produce aliasing, due to the regular sampling pattern of the sensor array, we also need to kind of avoid a full restoration of all sharpness, because that would unveil the aliasing in 'full glory'.
So, we are facing a balancing act; restore the original signal, but avoid emphasizing the aliasing artifacts.
The solution starts with the Raw converter. A good Raw converter will try to avoid the demosaicing and aliasing artifacts (zipper effect / mazing / stair-stepping / false color artifacts / etc.) and interpolate to a smooth version of the original input. Lower quality conversions will suffer more from subsequent sharpening.
We then need to remove the overall (lens/camera/sensor) blur in a process that's usually called 'Capture sharpening'. That blur is very much dependent on the aperture used, more so with narrower apertures due to diffraction, so it is somewhat variable. Since most Raw converters use sub-standard sharpening algorithms, I prefer to do most of my image editing on a TIFF output file.
Lightroom/Adobe Camera Raw are not too bad, but certainly leave things to be desired. Its deconvolution will produce very gritty conversions very fast, so you can only push the Detail slider so far (perhaps 50 is the limit), but you first need to get the radius setting exactly right (depends on lens and aperture used). The masking is useful to avoid sharpening of smooth surfaces, and is almost mandatory due to the gritty sharpening.
Other converters, like e.g. RawTherapee, which has some very good quality to offer, both the Amaze demosaicing algorithm, and Richardson-Lucy deconvolution sharpening, produce a lot of high quality image detail.
In a Photoshop oriented post-processing workflow, there are several other possibilities to suppress the negatives, and enhance the positives.
Cheers,
Bart
