The Deconvolution Sharpening tool included in C1 Pro 10 appears to be a "one size fits all" adjustment. There are no sliders that can be used to adjust the parameters of the sharpening such as radius and amount.
Hi,
In a way that's correct. The control is labeled 'Diffraction`Correction' and it looks like that is exactly what is being deconvolved, nothing more nothing less. It therefore does not require additional controls (assuming they did their homework). Diffraction is a constant "blur" per unit area based on Aperture and Wavelength. Because the Bayer CFA lumps many wavelengths in each (usually) R/G/B filter bandpass, one could take an assumed average wavelength for each filter.
The only things that can change the 'amount' of blur per pixel, are sensor element aperture shape/size as concentrated by optional micro-lenses, and focus distance (= magnification factor). Denser sampling (with smaller sensel pitch) of the given diffraction pattern will sub-sample the diffraction pattern in more detail, and thus it will be spread over more pixels. Closer than infinity focus, combined with focal length, will magnify the projected image and the diffraction pattern.
I do not yet know if the latter variables are actually used for deconvolution, e.g. by reading them from the EXIF (if available, which might be an issue with some lensmount adapters). It would take a series of additional tests to establish that, but it would be simpler if someone from PhaseOne could chime in and simply clarify that aspect.
These adjustments are possible with all of the other Deconvolution Sharpening tools I have used. I doubt that C1 10 is applying some sort of automated image analysis algorithm to determine the optimal settings, because the time delay in processing this adjustment is not that long at least on my Mac Pro.
A lot of the processing can be optimized, because there are not many variables. If the sensel pitch and magnification refinements are ignored things are really straight forward, but the result will be sub-optimal. Maybe some average (e.g. 5 metres, or e.g. hyperfocal distance if focal length and aperture are known) is used for distance and the camerera model (Raw file) profile triggers a sub-sampling factor lookup, who knows.
The deconvolution is probably relatively simple (not iterative), and thus won't consume that much time, especially if assisted by GPU an Open CL. Maybe it's more elaborate and adapts to image content, it's their secrect sauce. Anyway, it seems to do its job, even on close-ups.
I tried a close-up test shot taken at f/5.6 and the same shot at f/16, and they looked pretty much the same in the plane of best focus, once both were diffraction sharpened. In theory there is potentially an additional focus shift effect on sharpness when only aperture is changed, which would make testing/comparing a lot more time consuming.
The interesting thing is that it looks like they chose to treat diffraction as a separate blur source. Once that blur is removed (and it dominates at narrower apertures anyway), one can get pretty decent results with subsequent USM like edge-contrast boosts. However, even more detail from lens aberration induced blur could be restored with an additional deconvolution aimed at that different blur pattern., as would be possible for defocus blur with yet another blur pattern. Maybe those are things they want to implement in the future. There is also a risk of exagerating aliasing artifacts, like those caused by sensors without an Optical Low-Pass Filter (OLPF or AA-filter).
OTOH, I generally found that the same settings in Focus Magic worked quite well for almost all of my high res files.
Yes, but FocusMagic is pretty awesome, and requires an additional host-program like an image editor. The charm of the Capture One approach is that one can do most of the work in the Raw converter itself.
It will be interesting to compare the new Deconvolution Sharpening in C1 10 with Focus Magic. Has anyone taken the time yet to do that?
As said, FocusMagic can also tackle lens aberration and defocus blur. The mix of several blur sources usually creates a kind of modified Gaussain blur. So it should be able to do even more restoration of the original signal, but it is interesting to see/demonstrate how small the differences are becoming.
FocusMagic will probably need lower settings when the image is already diffraction deconvolved. Lots of testing ahead if we need to get to the bottom of things, but sofar it's a very good step in the right direction. Also the clear separation in the 3-step sharpening, and local sharpening in adjustment layers, is a huge benefit. Capture One Pro already had Sharpening Fall-off correction, we now need to test how that combines with diffraction deblur.
Maybe some answers will already come from the 3-step sharpening Webinar that is planned for tomorrow.
Cheers,
Bart