I had read some references about deconvolution sharpening before in regards to countering blur, specifically gaussian blur, but never connected any dots for an actual application. Then recently a fellow member tipped me off about using it as a diffraction correction technique in
this post. Thanks to a handy tool in my raw converter of choice, Rawtherapee, I immediately went to experimenting with it, to many satisfying results... Since the issue/concern/factor of diffraction comes up quite frequently here on the board, I figured we should have a thread here about deconvolution sharpening, along with some examples...
The linked post above talks about using it in the context of large printing, where I am sure it also works wonders, but I've found that it also helps in web-sized improvements, and therefore should be helpful in any situation where diffraction is a concern. Personally I've been using it or at least trying it on whenever I'm working with a F13 or smaller image, using whatever intensity fits best to my eye and purpose (always viewing the effect/result at 100% magnification).
This post (same post linked to above) also discusses how to use deconvolution sharpening in Lightroom, which I can't speak to, because I don't use Lightroom.
Since I've done my deconvolution sharpening examples below using Rawtherapee, here is a link to their little bit of info about it pertaining to Rawtherapee, where there is also a link to the Wikipedia on it:
Richardson Lucy (RL) Deconvolution Sharpening. I am also a Gimp user and there is also a deconvolution filter in the G'mic plugin as well*, if you're interested. It's not as robust as the filter in Rawtherapee, but it works. (It doesn't have the 'amount' and 'dampening' options, but it does have two blur type options that I haven't fully explored yet.)
Without further ado, here below are some examples. First, I offer a 100% crop from a
K-5, DA15, ISO80, F22, 30 second image. The only adjustment to the DNG in Rawtherapee was to turn on a K-5 dcp, using the embedded tone curve.
The cropped
image on the LEFT has the RL deconvolution sharpening turned ON and turned up high (settings also pictured below) and the
image on the RIGHT is with deconvolution turned OFF.
Here are the Rawtherapee deconvolution settings for the above example:
Next we have the full sized version that the above crop was cut from, just resized to 1024px wide. I also applied USM after resizing on both images (same amount to both versions).
This is with deconvolution sharpening turned OFF:
This is with deconvolution sharpening turned ON:
Next, the same image worked up rather aggressively into a black and white version using Rawtherapee. The resize, some dust spot healing and same amount of USM was applied to each in Gimp.
This is with deconvolution sharpening turned OFF:
This is with deconvolution sharpening turned ON:
These were the deconvolution settings used for the black and white version:
*Here's the deconvolution sharpening filter available in the G'mic plugin manager for Gimp:
Finally if anybody actually manages to make it to the bottom of this post, here's the full sized unprocessed image in 16-bit tiff export from Rawtherapee hosted at Mediafire, for anybody who might be interested at downloading both versions to compare:
With deconvolution sharpening turned ON (93MB)
With deconvolution sharpening turned OFF (93MB)
So, all that to say, go give it a try! And let me know what you think!
Last edited by todd; 04-13-2016 at 06:47 PM.