I developed the following raw image taken with my APS-C sony a6600 with both darktable and rawtherapee. DSC00662.ARW (24.0 MB)
This is a detail of the image coming from DT
What I find surprising is the difference in sharpness of the image between DT and RT; in a nutshell, RT seems by far sharper than DT.
Why is that and, in case, where am I wrong?
Is there any way to improve the sharpness of DT to make it comparable to that of RT?
Ciao
Vittorio
In RT you have three kinds of sharpening activated: capture sharpening, unsharp mask, and edge sharpening. You also have wavelets enabled, which effects perceived sharpness.
In DT, you have the sharpen module and high pass.
So, this isnât really good comparison as youâre not using the same modules.
Iâve posted about the difference between darktable and RT many times.
Itâs not just sharpening but amount of detail and the clean image.
I thought it might just have been because I use Fujis X-Trans sensor, and the RL deconvolution sharpening.
Over the years I have tried everything in dt from highpass, sharpen and contrast equaliser but nothing comes close.
Then along comes ART, in which I can achieve slightly better result then even RT using USM.
So, in summary, I always thought the difference was because dt did not have a deconvolution module but achieving the excellent results in ART using USM, I have to say I havenât a clue why.
But the difference is there.
This is what I get by only using two instances of the contrast equalizer. Still seems far away from de result in RawTherapee, but I think it already somewhat closer than your darktable attempt. Maybe this is what youâre looking for?
It might also have to do with parameter ranges and sensitivity. I find that I could easily exaggerate or destroy an image using max values in RT. As @Thanatomanic suggested, a way around the subdued nature of dt is to use more than one instance of a module; or use multiple modules to coordinate the enhancement, although iteration may sometimes introduce issues.
Concerning the topic: I use both programs and I think I have not noticed that one is sharper than the other, although the raws developed with RT seem to have more resolution. I think it has to do with the demosaicing. However, it is possible to sharpen photos very well in dt with two instances of the contrast equalizer.
Thanks a lot for the many answers.
Willing to stick to DT for many reasons, following your valuable suggestions I now develop my images according to these steps (as far as sharpness is concerned):
demosaic with Amaze;
exposure module;
Color calibration;
Filmic rgb;
Sharpen module disabled;
Local contrast;
(More important!!) 2 instances of the Contrast equalizer module tweaking luma and edges fine details.
In so doing I get very similar results in terms of sharpness to Rawtherapee.
Of course, Iâm not a âsharpness addictedâ but landscape and generic architectural photos require this attitude.
If you look closey to the discription of the âedgeâ tab you will see that it will only work if the luma or chrome curve has been set.
A separate instance of the contrast eq. with only the the edges curve activated wonât do anything. (unless Iâm wrong)
This level of sharpness is definitely hard to achieve in darktable (3.4.1). Hereâs my attempt, mostly by being careful with the denoise and using a combination of contrast equalizer and sharpen. Still not sure itâs quite as good as RT manages but itâs not bad (still feels a bit overdone).
.)
