Sharpening: Unsharp Mask or RL Deconvolution?

Software RawTherapee
1

Hello everyone,

i have made some test with an image that is quite sharp in the center, and is really soft on the sides.

I would like to share the results.

I have tried various radius values and various amount values.

Seems that rl works really well if the image is sharp whilst doesn’t improve perceived sharpness if the image is soft, whilst with unsharp mask you can make unsharp images look a bit better (with rl borders remains mushy, e.g. image 026).

With unsharp mask, borders seems to me better with higher amount but also with the same amount and higher radius. Probably borders become better increasing both radius and amount compared to the center (e.g. 003 is imho fine for the center, and 010 is one of the best for corners).

I provide also the original raw if someone wants to try: https://goo.gl/JEILce

Other ideas to obtain an even better perceived sharpness are welcome. I have tried wavelets but without great results.

Thank you :wink: .

2

Thank you for sharing your results! I think you’re correct that there isn’t a “best for all” sharpening method, but rather the method you choose should depend on the image and where you want to gain sharpening.

In my limited experience, wavelets work well when the image is already sharp and less well if the image is blurry or out of focus.

Have you looked at g’mic at all? There are some nice sharpening filters in there!