Thank you. I just tried it and it mostly works. It’s a good starting point but have to modify it for my images but I don’t really understand it.
In the wavelet > contrast: why is level 1 to 6 set to 0 and you apply the contrast to the coarser 7,8,9? I would I have done the other way, 0 contrast to the coarser and more contrast to the finer, similar to the detail > contrast by detail levels.
and what about wavelets > residual image > contrast?
Actually I proceeded with just some feeling, using an image that I had exported from a LR trial version as comparison, trying to match the clarity look up from the unprocessed image in RT.
Of course it’s far from good. I just wanted to share this pp3 because I knife like it on some images.
If you come up with something else, please share it with the community!
Thanks, though really hard to compare because images have different size…
A triple view, no changes, Contrast + 100, Clarity + 100 at once would be usefull, as you showed above with the Contrast + 100, Clarity + 100
Well , when using the Contrast slider the highlights is getting lighter , and the shadows is getting darker.
When using the Clarity , the highlights is not getting lighter , only the mid tones is getting more contrast.
Could we have the original histogram as well? We need every combination and histograms of each of the same thing to evaluate the differences.
From my experience with clarity, the algorithm changes once in a while. I think the recent versions incorporate machine learning, so it makes it even harder to understand well.
Using clarity many years ago, it seemed to be a combination of tone mapping, compression and local contrast enhancement. I used it in small amounts; large amounts looked bad. But with machine learning, as you can see from the +100 samples, it looks really good now.
I’ve been testing the clarity feature from “waveletnew”, and in conjunction with the settings I used earlier with classic wavelet and local contrast, I think I get something quite nice!
I guess that indeed, merging the clarity filter into the “dev” wavelet levels would be nice. Instead of hijacking this thread, I’ll probably share some results in another thread here, and in GitHub if it fits.