Local edit, colour appearance, smooth highlights: example

Was testing the new (lacam-16n branch) smooth highlights options and though I’d post some examples. In some files the difference was hard to spot but this particular file made it obvious. This post is just for information to those curious about the new tool.

None

Ev based

Gamma based roll off

Note that the file is clipped. This was actually useful for seeing the results imho. I experimented with the gamma/slope in source adjustments to get the overall contrast (no tone curve etc) that’s why those parameters are set quite strong.

In this case I prefer the gamma based option.

5 Likes

If you wish I can remove the “Ev based” choice. It’s 95% a GUI problem… No problems for me. :wink:

Jacques

In my experience there will probably be images that work better with ev based and others that work better with gamma based. This is just one example where gamma worked better. Unless you have some insight that suggest gamma based is better across the board that is.

A different thing is that when “basic” is active there are two gamma base options that appear to give the exact same results? Basic hides the slider for the slope option making the two gamma options identical?

@nosle

Yes, I know… but it’s easier to evaluate like this, at least for the test period… This allows you to see:

  • a) the smoothness of the highlights (lights…);
  • b) “Tone mapping”, because that is one, and I think very effective.

Here too it’s only 90% GUI, to remove.

And another question. I hid the slider in “Basic” mode because I am convinced that all “Tone mapping” tools are more or less destructive and that very often there is no need for them. Users use it because “it’s fashionable”. Of course easy to make it visible…

Thank you

Jacques

It makes sense to hide the slider in Basic. I’ve always used Standard or Advanced but with the recent changes I’m finding Basic does the job, which is excellent.

I’m not so sure about that. I think users (well this one at least) will generally tend to use whatever gives them an acceptable result in the least amount of time.

Being curious, is there some sort of shared archive of test images you are using?

This is just a recent shot of mine. I tend to test using my own images and use cases.

While waiting for a definitive choice, I propose for the moment, to not distort the “challenge” too much, to leave the GUI like this.

It will also be necessary to (re)see the Labels - which are very personal designations, the tooltip, and of course what do we keep.

I just pushed a little change:

  • Reduces the effect of Ev mode
  • Increases the possible value of “Slope” a little

Executables in progress.
https://github.com/Beep6581/RawTherapee/releases/tag/pre-dev-github-actions

The 2 Windows builds failed in ‘Bundle dependencies’, I don’t know why, nothing changed in my code. Github says : “cp: cannot stat ‘cursors/sb_v_double_arrow.cur’: No such file or directory”…

Yesterday, no problem… Sometimes with ‘Predev builds’, all is OK, sometimes MacOS fail, sometimes Appimages fails, sometimes Windows fail… :smiling_face_with_tear:

Jacques

Would you share this one? Seems to be interesting for some algo tests also for dt

@Jens-Hanno_Schwalm Feel free to download it here:

Note that it may download as a tiff. Just change the extension to dng. It’s a “native” sooc raw dng not a Adobe mangled dng.

As indicated in the copyright metadata I share it as:
https://creativecommons.org/publicdomain/zero/1.0/

1 Like

Thanks, i nice sample for highlights indeed. Hanno

@nosle

In addition to the challenge, here are a few possibilities for processing this image. I’ve only focused on the right, big building. With Local Adjustments.

There are 3 spots

  • the first is with Cam16 : Gamma based
  • the second -Color & Light: Merge file - Screen
  • the third - Original Retinex

You may or may not activate all 3 - of course the first…
You can adjust certain settings “a little better” (e.g. Laplacian threshold).

pp3 - hagalund-test01.dng-jdc.pp3

Of course with branch “lacam16n2”

Jacques