RawTherapee 5.12 Release Candidate 1

RawTherapee 5.12 is almost ready! To ensure the release goes smoothly, we have prepared a release candidate for all of you to try. If you encounter a bug or some other issue that we are not aware of, please let us know. As always, refer to RawPedia for instructions on how to write detailed bug reports. The preferred place to submit bug reports is on our GitHub page, but for this release candidate, you may report them here if you do not want to create a GitHub account.

Downloads for Windows and Linux are available on GitHub: Release 5.12-rc1 ¡ RawTherapee/RawTherapee ¡ GitHub. You can also use these direct download links.

For a list of what’s new in 5.12, see the following post.

Thanks to all who are testing!

11 Likes

Hello, little thing. I am on Xubuntu 24.04 and use the AppImage of RT 5.12 RC1. When I send a photo to Gimp, Gimp doesn’t show the icons AND is in English. My normal Gimp from the repos is in Dutch and shows the icons as usual. When I use the RT version from GitHub, self-compiled, everything is okay.

Same story for earlier dev versions of RT, like RawTherapee_Beep6581_ghs_release.AppImage or RawTherapee_RawTherapee_captur-noise_release.AppImage

I guess something goes wrong when packaging the AppImage.

I have noticed the same thing on Ubuntu 22.04. I found that if you open GIMP before sending the image from RawTherapee, the icons display normally.

Ah, yes indeed!

1 Like

Everything seems to function correctly on Ubuntu 22.04 using the AppImage apart from the icon problem raised by @paulmatth
I haven’t been following the latest developments in the Selective Editing > Color Appearance (CAM16…) module so apologies if the following has already been discussed:

  1. There are two locations in this module that feature Sigmoid and slope-based tone mappers. The first is in the Source Data Adjustments > Tone Mapping Operators menu and if I understand correctly from the tooltip these act mainly on the highlights (although there are some changes to the overall histogram when invoked). The second is in the CAM16 Image Adjustments section and judging by the changes in the histogram, this targets the complete tonal range. This is also implied in the tooltip but not explicitly stated. If this is indeed correct, then it could be useful to change the label in Source Data Adjustments to Highlight Tone Mapping or something similar to avoid confusion.
  2. The tone-mapping label in CAM16 Image Adjustments is “Sigmoid Q & Slope based” with the default option in the drop-down menu that follows being the second item in the menu i.e. Slope based. It would seem to me to be logical to have Sigmoid Q as the default option but maybe there is a reason for this?

@Wayne_Sutton
Hello Wayne

Thank you for testing :grinning:

Indeed, there were discussions on the subject during the review, 1 or 3 months ago (I don’t remember exactly).

  1. I think we’ve both had some exchanges, albeit old ones, but the module hasn’t fundamentally changed, except for fewer bugs (including RGB channel slope) and a new sigmoid. But there’s nothing to stop us from changing Source Data Adjustments, to Highlight Tone Mapping. Just a note this module also uses gamma/slope which deals with the whole luminance spectrum and not just highlight. Can anyone give their point of view? But I admit that finding a wording that is explicit, especially in this data field, is not an easy exercise.

  2. I think Slope based is superior in rendering to Sigmoid (in Q mode, which I remind you is very different from RGB mode), that’s why I set it as default. I don’t know if there should be a logic between the order of the title and the default choice. In my logic, which I recognize may be curious, it is not always the order of the title (label) that sets the default method, but the overall performance, for example Local contrast & Wavelet has as default method Wavelet. But no problem to change, here it is not a Label but a little bit of GUI. Can anyone give their point of view?

Currently I am not at home, but in Paris with my family, I am returning tomorrow morning (by train) and will be really available on Monday

@Lawrence37
if I make these changes, so Monday, on which branch: release, release-5.12, or dev

Jacques

[quote=“jdc, post:6, topic:49669”]
But there’s nothing to stop us from changing Source Data Adjustments, to Highlight Tone Mapping.
[/quote

Jacques, it’s not the Source Data Adjustments label I was referring to but the Tone Mapping Operators label inside Source Data Adjustments. However if some operators are highlight related and some are global this complicates things. The point is that as things stand there are two Sigmoid tone mappers (one in Source Data Adjustments and one in CAM16 Image Adjustments) and it is not immediately clear what the difference is. Maybe more explicit tool tips would be the answer.
As far as the drop down menu is concerned my point is that the default is the second choice in the menu and you have to go up to find Sigmoid. In most drop down menus you start from the top and work down, but this is nitpicking I guess.

@Wayne_Sutton

Hello Wayne

Just before going to take the train, I just created a Pull Request (in release-5.12) without executables) which includes your suggestions.

  • change Tone Mapping Operators to Highlight Tone Mapping
  • change order Tone mapping Q

Pull Request

:wink:

Jacques

1 Like

I think this name change will create more confusion. The “Tone Mapping Operators” Slope and Sigmoid very much affect the shadows as well as the rest of the image. Imho it makes no sense to pick a name that directly contradicts the effect.

There’s the Q to help explain some of it plus one is in “source data adjustments” and the other in “CAM16 Image adjustments” suggesting where it happens.

1 Like

The AppImage adjusts some GTK settings which are required to run RawTherapee as an AppImage, and these settings carry over to external editors. I haven’t gotten around to fixing the issue (it was reported last year).

1 Like

Providing a clear ‘label,’ a wording, or a tooltip is not a simple thing. Indeed, everyone, depending on their culture, language, and training, gives different meanings to words.

Let’s take words like ‘highlights,’ ‘shadows,’ ‘tone mapping,’ although everyone has a sense of what they might mean, but everyone interprets them differently.

As soon as the designers, either for convenience or because it makes sense to them, calls a method by a certain name, or groups different methods together… understanding, or misunderstanding, inevitably arises.

Let’s take the example of highlights. Everyone understands that these are highlights… but beyond this obvious fact, what lies behind them?
What do the following have in common:

  • highlight reconstruction – color propagation;
  • a highlight slider in ‘exposure’;
  • the EV-based or Gamma-based method in ‘Source Data adjustment’;
  • etc.

Technically and mathematically, none.

Same question if we talk about ‘Tone mapping’… It’s a somewhat catch-all concept that hides enormous differences in mathematical terms and algorithms. Once the differences are understood (not easy), how do we translate them for the user?

  • If, for example, I had called ‘Tone mapping’ (tab exposure) by its mathematical name ‘Cholesky factorization’… what would the reactions have been?

  • If ‘Sigmoid’, which is mathematically a symmetric hyperbolic function (a function that approaches a maximum, or a minimum, without ever reaching it), hadn’t appeared in Darktable about 4 years ago, but would appear for the first time in Rawtherapee in 2024, what would the reactions have been? Who knew about the Sigmoid hyperbolic function? Today, its usage has almost entered common parlance… but what’s behind it?

Another example is ‘Wavelet’. I’m often told that the way Gimp handles Wavelets is simpler than Rawtherapee’s. We’re comparing a carp and a rabbit… The only common points are the name and the notion of decomposition (if we admit that this concept is simple !!). Beyond that, everything is different: Gimp composes decomposition levels with layers, without (or almost without) worrying about signal processing. Rawtherapee does the opposite… Hence the really different approaches and functionalities.

If I had to rank Rawtherapee’s most complex/innovative algorithms, I would list them in alphabetical order (there are others, of course…):

  • Capture sharpening;

  • Generalized hyperbolic stretch (GHS) – just look at the debate surrounding the Chinese translation…what a rich debate (thanks to @syyrmb )

  • Highlight reconstruction – Color propagation;

  • Tone mapping with Choselky factorization;

  • Wavelets;

  • White balance auto – temperature correlation

  • etc.

Developing labels and tooltips is a complex process, especially for these concepts and methods in the broadest sense. At the very least, it takes two people to develop something understandable and coherent. Beyond that, it becomes extremely complex.

But of course, I remain open to any suggestions.

And, excuse my bad english…

Jacques

@Wayne_Sutton @nosle @Lawrence37 and others…

Pull request

I just made some comments within the Pull Request. For your opinion