AP removed the color preservation modes and simplified saturation handling…
This was his description…
"Filmic v7 removes chroma preservation norms. The saturation parameter drives an user-defined mix between the former
max RGB chroma-preserving norm and the “no chroma preservation” mode:
+50 % saturation is equal to the former max RGB chroma-preserving mode,
-50 % saturation is equal to the former no preservation mode (at constant hue and luminance),
0 % is an equal mix of both (mean),
values beyond ±50 % are linear extrapolations.
In any case :
saturation is forced at most the the original value (the algo is forbidden to resaturate),
hue is forced at the original value, gamut mapping is the same as before in Yrg.
The saturation parameter therefore controls only the amount of saturation bleaching,
especially on skin tones and skies."
Anyone building and testing could share their impression and experiments on this thread…
I just downloaded the latest version of Ansel to try filmic V7. I am glad to see AP is still actively developing his modules. I gave filmic 7 a quick run and I must say I feel handcuffed now the choices for preserve chrominance are removed. It may just be something new and I will get use to it, but on first impressions I am not convinced that I like V7 over V5 and V6. But at least in Ansel the user still has the choice to use V6, V5 or earlier versions.
Well it sort of is not removed from the explanation…the slider runs between none and max RGB… positions in between are a blend… You can enter values of 200 and -200 … I wasn’t sure exactly what that would be equivalent to if anything…
I would think they are different math… I think the slider is a blend from no to maxRGB and nothing more…I don’t quite understand in what way going beyond +/- 50% extends the correction… Just going from the description provided by AP and not the code
(Edit: I didn’t do the port, and didn’t mean to imply that I did. I just patched my local copy for a quick experiment when I noticed the commit in Ansel. The actual port was done by Pascal.)
I’ve put the code into darktable (it was rather trivial).
Yes, you can control the strength of the chroma preservation. However, the ‘salmon issue’ is still there (there’s no fundamental change). I think this is quite a superficial change.
Default position (in the center) provides a nice balance between these two different options that, in my experience, provides good results for most cases. The saturation in highlights looks much more natural:
Another advantage is that it’s relatively easy to adjust the saturation of the highlights to match the lighting conditions without having to fiddle with a lot of options.
Nowadays, with filmic v6, I often use the luminanceY or the Euclidean norm (or, in some cases, no chroma preservation). luminanceY seems to create good contrast in the highlights.
That would seem to agree with his comments … “not resaturating” I tried on a similar picture and you could by driving the slider quite hard see that the white clouds would transition from bleached out to having a bit of color tint… along with this was a perceived change in contrast… but it was not a massive effect… I suspect that fits with his original intent for filmic to be a global tone mapper set the base and then use the tools to develop color an contrast as desired…
I have used the “no chroma perservation” option almost exclusively.
If the highlights were on the edge of gamut and the color was bleached out, a subtle readjustment with Tone Equalizer was enough to bring them back into the range of higher saturation.
Everything else is controlled by numerous saturation and contrast options in the color balance rgb module.
I also find this separation between tone mapper, on the one hand, and regulation of contrasts and saturation on the other, very reasonable.
By color balance rgb you have much better control over the distribution of the saturation and contrasts through different brightness levels of the entire dynamic range. And Tone mapper ensures that the gamut limits are not exceeded.
Okay I love that you took the time to write this comment… this is also exactly what I have been doing for some time … now if I could just make it look as good as you do…
This puts a voice of reason into the changes in filmic 7 that I described as handcuffs. But at least we retain the options to use early versions if they suit our established workflows.
For sure it would seem to have simplified the color preservation aspect of v6 but as Kofa points out you can still run into that salmon color as it still does that gamut mapping…
Having all versions is nice as I think people have their preference…
Someone here on the forum got me to take a second look at V5 and converted me to the opinion that we have two very different beasts with V5 and V6. Many photos just work better with V5. So currently I often do a quick comparison of V5, V6 and Sigmoid. Soon I may need to throw V7 into that mix as well. But I am glad AP is still developing and his work finds its way into Darktable. He has created some nice modules.
I don’t think you will find it tremendously different from v6… the key also to comparing v5 with v6 can be the latitude settings… so simply starting in the defaults of v6 and trying v5 is not a true comparison as much of the difference comes from the part of the saturation control that is defined by the latitude setting which in v6 is largely a none contributor so… with v6 and v7 things are pretty similar so you can just flick the switch but with the v5 vs v6 part you do need to introduce the latitude that was zeroed by default with the introduction of v6…
The difference becomes obvious with the versions when you play with the controls. I am not saying one is better or worst, but they can produce very different results when color becomes a challenge. Time will tell with what V7 brings to the table.
The comment was more for the thread and those following than for you… I know that you have experimented a lot… Its just a very key point that opening an image in DT with filmic v6 starts with Latitude set to zero and expanding it does not impact saturation… this is quite different in v5 but someone starting with an image and opening it in v6 and then simply switching the color science to v5 will not see v5 with the intended or previous defaults which are setting that contribute massively to the difference that you will notice… that was my point as this would be easy to over look. I am sure new users have no clue about this evolution and nuance and so figure just toggling there is showing the difference between v5 and v6 ……
