Well it does not have to be a module, may be you can do it with the current tools.
Just keep the white point without change.
Then let me select the left part of the expansion range (for example -1 EV) and the number of Evs you want to darken that left extreme.
Then map all values in the range to the destination range preserving r/g/b relations (some kind of contrast curve in the range to contast the values may be that helps).
Then you can blend it with original image using parametric masks and all the arsenal to apply it locally to the sky and blend it smoothly.
Of course it needs to work over the original data, before that data is mapped to the color space gamut (since some of the color captured by the sensor might not be visible with high luminosity and perceive as almost white, but just due to the exposure you made, when you look to the sky your pupila closes and luminosity lowers and you see the blue tones).
Most of the times you wonât get out of gamut colors after transform, as the destination values have less value and no problem with saturation. If there remains some OOC value it should be mapped.
I donât know if you can do that with current tools, specially if the color in highlights have been altered in previous step by DT to keep them in gamut (which usually means desaturate them to convert them to mostly white).
I think that is exactly the kind of behaviour we are expermienting when trying to process skies and darkening and contrasting them: too dull skies and too few saturated (in other programs you have the opposite problem, but more easily solved blending in luminosity of desaturating afterwards).
exposure seems to lower the white point, the right extreme of the histogram goes to the left.
And if the original data from the sensoorr iis being manipulated to put it in gamut, you wonât get good results, just dull non saturated colors, not the same as if you had taken the photo with less exposure.
I left 0 EV unmodified â white point does not change.
I left everything below -3 EV unmodified â shadows donât change.
Note that the image has no pure white anywhere; the darkening you see is applied to non-white areas.
Here, since the sky has clipped pixels, reducing exposure creates a magenta cast. Lowering the filmic white point (so it matches the new, slightly darker brightest post) desaturates those highlight regions, though, and you get:
If I have understood @ariznaf correctly (and Iâm not sure if I have), I think he wants something like the Tone Equalizer functionality, but with the ability to expand the controllable EV range just in the highlights for finer control. So, in @kofaâs example, there is about 2EV range to work on, but if that whole TE mask could be shifted and expanded to cover just the highlights region. Then there would be more control points to work with, meaning itâs easier to get contrast between the subtle tonal differences in the highlights, and the white point can be kept in place.
Correct me if Iâm wrong @ariznaf!
But thatâs exactly what the mask exposure and constrast compensation controls are for â you define what range of exposure values in the image you want the 8 control points for your mask to be covering. If you want those control points to cover only the highlights, then simply set up the mask to cover the desired range of EV values.
Do you mind sharing a screenshot? How do you get the module to ignore the darker tones? I thought the mask sliders were designed to work with the whole dynamic range, and I donât even know where to start to get the mask compensation sliders to just isolate a part of the image.
Ok, Iâll dig up an image to shoiw an example. The basic procedure is you set the mask contrast to expand the range of the 8 control points to cover a 2EV range, then you use the mask exposure compensation to centre that 2EV window arouind the highlights. Darker tones will we mapped offto the left, below the maskâs â-8EVâ control point. You want to leave that â-8â control point on zero, and make your adjustments using the other 8 control points (-7 â 0).
There is no mask contrast with the simple tone curve preset; but you can have it with all the others.
With these parameters, the brightest part of the sky reads almost 0 EV:
Additionally, you can add a drawn and / or parametric mask to the whole tone equalizer, so you could exclude the parts of the wall that are almost as bright as the sky.
Unfortunately, I donât think this is what you are looking for, as this is âregion-orientedâ, not âpixel-orientedâ. Maybe the mask can be tweaked better (blurring it so much may have been a bad idea, but I have no more time today). You could also try the averaged modes, but those do not have a contrast enhancement slider, either.
Thanks. That would suggest you have about 5 control points that you could use to just work on the sky. Is that really almost -5EV for the darkest part of the sky? It doesnât look that dark to me. I have been experimenting on my own photos and I canât get more than about 2EV to work with. And thatâs using quite extreme values for the mask comp sliders. Are you doing this by trial and error, or is there a method to getting the mask to do exactly what you want? @Matt_Maguireâs explanation sounds like thereâs a method, but I couldnât work it out. This would be a game changer if I can master it, but it seems finicky and almost a bit of a âhackâ of tone equalizer. Iâm interested in seeing the actual results, because my own experiments have produced nasty halos and an unnatural look.
Yes â sorry, the screenshot was missing. However, Iâm not sure this will help (added a few sentences to the original comment). Will check tomorrow.
Yes, that could be a solution, there is few control in tone eq in the high tones, when you want to expand and contrast it.
And if I understood Aurelien correctly yhere is another problem with it,
Tone eq is quite high in the pipeline and it may be working with values that have been corrected and desaturated, as they have high tone values and you cannot have a high luminosity and too saturated value.
If it has been desaturated, it wonât get saturated when you lower its exposure usinf tone eq or other tool thought for that.
But it WAS saturated indeed in reality, if it has become desaturated is just because you have INTENTIONALLY overexposed it to have an ETTR image and get the maximum dynamic range.
So the tool should work over the original non corrected values, and only after the highlight expansion be checked against gamut and corrected if it needs for been OOG after that.
You understand that saturation is only defined in terms of perception ? Something that couldnât be more saturated in a large gamut space couldnât have been more saturated in reality because reality is assessed through they eyes of the observer, and the large gamut space is the closest digital representation we have. Welcome in the rabbit hole !
Also, tone EQ happens before any gamut mapping, input profile and final CAT. Itâs right after exposure.
