Well, this test raw is originated by this discussion about sigmoid and filmic tonal mapping.
There is people like me that says filmic is difficult to fight when you have skies that are over exposed and you want to get them back to a more natural and interesting look, and other tone mapping tool like sigmoid would be interesting.
So here am I trying to provide an example with overexposed sky (and some other part of the photo) that wanted to turn in a more natural look.
Not a good or interesting phot, just for trying, a photo with that kind of problem.
DSCF0175.RAF (24.6 MB)
This file is licensed Creative Commons, By-Attribution, NonCommercial, Share-Alike.
This is what I was getting, trying to revert the overexposed sky and get a bit more interesting sky with a natural look.
I used to employ color balance RGB and masks.
DSCF0175.RAF.xmp (12.1 KB)
No good details in sky.
We were adviced to use color calibration with masks to brighten red and darken red in sky, and tonal equalizer to adjust tones (not needed in this photo, I think).
I have tried and I have to admit it seems to provide very good results, at least in this photo.
DSCF0175_01.RAF.xmp (17.6 KB)
A problem is with the highlight recovery.
This photo has areas of clipped blue and green but not too difficult to reconstruct with other softwares, as it is quite uniform in color in that zones.
I have activated highlight recovery in color recovering mode, in order to recover the color and don’t get it grey.
It seems to do what intended, but creates some artifacts.
It seems a recognized problem and I was said it is being worked out with a new module based on guided 4th order laplacians so not problem, as it will be corrected.
Finally I have tried it with sigmoid.
I have only an old version based in 3.3 of DT and not 3.6, so I could not make the same exact processing.
This is what I get.
It would be great to see others results and comparison with sigmoid and filmic.
And results with RT too, in order to get a reference and see what can be obtained and with how much effort.
DSCF0175_02.RAF.xmp (23.6 KB)
May be there are much more better ways of developing it using filmic or sigmoid.
In particular, filmic seems to provide a way of recovering highlights on itself.
To do that, if I understood it well, you just do not use recovery at all in the base module, so filmic will receive overexposed values over 1.0 and compress them to 0-1 and try to recover them.
But I have tried with several photos to do that and get no good results, as it seems to turn them to a grey/white value with no propagation or surrounding colors.
I have to admit I don’t understand well that reconstruction tab in filmic.
May be @anon41087856 or others can show us a better way of using filmic in cases like this.