New Sigmoid Scene to Display mapping

Thanks for your detailed answer, very interesting.

Yes, JzAzBz

As far as I could observe, filmic shows considerable hue shift if you change the middle tone saturation slider but no shift when changing contrast.

I fiddled around a bit and think the following setup produces a “sigmoid-like” starting point with filmicrgb (dt 3.5) :

In the look-tab, set:

  • contrast to 1.75
  • latitude to 15%
  • shadows / highlights balance to 0%
  • middle tones saturation to 0%

In the options-tab, set:

  • color science to v5
  • preserve chrominance to no
  • contrast in shadows & highlights to safe

Thanks, it looks pretty good, I just reduced the contrast slightly to my taste. The best thing is that you gave me a nudge to get rid of the default mid tones 10% saturation increase. I hadn’t realised it was the cause of many woes in my images, and things look much better without it (IMO), so thanks for that!


As someone who isnt looking for absolute color accuracy, but decent adaptive exposure/shadow/highlight handling, and speed/simplicity of adjustment, Sigmod absolutely wins for me!
(I mainly shoot sporting events, where i need to edit hundreds of photos a day)
I really hope it does make it into 3.6.


The 3.6 release is in a month and the PR is still marked as WIP, so it isn’t looking good for inclusion in 3.6

I have found that the suggestion of @pass712 works pretty well for me on the meantime. Tonality is very close to OOC jpeg, the colours are very slightly different, but to me, look more realistic than the OOC. Pretty pleased with it. Haven’t had a chance to try sigmoid but, for now I don’t want to get hooked on something that might never make it to the final version…

1 Like

Hi @jandren thanks for the hard work!
It would be possible to reintroduce a little amount of hue shift in this case?

The preserve hue option preserve the hue and saturation ( saturation = max rgb - min rgb), it only interpolates the middle rgb value, however it reduces a little the luminance.
This is the reason why very bright orange could be a little off.

Using something like 70%-75% preserve hue and 30% - 25% rgb is visually the same correct hue but it handles better sunset photos.

Overall it looks close to how lightroom/camera raw renders colors, i have opened a request for rawtherapee some time ago


Yes! I have been toying with this idea myself :slight_smile:
Haven’t had the time to test it yet though, just some scribbles in my idea notebook. I had liked to go one step further and make it adjustable as a function of scene luminance. Middle grey could then be 100% corrected while very bright pixels converges to some percentage of crosstalk.

If you know about anyone doing something like that in a paper or similar, please share!

1 Like

This may be unrelated to the approach you want but I had come across this one in the past


Uhm i need more time to ready carefully this paper, i think it describes two steps

  1. restore hue after a per-channel tone rgb mapping , hue and saturation are the same of hsv/hsl color model…
    This is exactly what the preserve hue ( film like in rawtherapee) does.

  2. restore somehow the luminance channel (from the per-channel rgb tonemapped image?)

applsci-09-04658.pdf (3.6 MB)

This one looked interesting as well but the math is beyond me…

I read a good paper on colour preservation and compression last year but accidentally deleted the bookmark and can’t find it now.

The hue discussion reminds me of this comment the darktable 3.0 video series - #16 by sankos.

1 Like

Interesting settings that may be close to a rather bland JPG output. I am rather surprised that some users are trying to emulate the JPG style … I thought that the entire reason that we shot RAW was to have the ability to put a personal stamp on our images.


I noted that when playing around with the image that I think @davidvj shared. It had a lot of chromatic aberrations in the foliage. I wanted to compare darktable to ART or RT as they can use DCP files just to see what was up using that for color management and I think there are 3 options in ART/RT from the DCP profile , tone curve Base table and look table. Disabling the base table in color management dramatically improve the CA. I’m not sure how the overall colors fared but clearly there are tweak and ways that things are potentially altered before it gets to the tone mapper…

I’m not quite sure if that phrasing is a correct assertion. Starting points are just that…a start. The ability to put a personal stamp on sth. is tangential to the question of what is a starting point. This particular setting was derived from the question “can filmic and sigmoid be matched”. The answer to that is: in many cases no. @pass712 found a close match and he shared it. (If I understood his posting as he intended)

1 Like

Likely as well referring to the jpg as a common frame of reference for points of discussion and comparison of the two methods rather than necessarily trying to match it???


@davidvj @PhotoPhysicsGuy I may have been the person who gave the impression that I was trying to emulate the OOC jpg output. I just said that the tonality was similar to the OOC jpg, but that is just a starting point. Something bland with no unpleasant surprises works as an initial setting for me.


I know that this is somewhat away from the original discussion but … I simply ignore this ‘base’ JPG view (whatever that is) and use presets to go directly to my personal style. To me any JPG style is simply a distraction from my vision.
I use a Fujifilm camera so JPG can be any number of the manufacturer’s diverse visions.


I personally think the best starting point for discretionary editing is the linear RGB right after demosaic. I’ve had cases where the exposure and the scene worked out to need nothing beyond that, no tone curve required.


I agree, editing depends on the result you want to achieve and the point where you started. Sometimes both are close together.
I would always prefer a starting point which is close to my result so that I can reduce my time for editing to a minimum. And of course, the way and effort to reach your result also depends on your skills.