I doubt that caucasian skin is that yellowish - so the sample just give a comparison to lightroom color rendition…
4 Likes
It is in the sun but I was thinking the same thing wrt how yellow it looked… If there was sun DT filmic edit didn’t grab that either…
I am looking on a phone so maybe I’m skewed a bit
As others have mentioned the yellow is obviously from the scene. The salmon is not.
1 Like
Yes, the Photo was taken at sunset. Sorry but I couldn’t post the complete picture.
It’s more of an illustration, what I was struggling with.
My actual point is, the results of sigmoid are easier obtained and more in line with my expectations.
I concur all the points that nosle stated in his post 124 (edit: post no.).
Just wanted to share my experience, because I’m really happy with the new sigmoid module.
2 Likes
BTW, talking about fire: have you tried this file? Comparing filmic color science v5/v6 - #72 by age
With sigmoid, you can turn it into something like this:
filmic, set to RGB (display) merging space
and
lightness mode:
You can get slightly more/less yellow by tweaking filmic further.
But the same settings work horribly for blues:
1 Like
Does the extreme luminance slider do anything in this case to reverse or clean up the white??
Filmic doesn’t change middle grey , so for skin to be affected like this, your exposure is horribly wrong . I never never ever have seen skin color change by filmic , except If the sun is shining on it (so it are highlights )
2 Likes
I’d be very interested to play with a raw that shows this.
I might just add that filmic does ‘change’ colour (compared to sigmoid for example) in that it doesn’t increase any saturation as it increases contrast.
Most traditional tools that affect contrast have an effect on saturation at the same time.
Filmic goes to some length to avoid shifting colours… which (to me) in some cases leads to weird looking results, especially if one has increased contrast - until one goes to it’s ‘partner’, color balance rgb, and gives a big whack of chroma and saturation. Then things look good - usually.
This isn’t necessarily (i can never spell that word - thanks autocorrect!) applicable to the comparison image above - I personally agree that the sigmoid version looks much better than filmic and think that this is an example of filmic being mathematically correct, which looks wrong - to some people, like myself…
BUT I do think that this stuff is a matter of taste as well. And it should be.
After all, if we all agreed 100% on what things should look like in a photo, just think how horribly, mind meltingly boring the whole art scene would be! 
No variation, just an accurate representation of reality - what would be the point?
1 Like
I pointed this out to Aurelien way back but yes increasing contrast in tools such as LAB also does require chroma/saturation increase to look natural. But the filmic results are even worse than pure lab contrast though visually related. Colours go more haywire in filmic whereas lab contrast delivers a similar “dead” look.
1 Like
Mmmm… I’ve done very little with LAB so can’t compare, but interesting. I don’t really have enough technical knowledge for some of this discussion 
I like filmic for many things (one point is that I don’t many portraits?), especially with the other options it has, but I really like the new sigmoid module too. Some photos are so much easier to get looking really good.
So from my point of view we have the best of both worlds - filmic and sigmoid 
2 Likes
Reminder: to add value to our discussion, be sure to show qualitative and quantitative examples in as many places as you can. Many of you are already doing that. Thank you.
2 Likes
Which is exactly what is going to happen if you’re shooting someone who is illuminated by bright direct sunlight and is not wearing makeup, as happened in the example image.
That’s what happened here - filmic translated “bright sunlight highlights” into “dude is sunburned”
It’s pretty obvious, just from looking at the image, that the skin tones most negatively affected were ones that were direct-sun.
2 Likes
Yes Yes and Yes…
IMO none of the examples hit the mark but it could be my screen or the close crop… there is a sweet spot in there somewhere…warm sunny tones again IMO don’t look like any of those…
1 Like
Its a good discussion point and likely quite true for tone but with the gamut mapping in v6 some images for example of flowers and things with those reds yellows and oranges are very different looking across the board not just what might be areas outside “middle” gray so in my mind filmic is a curve type, its a tone mapper with a particular implementation in some software… and then the one being discussed here is DT’s version in a module but it is also combined with other elements now including the use of UCS which is marked different and you can show this a little by flipping to rgb CB to JZAZBZ to see the difference from UCS which is less contrasted more diffuse appearance with a nice spread of color which is normally a good option. Combine this with the gamut mapping and color preservation modes that come into play and the situation is not a simple as a filmic curve being applied to the data. Even the rethink between v5 and v6 makes them quite different to manage. Because of this complexity any comments and assertions should in every single case be backed up by a visual example as pointed out by @afre and in a form that will allow others to experiment and confirm or critique any assertions. I think at least 3 types of examples are useful… Can you get something in filmic that sigmoid can’t handle well, and the reverse and finally can the two be very similar and is the difference then merely how much time and effort does it take to get there… but this must be supported in all cases by examples so that any such assertions can be verified and don’t merely reflect opinions, bias and experience with the software…
Case in point your statement above commenting on that particular post that has itself generated a lot of back and forth might easily be avoided, tested, or demonstrated by someone having that file and taking a run at it and saying yes if you do x and y then you will get this but if you do it this way things are fine…
4 Likes
I will have to try the fire image… On your parade image sigmoid is bad in blue also if you try this lightness blend but in normal… basically default sigmoid or default filmic with a change only to v5 science, no for preservation and and safe for the shoulders ( as those now default to hard) ends up being very similar to sigmoid and both are about the same with some blue in the pants but both at a level that might actually be believable …
EV at 2.3… only denoise chroma preset and lens corrections added in addition to tonemappers… using legacy wb to avoid any possible issues with D65 values… wb set to as shot
Swapping sigmoid out for filmic as noted above white pants…
Default filmic… blue pants…
1 Like
To better understand the scene referred workflow, filmic and with these sigmoid a little bit more, I found the video series of Riley Brown starting here:
very enlightening (just ignoring the AgX thing). In particular the second video:
Hm, looking at the message preview, I should maybe also post the last one of this series:
I personally hardly touch filmic but go with the default settings and try to fix things before (same workflow as in blender, fix the “scene”), and if I have to touch it, it’s most of the time the highlight reconstruction or sometimes the contrast.
I’m looking forward to sigmoid as I have the impression that it could ease my workflow a bit in most cases.
Perhaps Aurélien’s original article will also be useful.
2 Likes