Brian I think this is worth adding back in your sunflower thread and then also maybe do the same thing with one or two more images and share the results. I have sometimes found that I have tested and worked through something that was ideal based on the image I was using to test but it didn’t always transfer well to other images… I am not in any way saying this is the case with your work but it might be nice to see a few more examples or later when you think its mature enough share your tweak so others can also give it a go…just a thought
I think some of what people see in v6 is interaction with highlights but also from what you can see when you flip the modes and as shown by Brian the clipping introduced by the gamut check can strongly impact the image. I have found that you need to be a bit more heavy handed in v6 with the white value. I used to trust the auto setting but sometimes now it requires more. Also maxrgb can sometime blur the image a bit so it looks the best overall for saturation out of the gate but its not as sharp…then you go to one of the other modes and they will have clipped areas that you have to address… It may be that I just got used to v5 but I find it easier to make the adjustment I need in the module whereas v6 needs more help from the other modules…perhaps by design as I think the manual even implies this… one thing that is interesting is when you use rgb color balance now the choice of color space can really impact the look…default UCS is good and spreads the color nicely for the most part but for duller images the contrast boost with JxAzBz can some times look better…but all this is a bit off topic
This looks interesting. I am generally satisfied with filmic v6 except with the results that you demonstrate of bright yellow objects turning salmon coloured. Perhaps I am not using it properly and need to RTFM although I thought I tried lots of the suggestions. Filmic v5 with chroma preserve = no generates black areas with saturated colours when used with colour balance rgb (I did report the issue a while back) so is not an option for me.
What version are you on. there are some random issues like this but also for me recently with Sigmoid and local contrast interacting but I think there have also been some fixes potentially??
I am on version 4.0, not updated to 4.01 yet as everything was working fine.
Lots of changes even beyond 4.01… wrt that issue you mentioned…can you save me searching and share it… I recall some issues showed up when rgb cb was set for one colorspace vs the other… the look of the image is for sure much different when you use the default perceptual UCS vs JZAzBz…
Will do, just on phone now. Beautiful autumn here in the UK (between torrential downpours) so going to go out for walk…
Same here and same plan after my coffee…headed into the Dundas Valley in Ontario … Natures artwork from last week…
6 Likes
Here is the image that gives trouble with filmic v5+chroma preserve = no + rgb colour balance
Just try changing from filmic v6 to v5
Borth 57.RW2 (14.7 MB)
Borth 57.RW2.xmp (9.8 KB)
This file is licensed CC BY-NC 4.0.
v6 is fine
v5 gives black areas
Apologies if we are hijacking thread on sigmoid.
Ya to further this we should move it… the CB preset adds saturation to the shadows and global. With the perceptual space and v6 it can handle it…If you simply turn off that preset and do the saturation with filmic when using v5 ie the midtone saturation and or use chroma and not saturation in these dark regions then no black happens… The black is happening in a few instances in DT when modules combine to drive the color out of gamut…
My take is a bit lighter so I don’t have issues in those dark regions
1 Like
Would you mind sharing a patch or a PR to your change? Looks interesting to me as I’m currently also struggling with an image which has predominantly yellow color tones.
Have you tried to edit it with v5 instead of v6 flimic??
Sorry- I’ve been away a few days. So rather than posting more of my own, I’ve been processing some sample raw files from free sources on the web, looking for ones that highlight what I’m talking about. I will post these in another thread soon- I’ve just been busy. In addition I’ll post a src code patch you can use if you are interested.
2 Likes
I would like to say this is by design. I wanted to make a module that is robust and smooth. This has been approached by having the following: any pixel color should converge to white as the exposure* increases towards infinity. Combine that with the observation that the display gamut is a tilted cube with one corner at black and one at white and highlights will soften for highlights. The available gamut decreases as we approach white.
The “fix” for too compressed highlights is to reduce the exposure of the affected area using either a masked exposure module, the tone equalizer, contrast in color balance rgb, etc, before the display transform.
*I think about exposure here as stops over (or under) middle grey.
9 Likes
I can happily report that the sigmoid branch has been merged to master now! Looking forward to seeing what you can do with it, hearing your feedback and fixing any bugs that might surface due to it being used more now! 
And of course spend some time on adding some additional features now that the basics are in!
32 Likes
Looking forward to testing this tonight. Thanks 
1 Like
Hello,
Great, thank you very much and congratulations for your work. 
I will carry out a series of tests and will come back to you with the results obtained
Greetings from Brussels,
Christian
1 Like
I started a new thread now that the module is officially in the master and this has been a long discussion…if someone thinks we should stay on this one feel free to delete or ignore my new one. I just thought that it will get a lot of interest and it might be nice to have targeted discussion based on this latest and “official” rendition… I have no problem if someone want to delete my thread if that doesn’t seem like the way to go for this topic…
2 Likes
PR for the sigmoid dtdocs module reference is up. Do you feel like I covered the basics well enough?
6 Likes
I like it. But it reads a bit like a technical document first, and an artistic description second. Softening it with an anslog-film analogy or something like that in the introduction might make it more approachable.
Also, the very first sentence (“easy to use and robust”) makes it sound like its alternatives are not easy, or not robust. While I don’t necessarily disagree, a more neutral wording might be less provocative.
Speaking of that, I am missing some justification re: filmic. What does sigmoid excel at, and what are its shortcomings? It might be good to explicitly spell that out (very neutrally) in the docs. Probably in both filmic and sigmoid’s docs, eventually.
6 Likes