Isn’t this what we have already with the module groups?
The choice is explicit because filmic is enabled by default. If a user wants to choose otherwise, then they can.
What is a “little something”? We have tooltips.
We’re drifting towards generalizations again, which I don’t think is helping anyone.
Please provide some links to where this is happening.
I’m still also interested in your images and the usage video if possible.
Impressive! But I got a bit lost at eqn (7). It seems you are setting the slope ψ’1 == ψ1 at x1. How does that follow from the constraints? Or have I missed something?
Just a thought, but I completely agree with this once the user has committed to using the software. However, in reality, many people want to try the software out before committing to it. I wonder if this is where we are seeing the dichotomy around modules like filmic. If you are coming from commercial software and interested in trying darktable, you may download it, install it and start playing with it to get a feel for it. If you do this, you are unlikely to just “get” filmic, color calibration, etc. straight away because they are very different to how other raw developers work. I wonder how many people abandon the software at this stage?
Although software companies, camera manufacturers, etc. provide a manual with their products, they also rely on their product being somewhat easy to pick up and learn, or at least familiar in the way it functions (I’m avoiding the “i” word!) and/or they provide a quick start guide. The manual is used for more in-depth reference later.
One of these days when I have more time, I intend to create a quick start guide/video clip series for darktable as my way of contributing a little something back. If something like this could be embedded or links provided from within the software (e.g. on the lighttable “start page” before importing pictures), maybe this will help or force people to read how things work rather than just dragging sliders and getting frustrated.
This is exactly the problem that @paperdigits is always talking about. Companies pay people and even the whole teams to make potential customers aware of their product. This includes something like a quick start guide. And these people don’t lose their free time for that.
Darktable, like all other free software, relies on the voluntary participation of the community. But if everyone here lacks the time to do something themselves to make it happen, then there is no point in pointing out the lack of such introductions.
This is a discussion forum, surely we can still discuss it. Besides, someone here might be inspired to do it after these discussions. I fully intend to do it, but I work full time and my spare time is spent looking after young kids. But one day I know that will change.
Of course we ca discuss this, but this won’t result in a better documentation. grass doesn’t grow faster when you pull it
Excellent!
You’d need to generate organic traffic to your video. If people made it to the manual, found the section they needed… Why not just read it? We do need quick starts, but that is separate stuff from the user manual. We have a request open to put a quick start “right at the beginning” but doing that only encourages people not to read the documentation proper.
If you’re new, you found the docs, and read the first section, Overview, you’d be in pretty decent shape. It isn’t that long nor is it complex.
Thanks for the positive words and advice!
I implemented @Mister_Teatime’s equation in filmic today, both in C and OpenCL:
It works well for its purpose, doesn’t overshoot, but makes the module a bit slower and produces muted contrast near white/black.
I’ll be very interested to try it out once it’s released!
Kudos for the twice-a-year release cadence, by the way. It’d be a shame to have to wait until Christmas for all the fantastic new features in Darktable. By Midsummer, I’d typically break down and use the unstable version, and fight with all the intermediate bugs. It’s much nicer this way,with two stable releases per year.
Common guys, after all the heated debates and many opinionated opinions, I can’t believe nobody will git pull https://github.com/aurelienpierre/darktable.git filmic-safe to test and report before release. It’s your time to actually make a difference, shine !
I will test it as far as it is merged into the developer version. I can hardly wait! 
Take it easy dude… This addition isn’t even 24 hours old yet and was posted late in the evening (for us Europeans). You really expect everybody to jump at it at the first sign of life in 25 days? 
Anyway, just pulled your code and gave it a quick test: Only thing I immediately noticed is the slightly less saturation result between v4 and v5. New one being the one that is less saturated. This is mostly seen in the darker and lighter parts of the image.
Maybe my test images (3 in this case) are not representative and the testing not rigorousness enough, but I do get the idea that it hits green and blue more then the reds.
I don’t think I telling you anything that you don’t already know. Well, maybe the red/green/blue remark is new…
This was just merged, for anyone building master.
I’m pulling the latest right now…
There is no change in there affecting saturation, it’s unrelated to v4 or v5.
Dear God! V5. This is crazy stuff! 
These last couple of days are like Christmas! You guys are outdoing yourselves! Hats off! 
So what do I see when switching from v5 to v4 and back?
Saturation might be the wrong terminology, but something is slightly different and it is mostly visible in the areas that I associate with being the most saturated/colourful. Could this be what you called “muted contrast”? I do believe that, from an editing point of view, contrast and saturation should go hand in hand (one up, the other one down).
The default latitude has changed. Could it be that?
v5 was introduced in this PR and is just a slight improvement of the chroma-preservation algorithm. With v4 the Filmic desaturation/resaturation curve can cause a change of the pixel norm - v5 does an additional normalization to prevent that and thus have the pixels follow the tone curve more closely even when the saturation is applied. It’s a subtle change but hopefully one for the better.
Regarding the new toe and shoulder: at an initial test, the behaviour seems very nice. I feel the latitude parameter now has a closer connection to the shadows and highlights rolloff. Very well done! The muted contrast doesn’t bother me and looks quite pleasant.
