Me too. It is highly descriptive and precise. From what I could gather it was invented by christianvincesfoto from Peru who first used that word as a hashtag.
I’m going on a hunch here, but 2025/2026 and (package design) students - chances their (i)phones can take pictures in RAW format are extremely high. On the other hand the amount of “real” cameras in the class will be rather small.
Interesting. As I said, I mostly use it for video, but I have shot RAW with it and not noticed any difference to the DNGs from the native app. (on my Nokia X30)
I’ll have a closer look sometime.
True!
Nice to know! I thought maybe you’d invented it, not having come across it before. 
If I look at the ‘photographers’ around me, I see an ongoing surge away from simple processing to the use of image over-painting with brushes and multiple contrived layers.
It appears simply easier to ‘fix’ a blown sky with a paint brush or an insert and the cost of commercial software appears to be inconsequential.
Photographic processing, as is performed in dt, I feel is rapidly giving way some sort of hybrid craft as camera imagery merges with some version of AI.
And what will that raw even be… I need to go back and read but there is some change to the DNG saved by the Pro version of the new Pixel phone. If I am correct one of the features in the pro version of the new pixel is that the raw it saves is a full size computational raw image used to generate the final image…I know on my now ancient 3a the DNG were useless if you zoomed… They were digitally cropped in the same way you would crop a zoomed jpg if you cropped in … you would end up with a shitty little 800K DNG and a 4 MB jpg… My old Lumia saved raw DNG but you got the full sensor image even if you zoomed for the jpg… I guess as it was not doing any voodoo on the “raw” that was strait forward… At the very least by that time there might still be raw but them there might be this computational variant as well
Exactly, image making – to separate it from photography – is getting more and more about post-production.
To take an analogy from the music world: it still helps a producer to know how a raw instrument or voice sounds and works even if they throw auto-tune at them and then add tons of equalizers, compressors, etc etc at the single tracks, then chop them up, put them together, mix them up before giving it to a mastering service.
And the producer does not have to be a great musician but they have to understand what the pure craft can create.
Hence my suggestion to teach the students about photo-graphy instead of all the stuff that comes later.
Yes. I’m very grateful that we have tools like darktable that give extensive options, without actually altering (in the sense of replacing) any of the original data.
Hi, from my current state of my thoughts what to teach, I’d start with “looking”. Rather than to use cardboard frames as in times when I was student, I’d let them use the smart phone camera. From this to let them select image sections and to let them play with contrast in black white in order to deal with basic design elements (lines, areas, pattern). Thats why tone curves repectively gradation for me is an important thing. That would be the first contact with digital image processing rather than anlogue darkroom work. From there I’d continue with light, most likely still using smart phones. After that I’d introduce system cameras. Most likely I will have the possibility to obtain some seven to ten cameras, most likely something like Sony ILCE 7III. This will be a little dependent how much money I will have available and what will be available at the market. The last big part may be product photography. Indeed, light and product photography I see as the biggest problem, because of missing space for studio conditions for parallel groups. I do not have a good idea so far how to handle this.
AI related stuff I’d include to a certain extent. It cannot be neglected now. We need to take into account that students use them, also in grade related course works. Course work submits will always comprise all steps of work, from original picture with exif data to final picture handed in. Most likely I’d allow AI if it is properly referenced, but lets see.
I believe that what comes after pressing the button was always decisive in photography, so I’d not neglect it though I thought about using Fuji cameras with their Fujifilm x raw studio in camera processing. But that would bind them way to strong to a certain manufacturer most likely much more as when using Sony. Even if from a teaching point of view I could get out of saying to much about raw development…
I’d also regard “planning” as important, but most likely this will be under exposed…
Hi,
I could not let it be and spent some further 4h…
Most time today on rawtherapee. My feeling for now is that I stick with it and maybe teach it after having used it as raw file opener for gimp only.
I did not know before that rawtherapee automatically creates a tonecurve so that the result looks close to the jpg included in RAW file.
See third break here: https://rawpedia.rawtherapee.com/Getting_Started#Edit_your_first_image
That of course makes it very easy and quick to use. No time waste to obtain the jpgs look and feel, which I saw in the camera and on which I did expose (at least one shot even if I make a second and third shot for “expose to the right”).
And afterwards I can change the tone curve or lay on top a second tone curve or just can work with the typical contrast slider which does the same as using a symmetrical second tone curve:
https://rawpedia.rawtherapee.com/Exposure#Tone_Curves
When playing with it I feel that I have less problems to what I want to achieve. If I e.g. face clipping then I simply adjust the tone curve to this:
And I obtain this histogram:
respectively
While in Filmic V7 I obtain:
So somehow rawtherapee handles the color issues for high brightness in standard setting much nicer than at least I can obtain by filmic, while I am free to handle the tone curve as I want.
The tone curves are so important for me since they allow basically the same approach for black/white and color, while giving the feeling of gradation from black/white times.
Best regards,
Andreas
There are about 6 models of that curve and they can look very different. Also the auto curve works fine many times and many times not… for consistent jpg like look you would be better to use Adobe DCP profiles that you can access in the color setting in RT
If you wanted to use darktable, I would suggest using sigmoid instead of filmic - many people find it much easier to handle colours with sigmoid. It makes it pretty easy to get near to a jpeg look too.
It’s available as a workflow option in preferences so easy to use too.
I’m not sure the two use the same information for the histogram … you would need to confirm this… DT by default uses rec2020… ie the working profile. I am not sure if RT uses that or the output profile .
sigmoid is even just a hardcoded curve like filmic - just less possible settings.
It’s just easier because some thinks, the sliders in filmic can be used to do tonal corrections - and it’s obvious that you can’t even think about this in sigmoid.
Both tonemappers have their strength an weaknesses e.g. filmic gives less contrasty skin tones, while sigmoid gives more color contrast when used with default settings.
The main topic is: don‘t use tonemappers to do the job you do with curves in photoshop, gimp, …
There are different tools for that job in darktable.
To deal with different color issues in shadows, midtones and highlights the 4ways tab in colorbalancergb or several instances of color calibration for more complex situations are better tools.
But if the main idea is to use curves for most use cases, then darktable isn‘t the proper tool. But then don’t call the course photo editing but applying curves
Hi, thanks for all of your feedback!
Martin, I agree that there are often more sophisticated tools than the curves, also if it is for contrast and tone mapping. But whenever possible it is not bad to have curves, like e.g. capture one has for Luma masks which at least for me makes it much simpler to understand what happens.
Also for teaching I believe that things need to be as transparent as possible and not too complicated. And there curves in general are very handy, while quite powerful.
To be honest I do not understand why the darktable development team does not like to support those curves including to set them. Even if it may not be the best tool for issues like contrast and range compression it is quite powerful and understandable what happens especially in black/white.
Best regards
Andreas
Of course you can use rgb curve for almost everything - nd if you want to deal with Lab curves -even those are possible. But you need to take care how to press the up to14 bit raw input into 8 bit output. There’s also a customisable curve for that: simply use basecurve as a tone mapper.
But to be honest - then you really don’t need darktable. There are plenty of curve related apps around …
As Martin notes the curves are there but UI for example with the tone curve module assumes 0-1
mon to max and gray 50% working in LAB… So this needs to be considered when you use one of the none scene referred modules.
Filmic and Sigmoid are in the software for this purpose and to allow DT to manage HDR coming from the camera images to the display…
Hi Martin and Todd,
I believe that I have fairly understood the actual problem.
Somehow the tones of the RAW need to be mapped in order to get a “normal looking” picture and this is a very basic function most likely always and ever made by tone curves may they be visible or unvisible, fixed or changeable.
In doing so one may split it up into up to three process steps:
A maybe optional first step is to somehow stretch or shrink the original EV into the “useable zone” (typically be referred 0…255 in C1) . Like in Capture One as shown in this Video:
https://www.youtube.com/watch?v=H5I-TsINP_c
In general this step could be also done by a appropiately shaped tone curve (with focus where it begins and ends).
The next is to make the flat tonality of the RAW looking more natural respectively like we have seen in the camera. That we might call the filmic aspect. In C1 there is an automatic mode which can be switched of similar to rawtherapee.
The next is to control contrast. That is the typical s-shaped curve.
The question is how it is done by whatever tool; for teaching I need to be able to explain it so that I feel confident in teaching.
Explanation for me is ok for filmic. But using it is strange. At least I can set it so that switching it on and off does not change the picture.
For sigmoid I do not understand what it does. Using it with standard values except contrast set to 1, switching it on generally does somehow “eat” some EV: the picture gets darker. switching exposure on to +0.7 picture regains its brightness and histogram moves ro the right but only a little, not the full amount. So that does not make me confident and I have absolutely no idea what I shoul tell students on this. Some magic internal behavior?
If it would be clear what sigmoid does and if it would be possible to obtain also a linear curve in sigmoid, then sigmoid plus optional underlying RGB curves for manual camera mapping and or contrast may be what I’d need.
At least so long for me Darktable is in development and in a status that I do not use it for teaching. But there is still some time until first time teaching!
Best regards
Andreas
In my understanding the main point with scene referred processing in dt is to the contrary that all what follows hereafter in your posting, is performed in dt before the first step you describe as "stretch or shrink the original EV into the “useable zone” – and that this is the main function of sigmoid/filmic to ensure that whatever processing we do, it should finally be shoehorned into the “usable zone”.
the image from Todd is also valid for sigmoid - there’s just a different curve applied and the handling of colors during that transfer is different.
Hi, my point is that I do not understand how and why the histogram behaves as it behaves. In my trials it did not make sense for me what I saw with respect to histogram.
I’ve spent now some time on rawtherapee - going through the local adjustement tutorial. There I came across filmic and sigmoid - here under “Log Encoding”:
https://rawpedia.rawtherapee.com/Local_Adjustments#HDR_to_SDR:_A_First_Approach_(Log_Encoding_-_CAM16_-_JzCzHz_-_Sigmoid)
I did not look at this in detail, yet, since there is a lot other powerful stuff in local adjustments. Even masks which can be influenced by tone curves 
.
