This is part of my everyday work. And I dare to say the opposite. Processing photos has never been as fast as it is now with darktable.
And in my experience, when documenting events, it is very rare that each picture has completely different lighting conditions. It is rather the case that you always have a whole series of similar photos and only with the change of position and situation do the lighting conditions change again.
If you then process a photo, you can transfer the change (history in darktable or style) very quickly to the other photos in the series and - only if necessary - you can quickly adjust the style of individual photos.
Besides, you can pre-set some styles according to the lighting conditions and your experience and then apply them with just a few mouse clicks.
In this respect, my suggestion remains that you show a couple of examples about which we can talk very specifically.
How do you envision an example when the description is this?
What should this example look like?
A recorded editing session of one hour would be one way of doing this. But to show ‘issues’ his whole interaction with the module would have to be scrutinized. Probably including his comments what he wants to do at which point and time measurements and clicks taken.
Sharing a difficult raw-file is a minute part of what he describes. At least that’s how I read it.
I answer to you “transfer very quickly” opinion with “I’ve seen better solutions to this”, but yes, that is how one works.
It’s just that the overly detailed interface get’s in the way of doing the individual edits in a transparent way. Too many tab-tool-change-clicks - and yes, I experiment a lot with hotkeys etc.
Hm, I’m not gonna pretend to understand all the math discussed above but as a user (let’s say that third group), I find that I mostly nowdays use just 2-3 sliders in filmic. The White/Black relative exposure and middle tone saturation.
But if the image is complex, then yeah, it can be a bit of a challenge because Filmic is not really that smart I guess so I have to fiddle with “options” tab etc.
But again, for most images tho I really don’t have to touch anything at all.
When thinking about the Filmic from real world usage I mostly encounter 2 problems:
users are likely to use it as a creative tool. It’s way to convenient to just pull the “dynamic range scaling” slider down and get your desired look even tho the same is achieved with a tone eq preset.
Reconstruction is just useless. Produces artefact looking result more often than not etc. I haven’t found a single image where it served any purpose. When editing Christmas images, all of the bokeh balls from the lights are either desaturated or non recoverable or something else. I’ve seen someone here sharing their result from RawTherapee which was perfect.
One unrelated side note that is true for Filmic but also for DT in general (open a new discussion if you wish to respond to this so that we don’t hijack the thread.).
DT and Filmic give you a blank canvas, a dull looking desaturated image with no “look” at all. User is expected to decide everything. And then it becomes a problem because that’s not how the cameras work, cameras apply their own look when I’m shooting etc. So in order to actually get a reference and a representation of an image that I took when shooting I must rely on constantly referring to out of camera JPEGs. That’s especially true when doing anyhing where color accuracy is important. It’s very easy to make a human mistake and make red a bit more orange or something else when you have a 0 starting point and are not starting from the look you had in your camera when you took the image.
It is almost like a camera profile is bad or something so I find myself fiddling around with color calibration a lot and too much to nail the colors and the look. The problem is that channel mixer isn’t really something that can be applied from one image to the next easily so every single image will require tweaking.
Edit: that said, I love the new channel mixer so much. If I want to I can just get lost in that single module and really explore the creative possibilities for hours and hours.
Then I encounter another (personal) issue. Too much image duplicates because I make 10 different looks and can’t decide which I like the most But that’s a sweet burden
I would love to see more video feedback of usability issues. I also think, that what you showed is not the problem that @grubernd had. But I absolutely agree, videos describing the problems will help clarify issues with modules.
I think it is a very nice example. If you would screencast me working - pretty much identical.
But have you seen how many mouse movements and tab-change-clicks he had to do just for the initial setup? Something that non-scenic can do with one to four points on a curve - depending on your experience using a curve tool that takes a few seconds. And you can still plus-minus the exposure afterwards for those images that were off.
(Please let’s leave the “but filmic can handle the dynamic range shift much better” argument out of this, because that is one of the many reasons we want to use it, right?)
The quick access panel in the next version of darktable should make some of these workflows a little less painful (because you can line up all of the sliders you want to use on a single panel).
And also there are the dynamic shortcuts where you can scroll while maintaining a key pressed to change a slider without having to change tabs (for example, I change exposure with E + wheel).
I often wonder if some of the feelings around filmic come from attempting to use it on every image and expecting it to do more than it is designed to do. Call it maybe over reliance on filmic. By placing filmic in that role maybe this is where some of the problem might lie. Sometimes the default settings of filmic will absolutely nail and image and it looks so fantastic…great contrast color balance…the results are amazing just turning it on. Then there are images where it gets pretty much the same result but you need to tweak a bit…at least white and black relative. And then some images it will just not come to this type of solution. The highlights will lose all detail or the shadows just can’t come to a good solution. But isn’t this normal. Given the infinite range of data presented to it by thousands of images is this not really the expected result. I think what happens is people expect magic from filmic as the solution to complete 80% of their edit and when despite moving every slider they still don’t get that amazing look then filmic is now problematic. Probably the issue is does darktable as a whole with its currently approach require significantly more work to get to a finished or mostly finished image (technically superior or not) than other products?. If so is this the price of complete technical freedom? If so are people willing to work with that to have what DT offers?. To what extent should this be a focus of the software, ie technical progress versus usability on a broader scale. In the end DT is a cumulative product that also has to support legacy edits while adding new features and technology. I think some efforts are being made to streamline the workflow…the thing that must be remembered is that as a user we are the beneficiaries of all the hard work of the contributors and have been given the privilege and access to the software. What they do with it and where they take it will be guided by their interests and skills and we are free to use it as it suits our needs or make another choice…
The problem here is not filmic or darktable per se, but a phenomenon that applies to Free Software in general - how to educate users to participate in optimization. This is actually a paradox: on one hand, you have the chance to influence the development, but on the other hand, you are not willing to put minimal effort into documenting your concerns, difficulties and suggestions in an adequate way.
If people come to you with “it’s not intuitive” “it’s too complicated” “I can’t work effectively” “I can’t get good results” “there are too many modules, buttons and sliders” but they are not willing to invest time and energy to explain exactly what they are doing with examples, raw files, screenshots, etc. how can you do anything to solve the problems?
How do you get further if you have to ask people again and again to be more concrete with their claims and difficulties?
In my experience, the main reason for this phenomenon is the fact that people do not dare to be concrete because they run the risk of exposing themselves. That they show how little they know, how badly they take pictures, how little competence they have to bring things to the point, that they run the risk of getting into eternal and tough discussions etc. It is much easier to hide your frustration behind generalizations.
This danger to expose yourself is always present although I am convinced that in this forum for the most part is very civilized and understanding with each other.
But if you don’t dare to do that, then these above mentioned questions and problems are just empty phrases with which neither the developers nor the experienced users can do anything.
Good points…for sure. I think we are progressing towards viewing software like phone apps…a tool to get something done fast with very little effort or user input. This and comparing DT to commercial products where there is a huge need to capture a target market and lots of money to training and marketing…Just the pace of change in something like the darktable open source project should be a hint that this is not something that can be used without investing some time. I also echo your comments this is a great community that has always been quick to engage and been very respectful…I am always grateful for the help and information I receive from you and others…
I have a new found interest in streamlined UI and it has nothing to do with features and everything to do with my mouser’s wrist…I need to look at cleaning up how I edit to minimize all that mousing around…or I need to give my wrist a break…maybe try my left hand…now that would be a good exercise for my brain and a terrible exercise in execution…take care…
I’m probably guilty of that but then again I’m still learning how to use darktable as it’s intended to be used.
It would be great help if you can share images which fit the 3 criteria above or possibly point to 3 play RAW discussions. For me it will help set my expectations with filmic and be a great learning experience.
In my short test with around 500k images that I have in my archive. I find the following filmic RGB settings to be much closer to the sooc jpegs
White relative exposure : +3.5
Black relative exposure : -9.0
Preserve chrominance: none
Target back illuminance: 0
Exposure: +1
Camera I have tested with
Canon 5d2, 5dsr,1dxm2
Olympus em5, e-p5, em1,em1m2,em1m2, em1x,em5m2,em5m2,em10m2,em10m2,em10m4,xz-1
Nikon d800
Panasonic g9
Initially I adjusted the latitude to 50% which was even closer to the sooc but then you loose the ability to adjust the contrast without clipping.
After setting all this I create a store preset as default for all my images.
Once set at least everytime.you switch from.light table to darkroom the images won’t suddenly go dark due to the darktable preset that was there.
What I did is most probably wrong like using preserve chrominance of none or trying to make my.initial setting like the sooc. So if you like go it a try maybe I am.wrong but there might be a 0.0000001 % chance you might like there settings
PS: for Olympus camera if you are using low iso as in iso below 200 the exposure compensation should be around + 0.3 instead of 1.
And
If you adjust your white relative exposure above 4 you will loose contrast and desaturation and you have to add contrast and change preserve chrominance to norm to get the color back. That’s what I observe
Possibly this is the cause. Maybe another argument for more than one dedicated scene-to-display mapper.
That strongly depends on what people expect filmic to do. That is one of the reasons why discussions about the intended use and the place within the pipeline (in addition to the overall pipeline design) are very important. Is it a technical output transform? Does the user mainly use it as a preset without ever touching it’s parameters? Are the artistic look decisions at the right place in the pipeline, or should those artistic decisions only come before the module?
Maybe that is connected with how it’s presented? It’s supposedly the bees-knees of control over highlights, hightlight-contrast, clipping, shadows, shadow-contrast AND midtone contrast. Oh AND produce a specific look!
This quite often heard suspicion that some users don’t really know what they’re doing and the devs can’t really trust the users when they describe a problem is unfortunately also a problem. While, trust me, I fully understand where you’re coming from with this argument, it’s also very easy to fall into the trap of “there is no problem…it’s fine” where it actually isn’t. Mixing in the skill-level of the user is a convenience to disregard valid points (that still have to be made thoroughly, I agree on that). Criticism is not dependent on skill-level. If a total noob finds a flaw in how things don’t work in software xyz, you don’t ask for credentials first. Yes, credentials make it easier to talk about the problem, for sure, but the problem description does not become invalid if the one who brings it up is unskilled.
In my opinion it has a technical part that is the white point scaling down, and an artistical part, the contrast amount and shape.
Probably the latter could be called pseudo-technical when the base curve is matched.
This is one of the main issues i’ve observed too with filmic, there are two way to solve this:
scaling the contrast amount so it remains perceptually constant
add a highlights only roll-off tone mapper operator before filmic, here the bt.2390 is the best choice because the knee point (when the roll-off start) is calculated internally, less parameters for the end user.
For now it’s possible to use the highlights slider in the basic adjustments module.
Edit: probably is better to use it inside filmic for the highlights roll-off, then use the log2 shaper for the contrast part however in this way we have the 10000 nits limit
But that’s a sweet burden 
however in this way we have the 10000 nits limit