Seeking a bit of workflow clarity

I have just started a week or two on DT, aiming to move from Capture One because I am moving everything to Linux given that Ms licenses will expire forcing me into completely unneeded purchases of new hardware. I have really liked the results in Capture One and how quickly I could get to a good image though.

In DT the flow is often confusing to me, maybe because of the scene referred tools mixed in with the usual display referred tools.

I cannot get used to leaving cropping to late in the process… Feels very weird to me… And I can not position the tool earlier in the flow either… Or can I?

Second, for scene referred I read a recommendation to use flmic white and black point first… And only then get going with the rest… But according to DT workflow, the tool should then sit at the bottom… And it is at the top in my quick access.

The other struggle is between using the tone curve tool and the tone equalizer for example for shadows and highlights. With TE I just cannot seem to get good results… And I also read different opinions on this forum.
Is there a secret sauce to using TE which supposedly is better than a curve tool?

I would appreciate a view on which flow and tools are best used for an effective result that could match the outcomes I get in Capture One (since I have Sony cameras, CO does a much better job than LR on Sony raw in my humble opinion… Not sure yet on DT since part of the result is still me not knowing DT well I think).

You can crop early. Editing history order and pipeline (internal execution order) are independent.

The rest requires a longer discussion, and I cannot do that now. I’m sure the others will help.

2 Likes

I would say that the TC is more classic tool and works in LAB colour space while TE is more modern and advanced tool that works with internal grey mask. There are good videos on YT. i.e https://youtu.be/FqA1Mk_ulVk?si=ZQmrsJSPTfdSAepT

Recommended reading to get started:
https://darktable-org.github.io/dtdocs/en/overview/workflow/process/#image-processing-in-3-modules

https://darktable-org.github.io/dtdocs/en/darkroom/pixelpipe/the-pixelpipe-and-module-order/#module-order-and-workflows

https://darktable-org.github.io/dtdocs/en/darkroom/pixelpipe/history-stack/

3 Likes

It is important to remember that, as @Kofa said, workflow/history order and pipeline/application order are independant.
The order in the history stack is the order in which modules were activated or edited,
the order in the module list (right side pane) is the application order (not implying modules should be used in that order!).

E.g. filmic sits at the top of the module list because it is one of the last modules to be applied in the pixel pipe (the last scene referred module, as it maps the image to display-referred).
But it’s one of the earliest in terms of workflow (just after exposure and color calibration).

Also, some modules may seem to ignore crops while you are working with them (retouch, rotate and perspective). That is because they work best on the full image. They are also applied before crop…

As for workflow: based on recomendations from when the scene-referred workflow was introduced, I like to start with setting the exposure to where I want it (as that determines a fixed point for later operations). Then usually “color calibration” adjustment and “filmic” (black and white reference, contrast, etc.). Then “color balance” to get chroma/saturation adjusted. From there, it depends on the image.

Note that it may be useful to delay “profiled denoising” and “diffuse or sharpen”, as those can be calculation-intensive (esp. DorS)


“Tone equaliser” is a fairly complex module to use. One advantage is the way it can protect local contrast. Another that is works in scene-referred. That means it can safely increase the dynamic range of the image. “Tone curve” maps a 0…1 input to a 0…1 output, so you cannot change the total dynamic range.

The struggle for most is the exposure mask. What this does is set the adjustments for 9 regions in your image’s tonal range. This is similar to the zone system, except that the zones are adjustable! Which tonal region is adjusted is shown by the exposure mask. Other than that, the mask has no influence on your image: if you only want to change a small part of the tonal range, you can have your zone of interest show up as light or dark gray in the exposure mask, with no change in the final result (avoid black or white…).

Some want the exposure mask to perfectly cover the full dynamic range of the image. Personally, I rarely care about that, as in most cases I only want to adjust part of the tonal range. As long as the exposure mask covers that part, I’m fine. I don’t care if the mask seems burned out in the highlight when I want to adjust the shadows, or if the shadows are deep black when I want to deal with highlights… (note: that’s still the exposure mask, not the image, not the blend mask).

And if I have to adjust both shadows and highlights, I can use multiple copies of the tone equaliser, to work on different parts of the image (e.g. one copy for highlights, one for shadows). That has the advantage that the masking doesn’t have to be very precise (that’s the exposure mask specific to this module, not the blend masks you find in almost all modules!).


Not knowing capture one, I have no idea what kind of rendering that gives. As a Sony user, I do know I usually have to add 1EV exposure to compensate for Sony’s idea of a good exposure (I could do that in camera, but then I risk even more blown out skies :wink: ).

3 Likes

For tone equalzer, I often use a very heavily blurred mask. The strength of that module is to manipulate regions, not individual pixels, which is very important when you compress the dynamic range, in order to avoid losing local contrast.

For a recent example, you can check out this post from me (note that it uses the in-development agx tone mapper, so it won’t work perfectly on the released version of damage; simply enable filmic rgb or sigmoid, instead).

2 Likes

I had indeed read through the links you sent as well.
I kind of forgot the exposure first then filmic white/black… but filmic is supposedly the last step of scene-referred before going into the compressed dynamic range modules… yet tone equalizer is lower in the stack but is not really talked about until after the use of filmic, which seems odd to me at least. But I guess I am confusing where modules apply and where they are in the history stack… and apparently that is two wholly different things.

I intend to do most of the basic adjustments via the quick panel, but also noticed that the TE doesn’t work with the mouse scroll via quick access… it requires pulling the graph. On top, a big part of my struggle I now see is the fact that I have pictures where the tone curve histogram is actually quite narrow… which makes adjusting very difficult because you pull too many tone regions in one go… hence the need to widen the histogram first. Seems convoluted instead of the module auto-applying an expansion of that histogram (not sure if that is possible).

The TE module seems indeed quite a lot more complex than what my normal shadows/highlights correction would entail… and that of course reduces speed of work. It’s almost as if I want a basic but effective functionality on the quick access and then when required can dive deeper as needed or for a pic where I want to drive to perfection.
But as I undestand it from the comments, the tone curve tool won’t work on the full dynamic range… I made a few presets in it (typically much smaller adjustments than the presets that are in there), which would allow a quick pass-through…
The presets on the tone equalizer tool however depend a lot on the tone histogram it seems and therefore do not work well at all for me.

I didn’t post any links…

That’s indeed again the difference between pipeline order and workflow.

The order in which the modules are applied to the image in the pixel pipe is based on technical considerations, and usually the optimal order for applying the modules (not always!). But that order is not relevant for the “logical” editing sequence (usually from global to more detailed).
So while being aware of the pipeline order is useful, it’s not all that relevant for most of the editing (in basic cases). Unfortunately, the order in the right sidebar cannot be changed to a “workflow order”, as those tabs are now used to allow modifying the pipeline order (and not just show it)

For that, there is a “mask contrast compensation” which changes the range covered by the exposure mask (for some of the options for “preserve details”). Then again, if the tonal range in your image is quite narrow, why would you want or need a “shadows and highlights” correction?

As it is placed in the pipeline after filmic (or sigmoid), its input is already restricted to the range 0…1. So it does work on the full dynamic range, but you cannot increase that dynamic range (which “tone equaliser” can do)

1 Like

It is unclear what you are asking for here. What specific outcome? Eg you want to fix WB, do something more subtle to colors, compress the tonal range, etc?

Generally, it is much better to start from what you want to achive than how you want to achiveve it.

That said, this is my basic workflow:

  1. Use sigmoid as the tone mapper. The defaults give you a reasonable mapping straight out of the box and are easy to tweak. Sure, try filmic rgb later on, but for now sigmoid is fine.

  2. always enable lens correction, and enable raw chromatic aberration and denoise: profiled if CA or noise makes it necessary.

  3. crop and rotate to taste

  4. check white balance and fix if necessary in color calibration

  5. optionally, apply local contrast, haze removal (I find them much faster than diffuse & sharpen, and equally good), maybe apply a mild saturation boost in color balance rgb but these days I prefer the natural look

Some images may require more, eg tone compression with tone equalizer, simulating a GND filter with an exposure instance, tweaking colors with color equalizer or color calibration. But those are just refinements.

Sorry, this is long, but here are some thoughts that might help on just the workflow aspect (I’ll leave the more technical aspects of what you are trying to accomplish to others).

Keep in mind that one of the best things about DT is the thing that many newcomers find the most frustrating: there are many different modules and parameters that will accomplish the same editing task’s visual appearance results on your image. I’ve been using DT for about 7 years and have printed many of my images on large art paper and it looks great, but someone who is more proficient at DT may look at my edits and wonder what the hell I was doing—if you are getting the results you want on screen/in print, you’re doing it “correctly”.

In the darkroom there are two main areas where you see modules: the left panel’s “history” and pretty much the entire right panel. The order in which modules appear in either of those two spots has almost nothing to do with the order in which you choose to work.

The “history” section in the left panel shows you the order in which you have worked with the most recent edit appearing at the top. In that section you may have many modules: automatically applied modules that DT applies, module presets you had DT apply on import, duplicates of modules, unused modules (they are grayed out), etc.

The right panel is where you open modules and do the editing work. The order in which the modules appear in that panel typically has nothing to do with the order in which you choose to open a module and do the work.

At the top of the panel, above the modules, are icons that have predefined sets of modules. On the far right of those icons is a pancake menu which allows you to change which presets of modules you are viewing and it gives you the option to define your own presets. (I have set up a preset for myself with the modules I most frequently use.)

It is possible to “change the order” of the modules seen in the right panel by simply dragging them on the screen to the order you would like. This is very tempting to do because for many people seeing the linear order of the modules on the screen, in the same order (usually top to bottom) as they work,. is visually/intuitively appealing. However, that also changes the order of the processing pipeline and it strongly urged that until you figure out the basics, just leave the order of the modules alone for now, and live with the fact that the linear order you see the modules arranged in in the right panel is not necessarily the way you will work through them.

That being said, choose your own workflow. As an example, the following is typically my workflow.

[based on kofa’s excellent observation, I deleted the paragraph kofa quoted about white balance and ‘color calibration’ module as it’s not really germane to the workflow anyhow]

My first step is to make sure the image is level, so I use the right mouse button and drag a horizontal across the image on an object I want level (or vertical line if there is a better vertical line on the image for plumbing). Sometimes I need to change the keystoning and use the rotate and perspective module for that.

Next, I crop (I set this to keyboard shortcut ‘c’).

Then I retouch (which momentarily disables my cropping, but when the retouch module is closed, the cropping reapplies to what I had).

I make sure the clipping indicator in enabled (already mapped as shortcut key ‘o’).

If the shot looks either over/under exposed based on the clipping indicator or visually, my next step is to have DT automatically adjust the ‘exposure’ (I usually do this step regardless of what my eyes tell me). In the ‘exposure’ module, the little eyedropper to the far right, when pressed, reads the bulk of the picture (you can see how much of the picture is used for the adjustment when you press the eyedropper as a box appears over the image with handles in each corner which can be used to adjust the area being read). I have created a keyboard shortcut for this: ‘e’. I don’t worry too much about the result I visually see but rather accept it for now as the next step usually compensates for any over/under exposed areas.

Next I open the ‘filmic’ module and in the ‘scene’ tab (the one that opens by default), on the far right are three eyedroppers: press on the bottom one which automatically adjusts the image like the ‘exposure’ module’s eyedropper. I also have a shortcut key for this: ‘f’ (‘Filmic’ is a fiddly module and as opposed to ‘sigmoid’ and occasionally needs to be adjusted throughout my workflow, so I usually hit the ‘f’ key throughout to see if the results are better, or get into it and readjust manually.)

So to quickly apply those two modules I simply press ‘e’ then ‘f’ and my picture is usually ready for editing.

If I don’t like the results after those two modules, I’ll tinker with the parameters in one/both of those modules.

I rarely use the “tone curve” module and instead I most rely on the ‘color balance RGB’ to do almost all of my shadow/highlight work and basic color correction. The “master” tab at the top of that module is a good starting point, but the “4 ways” and “mask” tabs are very important in my workflow. I often dial down the contrast in this module to bring out more subtle details then fix the contrast with the “4 ways” and “mask” tabs.

For additional contrast work I often use the ‘diffuse or sharpen’ “dehaze” preset and adjust. As well as the ‘contrast equalizer’ “clarity” preset, then dial the “mix” parameter from +1 down to somewhere between +.3 - +.5. I typically only use the ‘tone equalizer’ for fine tuning. Occasionally I’ll use the ‘local contrast’ module.

Lastly, I denoise using whatever modules work best (sometimes its ‘astrophoto denoise’, sometimes ‘diffuse or sharpen’ sometimes ‘denoise (profiled)’, sometimes all of these, sometimes several instances of one or more of these).

Granted, there’s plenty more to do on any given image, and, as I said at the start, there are many ways to accomplish what you need to accomplish. Luckily, there are a lot of tutorials on the modules (YouTube is your friend), so figuring out individual modules shouldn’t be too hard once you get the hang of things and have developed a workflow.

1 Like

That has nothing to do with the camera using the correct white balance or not. It has to do with how white balance is applied, and is a singular complex topic. Let’s not confuse newcomers.

1 Like

Thanks for the clarification.

1 Like

From what I see in the comments, everyone seems to figure out their own way through, from the “basis” of the early guidance.

The pixelpipe sequence sitting on the right instead of “user sequence” is unfortunately utterly counterintuitive to me… I tend to be of the opinion that in the end an efficient workflow is what the user should experience and the software is intended to process that in the background rather than pushing it in the foreground… but I guess this is what comes with DT.
I do wonder what the exact user benefit is of having the pixelpipe shown in the right instead of a customizable user workflow. Keyboard shortcuts may help a bit as a workaround indeed.

On other things…
For the moment I prefer filmic to sigmoid actually as you.

The TE is a question mark, but from a YT video I watched earlier, the color balance RGB does a lot of the adjustments I would lean into indeed (including for example the perceptual brilliance sliders which seem to finetune my result from TE in a way I like in terms of result).
Not sure I could do without TE or a tone curve tool as such… but I will have to experience that with going through multiple photos in different contexts.

I was very overwhelmed after week 1, but I am sort of coming to at least a beginning of a workflow it seems.

1 Like

Give it time… It took me about 100 hours of editing to get comfortable with just the basics in Darktable. Now I find it a really versatile tool that is continuously extended with useful functionality. It is a worthwhile investment.

Even though I’ve been using dt for so many years, i sometimes still have trouble locating specific modules because they are in the order they are in (even alphabetical would be helpful for me), so shortcuts and the ‘search modules…’ at the top of the right panel help me a lot. Best of luck in finding your way…it does get easier.

@gevk , you are perhaps getting overloaded with ideas and advice, so here’s a bit more!..
These are things I find invaluable.

  1. Check out masks (in the manual) in order to make local changes. This is one of DT’s best features IMHO. Then see how a mask in a “dummy” exposure module can be used multiple times later in the pipe (“raster mask”). I find this a very intuitive way to adjust the image and in particular prefer it to Tone Equalizer.
  2. Check out Styles which allow you to set up your favourite modules with your favourite initial settings in one quick action.

If you post up a sample raw plus a C1 rendition that you are very happy with, ppl will probably try to reproduce it in DT and looking at their recipe / .XMP will probably be useful.

1 Like

You can change the order in which modules are applied in that right panel. Changing that panel to show a “workflow” order has a few issues:

  • the devs would have to find another way to allow changes in the pixelpipe order (and such changes are essential in certain cases);
  • which workflow order would you set?

What can make use a lot easier is customising that right panel, to have it show only the modules you want to use, in groups that make sense to you. In case you need a module that you did not keep in your layout, clicking on the group header shows a full list of available modules.

One caution here: don’t apply too much perceptual brilliance to the highlights (a safe limit is about 20%), the algorithm is such that it can make highlights “go nuclear”, i.e. reach ridiculously high values. That can give issues with other modules, esp. highlight correction in filmic.

1 Like

I appreciate that… But it seems the large majority of use cases now needs to work with a workflow that is helpful for border cases while it is of no benefit for the majority of cases. Then again… Until devs decide something else, I will live with it, like with other software…

Have been experimenting with masks but did not realize masks are re-usable… So a mask on a dummy non-adjust module can be picked up elsewhere. I will go exploring…

I use at least one module in a custom position on every single edit.

While it may seem like an annoyance now, once you need to move a module position it is invaluable.