On the search for a good DT workflow

darktable
rawprocessing

#1

Hi folks,

for my question, it needs a little bit more of explanation.

I used to use both lightroom and Capture one for a couple of years and after some days of learning, I was always quite impressed with the results and was really happy with both software (I switched to C1 because I was pissed by the politics by adobe). I had my workflow with which I was pretty fast, even if i was doing local edits. At all I was quite happy.

Besides that I am an idealist and a huge fan of open source technology and did the switch to linux two years ago. First I had a VM with my preferred software in which I edited my images. But my goal was always to do the complete switch. The main projects I was digging into were darktable(more) and rawtherapee(less).

But even after months of using it I still do not feel 100% home. I do not really get into my editing flow and sometimes I am unhappy with the results (especially in some more extreme lighting conditions). When using my former software I did a whole holiday session in one evening and was happy. With darktable I always go back and forth, do not finish everything in one evening and most of the time come back to the pictures again, because I still think something is missing. Often I am overwhelmed by the choice and do not really know if I am using the right module for the job (maybe this is just a “missing tutoral” issue for me). At all I feel like I do not get the same picture quality like in the commercial software (without hours of fiddling). And if I have a result which pleases me and I then copy the editing to another (in my eyes similar) picture, the result looks rubbish…

I know that opensource software always has its quirks and that neither darktable, nor rawtherapee wants to clone the look and feel and results of any commercial software. I am also willed to adopt a new workflow but I just didn’t find it so far.

Additionaly I didn’t see any dt edited images (especially landscape) which blew me away and I am not sure if this is because of the lack of IQ of dt or just because there aren’t any (semi)-pros using this software.

Do you have any tipps for my on how to adopt my workflow and editing completely to dt? How did you start? Does anybody else have similar feelings? Can anybody describe his workflow? Do you work with a lots of presets?

And don’t get me wrong, in no matter this should be a rant against opensource in general or dt in special. I am absolutley impressed by the development over the last months and year and about all the work which is done by the developers.

I am looking forward for your answers.

Thanks
joe

PS: May I provide a RAW and some of the more capable persons can show me, what they are able to do?


(Mica) #2

Absolutely you may provide raw files. We ask that they be creative commons licensed if you post them, as this protects us legally.

You can also have a look at our play raw threads, where people share images and others post their processing files (XMP for darktable).

After that, you can check out the Weekly Edit videos by @harry_durgin. He does a fantastic job of talking through the edits he makes. Also the videos of RobertHutton who is also great at explaining.

Lastly, you haven’t told us what you’ve tried and why its failed you. Those would be helpful details.


#3

Thanks for your answer. I will pick a file and upload it later (Edit: by-nc-sa should fit?). And thanks for the links I didn’t know @harry_durgin so far. I scrolled through one of his edits (and will have a closer look later) and while watching it I have to say, that I think I am still missing some general editing skills (highpass/lowpass for example). I think this is one of the things where I fail so far. I never had the feeling that I had to “dig this deep” while using LR or C1.


(Mica) #4

Change can be intimidating, but if you’re will in to dig in and learn, I think DT is very powerful and will serve you well.


#5

I think I will not sit on my workstation today, but I will provide a picture tomorrow. In the meantime I have a couple of questions which are in my mind for a long time.
Are there any modules which should be preferred?
Any modules which should be avoided?
Is there a recommended order, in which the modules should be applied?
Is there a good way to apply defaults depending on the picture (Denoise, lens correction, …)?
Which modules do you apply by default?


#6

The modules are in a fix pipeline, so it depends on your workflow only when to apply them, not on the desired result. You can see the order when you click on the “on/off” symbol left of the “star” of the preferred modules. You have to read the list from bottom to top.

The fix order makes perfect sense to most of the modules since there are interdependencies or other reasons why they have exactly their place in the pipeline.

Some operations are more time consuming than others, and, on slower hardware, it might be faster to wait to apply them, especially if their output is not crucial for the overall assessment of the image. What I think of is e.g. denoising. It has little impact on the artistic result but slows down the pipeline a lot.


#7

Regarding the speed, I thought about picking a RX480. The openCL Tests of some people suggest that the impact is huge (I dont know if all modules are affected by the opencl accelleration). Unfortunatley, prices are a bit high at the moment because of the stupid ethereum rush.


#8

When you post your raw, maybe also point us to the looks you want to achieve (Link us to some images you like). You mentioned you used LR and where talking about landscape photography, so I kind of have a guess what you are looking for, but I may be totally wrong. So if you give us some hints (images by pros or whatever) on what you want to achieve, then we can better help.


#9

I learned how to use DT in large parts from Robert Hutton’s videos on youtube. Start with simple ones and go from there.


#10

Seems to be a good idea, first video taught a lot. But I’m still overwhelmed by the number of modules :slight_smile:
And yes I will post some according examples.

I will also create a custom Keyboard mapping to accelerate my workflow

You can see the order when you click on the “on/off” symbol left of the “star” of the preferred modules. You have to read the list from bottom to top.

Viceversa this means, this order can’t be changed like I wish?


#11

It cannot be changed, but in > 99 % of cases (rough estimation, YMMV) this is not an issue. For the majority of modules there is only one useful location or at least not too many possible steps up or down. E.g., all the raw stuff is at the very beginning of the pipeline, it would not make sense at the end.

However, one has too keep it in mind to find a useful workflow. One example: Whenever I watch tutorials for lightroom, people tend to use the white balance for colour grading. While I am doing this in dt as well, it is probably not the best you can do (in both programs). The white balance is early in the pipeline because it deals with the colour information from the bayer array and relates the colour channels to each other. Therefore, every change there will change the output of every other colour grading module later in the pipeline. It would be more straight forward to use the white balance to get a neutral picture and use other colour grading tools (late in the pipeline) to adjust the overall colour look.

For the cases where the fix pipeline is an issue, you could look into @harry_durgin’s workflow videos. The fix pipeline issue is one reason for him to save intermediate tiff files. But it is not the only reason, and I guess in most of his cases one could have found a way to go without the intermediate step. However, I hope there will some day be a functionality in darktable to automate the intermediate .tiff generation @harry_durgin uses¹, but it would be a convenience feature, and there may be other issues that have to be solved first.

¹To make it clear: What I think of is that you have still only one picture representation in lighttable, but you can have several chained pipelines and select on which to work. Darktable would then automatically work only on this particular pipeline and save the result as tiff in a cache or regenerate as necessary, which would serve as input to the subsequent pipelines. Only the last one would result in the final image.


#12

Not in general. But there are modules that I think are more simple for the beginner than others.

  • At the beginning, I found it particularly difficult to get a good denoising with the profiled denoise module, but I know that some people are doing fantastic things with it. However, I was much happier with the denoise nl means for a long time until recently. And I was never able to get something useful out of the equalizer, even knowing what it does mathematically (in principle, not all the details :wink:).
  • Velvia is very often causing a lot of out-of-gamut colours for me, therefore I tend to avoid it.
  • I struggle with the colour lookup table, but I did not have much time to learn it. Still waiting for a tutorial video by @rwh :wink:.
  • There are some modules that are for very special purposes and you would not need them in general. Raw black/white point is one of these, another one is the invert module, which is fantastic when you are dealing with negative film scans but is hardly useful for digital images.

My personal highlight of recent additions is the perspective correction, it is amazing, especially for architecture.


(keith) #13

I was 100% happy with darktable, until I went over to Fuji and it’s xtrans sensor, I could never get that’ crisp’ look.
It resulted in me using RT for demosaicing and using their RL deconvolution sharpening with damping set to zero (then adjusting).
I then export as a 16 bit tiff and do the rest in dt.
Results are impressive.
i don’t do loads of photos.
Harry Durgins and Robert Huttons ae well worth the time watching.
Plus a good video on dt’s perspective control.

Stick with it and you won’t be disappointed.


#14

OK, in the meantime I watched the tutorial video with the senja peninsula (https://www.youtube.com/watch?v=CY3Zgty2rt0). And I must admit, that this is veeerryyy impressive. And I also learned that I have to learn a lot about general picture processing (eg. tonemapping).

ukbanko, so this could also be my problem. I am also a Fuji X user. The description ‘crisp’ or ‘clean’ matches what I am missing in DT. But I always thought that dt is very capable of developing XTrans?

Edit: Additionally I have to say, that I got some good results with RT, but for me, the usabilty isn’t nearly as good as in DT. And converting vise versa is something I absolutley want to avoid. Especially for my day to day photography…

Another edit: Does anybody know the online trainings from Riley Brandt? I just stumbled over them… :slight_smile:
http://lessons.rileybrandt.com/


#15

Yes, they are really good. I did a review back when the first ones came out:


#16

hi, although I didn’t manage it, to upload a good example (sorry for that) I get closer to what I am expecting. Both IQ-wise and workflow-wise. Regarding IQ, DT forces one to think more. In LR it is somehow easy to jst fiddle with some sliders and what you get does fit very often. In DT you have to think about what you want to receive and then go step by step (I know, this should be the normal approach :smile:)
There is one major drawback for me, which I have now identified regarding the workflow. And this is the whole preset/style stuff. Normally, when processing a huge amount of pictures, I do work a lot with presets. I circle through my images one by one, select a preset, adjust it a little bit and go to the next one. So first important question:

  • is There a way to access the styles from Darkroom module (without keyboardshortcuts)

And the second drawback is that (at least I have the feeling) it is much more hard to create styles which are reusable without much more editing afterwards. So second question:

  • How do you work with presets, are they reusable for you? What parameters do you normally set, when using presets?

Thank you very much.


#17

There is an icon on the lower left to access the styles. I use them mainly for denoising.


(Mica) #18

You can apply a bunch of module presets at once, via the menu in the lower left, as @chris has said, or you can apply presets per module, using the hamburger menu n each module.


#19

Hi. I create my presets usually with the individual modules and tend not to use styles on the global image, at least not yet. Specifically where I have to set up multiple parameters on a module, I try to get myself into the habit of setting up presets. In denoising for example, I have one preset to control luma noise and one setup to control color noise.
I tend not to set specific target values but to setup the module ready for fine tuning. With profiled denoise, my preset contains the blend mode and the algorithm used. The specific values I experiment with based on the picture.

I really like Riley Brandt’s tutorials. They got me up and running with a reasonable amount of time invested. They also cover other parts of the workflow (ingesting, a little monitor profiling, etc).
In addition I added some browsing in the really good manual and threw in some other video sources (I watch Weekly Edit myself, but also a couple of other sources).


(Jeff) #20

LR was OK, but I never loved it. It always felt like I was missing something and in turn my photo edits seemed to lack that something as well.

I feel presets are personal tastes of the photographer or editor and there is no, “One for all”. I’ve never used presets in the past (3+ years ago) with LR.

For me dt was love at 1st sight, so powerful and I liked in the beginning you do have to roll up your sleeves and get dirty a bit. Once you dig in and start to understand how things work the payback is edits come easier and your photos and photography will benefit.

I use styles and modules presents of my own design in dt a lot. I’ve gone through the styles on https://dtstyle.net/ in the past to download a few I liked just so I could see their changes in the modules that made up the style.

I only have a half dozen general styles that are very similar because each one is a step above the previous in my constant progression of my skills and tastes. Through experimentation of existing module presets I find setting variations I like and save them. Just combine your favorite module settings to suite your tastes then save that as a style. When I define a style I will save most all active modules to it with the exception of ones like crop, orientation, etc…

It’s not too much unlike LR for some I bet, but my workflow goes a little something like this. I import photos, applying my name and copyright info to all. After import I use lighttable to quickly scan through the images to find ones of interest and likely keepers at 1st glance giving each one at least one star, more for something I might really want to make work and [R] for all the instant rejects. In lighttable hover you mouse cursor over a thumbnail and pressing the [Z] key will popup a larger preview of that image. It’s fairly quick can be very useful when weeding out the good from the bad without having to open each one in darkroom. Filter the images shown to be one star or greater rating. If there is a lot, or many of the same I’ll now go through those now adding more stars if warranted or removing all star ratings if I find I don’t want it. Note I only filter out rejects and no star ratings. I don’t delete anything at this point, not even the rejects. I’ll select all prospective keepers visible at this point in lighttable and apply my current working preferred general style to all. I’ll go though the results for any stand out issues. If none I leave them alone otherwise I’ll tweak them individually from here on a need basis.

It seems when there are any issues it’s from my composition when shooting, it was to dark, to bright, needs a crop that sort of thing. Sometimes it’s just strong mixed lighting I was shooting in natural light like a recent fair we went to late day. Overbright highlights are my main issue when it happens but I have a few presets for the exposure module that employ parametric masks on the grey channel with tweaked exposure setting that only affect the overbright parts of the image which at a fair at night can be any light source to strong. I’m actually editing a ferris wheel image with the issue. It’s not a bullet proof preset, it will need tweaking from time to time, but it does a pretty good for me most the time.

This is an example of what I was talking about earlier the more you dive in the more you’ll learn, learn to deal with everyday issues in photos that can be fixed with some tweaking and experimentation in dt.

If your worried of screwing any of the settings up or losing any pre-existing presets, or your photo library database in dt just back up your “/home/username/.config/darktable” directory. Which I recommend you do from time to time anyway.

If you want to see a large group of darktable user image edit results then visit our dt group on Flickr if you haven’t yet and check out the photo pool. https://www.flickr.com/groups/darktable/

Good luck with dt! I think you’ll find it quite awesome and easier over time.