MStraeten already ansered your question in the other thread.
As said in the other thread : this is to be expected. See http://darktable.secretlifeofmatt.com/overview/workflow/edit-overview/ to understand the differences between the modes. Or PIXLS.US - Darktable 3:RGB or Lab? Which Modules? Help! for a deeper insight. And there are videos by Aurelien Pierre and Bruce Williams dealing with this aspect.
Is that a guess like the people in the other thread? Can you point me to the documentation or the developer’s post that says this? Thank you!
No, MStraeten, did not answer.
By the way, just a note: If someone doesn’t know the answer then that is okay. Either pass this thread by or clearly say that what you have to say is a guess. Thank you. I would appreciate a link to the documentation that gives the answer or a post from the developer that explicitly tells the answer.
I want to let people know that I do appreciate responses. I just want everyone know that I too can make guesses. That isn’t the problem. I am hoping to find out the answer.
From the manual (git master 3.3.0+1006~ge31529f63) :
auto-apply pixel workflow defaults
With that setting, you can define which default workflow to work on. Two workflows are
• Scene-referred workflow is based on linear modules and will auto-apply filmic and
exposure. This workflow consist of adding, by default, +0,5 EV of exposure (in exposure
module), to adjust the global brightness and mid-grey to a similar value to that used
by most JPEGs produced by SLR cameras. The interest of filmic is to apply a tone
mapping curve at the end of the pipeline, allowing a physically realistic retouching
on RGB values encoded linearly before, but also to manage color consistency through
the tone mapping, in order to preserve hues as much as possible and to avoid gamut
This default +0,5 EV value can be overridden using an auto-apply preset if it is found the
images from a particular camera are consistently too dark.
The scene-referred workflow also enables by default the “compensate camera
exposure” option in the exposure module, to adjust the global brightness appropriately
in cases where the camera’s exposure compensation dial was used to protect highlights
in the image or Expose To The Right (ETTR) to optimally make use of the sensor’s
• Display-referred workflow, used by default, is based on Lab modules and will auto-apply
base curve and the legacy module pipe order (darktable 2.6 and older). It will apply the
non-linear transformation at the beginning of the pipeline, and all editing will then be
done on non-linear RGB.
A third option is available to not use any workflow and so deactivate all automatic
modules. That allow you to manage your pipeline manually.
You have tried all three methods, which one did you prefer?
That might give you your answer for your particular style of photography
No. What gives you that idea?
No. I can’t. But I know that all darktable tries to do with the Lightroom settings import is map the LR tool onto am equivalent darktable tool. All the tools that are mapped are LAB modules, which puts you firmly in the display referred workflow.
So in the end, darktable is open source, and the code is there for all to see.
What I see when I look at the code is that there is no application of base curve or filmic. The code will create an instance of the “tone curve” module if you have made adjustments in Lightroom. I still have a couple of xmp files from Lightroom, but there the tone curve definition is linear, and so darktable doesn’t bother to create any tone curve instance.
So, you probably know that if you look at a flat raw file with no tone curve at all, it will look dark and low contrast. Therefore, it seems clear to me that the intention was for the base-curve to be applied, and then other modules from the lightroom XMP applied on top (like crop and rotate, vignetting, etc.). However, if you use the scene-referred workflow, then the basecurve is substituted with the filmic rgb + exposure modules, and this is equally fine if you set up an exposure auto-preset suitable for your camera. It just comes down to which workflow you prefer to use.
The only area of doubt I have is about this tone curve that could be created. I don’t have any lightroom XMPs which contain this, and I no longer have a copy of lightroom with which to make a test edit. The concern is whether the tone curve coming from lightroom is relative to lightroom’s “base curve equivalent”, or whether it is relative to the original raw. If the latter is the case, and you have base curve or filmic rgb also active, then you will end up with an overly bright overly contrasted image.
I would “guess” that it is actually the former case rather than the latter, but perhaps if you still have lightroom, you can make a test (make some tone curve modification in lightroom, import the image into darktable, and see whether the darktable tone curve module is complementing or is “doubling up” with the base curve). Then you can let us all know if my guess is correct
[edit: in both display-referred and scene-referred, the module order still places tone curve AFTER base curve or filmic, so the different module ordering between the two workflows doesn’t matter in this case)
Does it …I don’t think any edits are carried across maybe rating or something but I don’t believe any LR edits are preserved???
The edits that are carried across are described here:
Whether this is still true for later editions of lightroom, I am not sure. I know at least crop still worked, based on last year’s release of Lightroom Classic. I don’t have an active subscription anymore, so I can’t check further…,
Ya I wasn’t really sure but from most of what I had read people were successful with tags and rating but not much else but mileage may vary and as you say LR keeps changing and so does DT…there are so many nuances in the last year with module changes and order and versions of xmp files in DT that there are bound to be issues…heck DT keeps messing with my xmp files. Over the last two years subsequent builds have erased or altered my xmp files from older edits…
Matt, I have been away for a few days so I just saw your interesting post. Thank you. I still have my LR/PS subscription so later I will do some more experimenting by selecting some more photos, saving the LR XMP, and then importing them into dt. From what I have seen already where I imported the same image with LR XMP into dt 3 times using scene-referred, display-referred, and none they were all quite different. Only one image though so it wasn’t clear which I was supposed to use. Your suggestion from looking at the code that it expects the base curve (display-referred) makes sense.
I have an earlier post where I asked how I could have dt import only the metadata (tags, star ratings, etc.) and not the LR developing steps since dt doesn’t do a good job with them. No one knew a way to do that. The suggestion was to go through all 120k or so images after importing and one by one manually remove the dt development steps that came from LR.
If dt did a good job with the LR development stuff that would be great, but since it doesn’t it would probably be better to have an option to tell dt to only import the metadata.
Importing Lightroom XMP sidecars
That was definitely not the suggestion. You can highlight all the imported image by select the film roll, then use the Discard button from the history module to discard the edit history from all 120k photos at once.
Yes, that was suggested. And as you probably read further down, it didn’t work. Then the next suggestion was to go through one by one and remove steps with the compress history command.
Thank you to everyone for your input.
I thought I would update this thread with a bit more info about bug reports I submitted some time back.
Discard history does not remove Lightroom emulated steps #7594
And a feature request:
Importer for Lightroom XMP sidecars should allow only getting metadata #7595