In other languages: Français, Deutsch. Italiano.
For the first time in its history, darktable breaks its one year release cycle by releasing version 3.2 in August of 2020. The unfortunate state of global health has led to a marked increase in contributions and improvements. On top of that, version 3.4 is still scheduled for Christmas 2020. 2020 will therefore be the first year in which the darktable team will have the pleasure to offer you two major versions.
This is a companion discussion topic for the original entry at https://www.darktable.org/2020/08/darktable-3-2/
negadoctor is a game-changer
2 Likes
Although not a native English speaker, I believe this is a typo, right?
Regardless, thanks to all devs and thorough testers for all your work!
After some time using the master branch, I will start using this stable release from now on and finally start using the dam features.
Many reviews, but a miss one indeed. As a comma at the wrong place in the beginning of the article… Sorry for that!
1 Like
First, thanks for another impressive release!
I’m on Windows and wonder about the AVIF support requiring libavif. I couldn’t find a download to libavif.dll and its requirements (the official Github repository for libavif linked to a static .lib but that won’t do here), but discovered that MSYS2 has a mingw-w64-x86_64-libavif package with shared .dll’s!
So I installed the package, looked up the dependencies and ended up with a libavif.dll, libaom.dll, libdav1d.dll and libwinpthread-1.dll. Looked good so far! But I couldn’t for the life of me know where to put these for darktable? I tried placing them in darktable\lib\darktable\plugins\imageio\format (seemed to make sense) but I still got no option for this.
Do I need to compile Darktable myself for this??
Any other guidance for me as a poor Windows user here?
I was hoping for just a few binaries to throw into a folder.
Yes. If you have the packages installed, cmake should find them and enable the feature
Great, thanks! I’ll stop trying here and look into the more involved way when I have time. 
I must say I was a little surprised it didn’t simply come with this in the standard release given that it’s an open codec, but maybe it’s a decision based on format maturity and adoption which I’d understand.
I don’t have the option to install the Lua Script Manager. I built from source and got no complaints about missing anything for lua. I had it installed in the previous version and changed nothing in my luarc. When I run darktable -d lua I get:
1.041367 LUA ERROR : .../.config/darktable/lua/script-manager/script_manager.lua:645: Module script-manager/script_manager is not compatible with API 6.0.0
I’m still new to building from source so I’m not sure what I might have missed.
I love the detailed explanations provided here, thanks so much!
My 2 cents, for what its worth. (Unfortunately I can’t contribute code, but I can at least contribute ideas.) A module currently can be adjusted in 3 different areas: Active Modules, Favourites, and Group. It feels a little messy to me. I think it could be simplified to just Active Modules. (With all the other modules found via search, or via the ‘More Modules’ section underneath. To make them easy to find, they could be grouped by category - much like how an OS start button has a list of applications, grouped by category). The other thing that gets messy is having multiple modules open at once and large scrolling bars. If all the other current tabs (like Favourites and Groups) were removed, they could be replaced by a Properties tab and Masking tab. So the Active Modules tab shows only the power button, module name and icons. Adjustments would be made in the Properties tab, and masking adjustments in the Masking tab. To me, that would be much simpler and neater.
Also, is there a way to click on the module and drag it to a different position in the pipeline? My current method is clicking the little squares and selecting ‘Move Up’ or ‘Move Down’ which is fine, but a click and drag would be much easier. I think I’ve seen it done on a tutorial, but don’t know how.
Great work by all involved for this brilliant software!
This is really impressive!
Thank you all for you hard work!
I would be against removing favourites and groups as I use them all the time to just see the modules I use 90% of the time. I wouldn’t want to use search or more modules for these.
Plus using groups instils discipline into my workflow by moving from left to right through the groups.
I do agree with you about large scrolling bars being a bit messy, particularly in the colour balance module.
1 Like
There is a setting in the preferences allowing you to have only the active module open.
Ctrl-shift-click and drag.
2 Likes
Yes, I’m sure lots of people use it differently, which makes it tricky. With my suggestion, it should still be easy for one to access their favourites, perhaps with a load favourites button easily accessible. With groups, instead of moving left to right through the tabs, you could simply move top to bottom as they appear in the more modules list (grouped by category, so it is not just one long list like it is now). It would only take 2 clicks (more modules, module) - which is the same it currently takes (module group, module) - and only slightly more mouse movement. However its good to hear how you use it, devs no doubt need to take all use cases into account to make their decision, and if most people do it your way then a change might not be preferable.
1 Like
About your 2 cents on module reorganization, it would be better to discuss that on the PR where already some people discuss about it: regroup modules in 3 tabs : technical, grading and effects by aurelienpierre · Pull Request #5322 · darktable-org/darktable · GitHub
About moving module in pipeline, 2 ways: by clicking on multiple instances icon and menu or with Shit+Ctrl+drag&drop. You could read a more detailed thing about features like that introduced in 3.0 there: darktable 3.0 | darktable
1 Like
Being colour blind I just love the new RGB parade histogram option.
It’s always been a struggle to set the correct white balance, even using a colour picker etc.
The parade view makes it so much easier.
4 Likes
I am trying to replicate the part in the article about negadoctor and fail to do so as some steps are not described in the article or I simply misunderstand them. They could be about the white balance or any other previous step to be taken before using negadoctor. In my case this results in colors that look pretty good but the shadows (and true blacks) lifted so far that the images fall apart and the image overall become noisy and too bright. Workflow-wise, I make a linear scan with a semi-professional scanner at 16bit, load the image, open negadoctor and select an unexposed part of negative’s frame as color base. Next, I try to find a good balance with the dmax and scan exposure setting but not always with sucess. I also experimented with the corrections tab. Is there a detailed documentation somewhere? Otherwise, negadoctor looks great! I am using negfix8 so far.
We have a draft document here:
https://elstoc.github.io/dtdocs/module-reference/processing-modules/negadoctor/
Is this helpful at all, or do you think it needs improvement somewhere?
2 Likes
That was fast, thanks a lot! The documentation is pretty good. It also contains this key sentence ‘It is likely at this point that your image will still look too dark, but you can correct this later.’. In my case the image pretty bright and noisy in the shadows at this point already. Not sure why this is the case. I tried several images. I guess, I will experiment more and also adjust the exposure settings afterwards but this will risk bringing down the overall exposure too much and hence I will end up playing around with changing the shadows and highlights. I will also compare the results with negfix8 for those cases where negfix8 fails. Any ideas why negadoctor tends to lift the shadows so much (also leading to color noise in darker areas)? Maybe it is the way I scan (which is adjusted to the negfix8 workflow). Thanks again.
Okay, I found an interesting ‘effect’. Depending on how I draw the rectangle using the color picker for the film base will lead to very dark (sometimes almost black) or very light images even if I pick almost the same area but just draw the rectangle differently.
You need to draw the rectangle around an unexposed portion of the film, such as the little strip between the images. This lets negadoctor understand what a piece if film looks like that hasn’t been exposed to the light (Dmin). If you include any region that was exposed to light, this will mess up the calibration.
The other thing is to make sure you don’t have any tone curves enabled (no base curve, no filmic, etc.)
1 Like