[qurren]

I've used Darktable for almost a decade quite happily, but recently am moving off of it because they made one MASSIVE blunder with the new shitty color grading and white balancing tool that never gets any of my images right, all the tones are off, and it doesn't have sane defaults that renders things at least "almost correctly" at first go like the previous white balance tool did. It was way better a couple years ago with the vanilla "White Balance" and "Base Curve" tools.

Darktable also cannot correctly color any of my DJI Mini Pro imagery. None. They all have really bad greenish hues. I haven't been able to publish any of my DJI images for 4 years now because Darktable won't fix it, and it's a known issue that they explicitly plan to not fix (https://github.com/darktable-org/darktable/issues/13319).

I'm currently starting to love RapidRAW (https://github.com/cybertimon/rapidraw); the developer is super active and releases new features every couple of weeks. I'm also able to add a few minor nitpick features I want quite easily with AI code tools. I have a good feeling I can get AI to fix the DJI issue on RapidRAW since the code is quite nicely organized.

[blutack]

I don't think your assessment of the DJI Mini thing is very fair. Sounds like they would be more than happy to accept a pull request and indeed there is an open one the maintainers are helping get merged [1].

Shame that the new workflow doesn't work well for your particular camera and setup - but I would encourage other people to try it, because it works great for mine.

  We know :-) It very much seems there is no dt developer currently interested in this sort of correction, i suppose none are using such devices.

  Just two reminders:

  Such "missing support for my camera" started me to get into darktable code. A lot of code to be reused is there, there are already some hints, just go ahead :-)
  Some comments above i suggested something ... :-)

1: https://github.com/darktable-org/darktable/pull/20467

[pmarreck]

I now have a business idea: Curated Open-Source App Forks.

Would you pay to have access to a service which tweaked the open-source software you use just enough to make it much more useful to your workflow, and automatically updated it over time? (I'm not just talking about a one-time fork; I'm talking about a subscription to get access to a maintained fork that stays current with upstream, fixes any conflicts, does the rebuild, hosts the build, etc... This justifies a cost) This incompatibility with DJI images that the Darktable devs #wontfix is a perfect example use-case.

Needless to say this would be for professional use, with professional expectations, but if your business depends on having that slightly-tweaked tool, then it may be worth it; just exploring the idea, here

[qurren]

Honest answer from me: Probably not, because vanilla Claude Code already does a great job of maintaining my forks with open-source main, and also handles the packaging and installation beautifully.

There is a huge overlap between (OSS users, people who can code) and an even bigger overlap between (OSS users, people who can use coding AI tools)

The people who can't even use coding AI tools almost universally use closed-source software like Adobe Lightroom that you can't fork easily.

[pmarreck]

There's quite a difference in my experience between "using coding AI tools" and "using coding AI tools _well_", especially when it comes to editing code.

[qurren]

Doesn't really matter in this case. Doesn't matter enough for people to pull out their wallet. You can re-fork from clean OSS and re-implement any time while you sleep, using whatever AI subscription you already have.

[pmarreck]

I think you overestimate the number of photographers who are also comfortable with CLI tools, because doing this via web-facing UI's is quite less capable currently

[qurren]

Those photographers you speak of (95%) are using Lightroom and Photoshop and you wouldn't be able to build a fork business for them. The 5% who use Darktable are likely comfortable with CLI tools.

[jasonjayr]

Isn't this what happened with ElasticSearch and AWS OpenSearch? I imagine that OSS developers will not appreciate this at scale, especially if the core developers are not reaping some of the revenue from such a project.

[seemaze]

VPD - virtual private developer. Sell off parcels of developer attention by the hour!

[actionfromafar]

Somebody™ should do it with Ansel IMHO.

[jollymonATX]

Nope

[mikae1]

> but recently am moving off of it because they made one MASSIVE blunder with the new shitty color grading and white balancing tool

If you're talking about color calibration and color balance rgb you can disable them and use darktable exactly as before.

[c7b]

Stupid question, if there's an old version that you liked, couldn't you just keep using that version?

[qurren]

It's a pain in the ass to maintain when it has dependencies on older stuff while I upgrade the rest of the system

[epihelix]

Well, you shouldn't need to maintain - just compile a static linked binary and you should be good for the rest of your system's life.

But more practically - I've been using darktable for well over a decade, and it's remained backward compatible with my old styles. If you like the older modules, they should still be there, and you should still be able to apply your old styles the same as always.

My own problem is that darktable keeps improving, so much so that I want to reprocess all my old photos again :/

[jimnotgym]

I followed the github issue you posted.

There is a pr they are trying to merge, last comment was...10 hours ago. I'm not sure how this is 'actively plans not to fix'