When I wrote my entry on Linux RAW processors I said that Rawtherapee was a better choice than darktable. I have to take that back because it turns out my quick tests weren't a good representation of using either program for real.

I formed my initial views after test-processing just a couple of photos with each program. Now that I've used both to process batches of photos for real (and in one case I've run the same batch through both), I've had to change my opinion. It turns out that darktable is what you want to use, not Rawtherapee. For all of darktable's irritations, it works better. I summarized the main reasons why in a tweet:

Darktable drives me up the wall and I hate the experience of using it, but it delivers better results than Rawtherapee and does it faster.

(I've since become more acclimatized (or numb) to darktable's interface issues.)

The first issue is that Rawtherapee turns out to be relatively terrible for sorting through a bunch of photos and figuring out which ones are worthwhile. I could do it, but it took too long and was a pain in the rear in all sorts of ways because Rawtherapee has fumbled multiple aspects of doing this efficiently. Darktable is not great at this but in practice I can go through a bunch of photos much quicker and more efficiently with it. Since this is a major part of my daily workflow, this matters a lot to me.

(For example, Rawtherapee has absolutely and utterly terrible downsizing of thumbnails in its directory overview, to the point where they are basically useless for telling you anything about the quality of the photos. Think of the most crude and jagged downsizing you've seen; that's Rawtherapee.)

The other part is that I get better processed photos with darktable, in that I like how they look and it's (much) easier to produce what I think of as good looking photos. Again, darktable is not perfect and there are some things that Rawtherapee unquestionably does better, but darktable wins overall for me. I find it very hard to argue with clearly better results, especially when I can get them surprisingly rapidly and easily.

Now I'm going to say something that may make people especially unhappy, because there's a third advantage to darktable. Namely, it's under much more active development than Rawtherapee (I track the source repositories for both and darktable sees multiple commits a day whereas Rawtherapee moves much slower). I know that development activity doesn't necessarily equal quality, but both programs are highly imperfect right now so the one that's under much more active development is much more likely to improve into something good (or at least have your favorite irritation fixed).

(Note that with either program you want to be using the latest version compiled from the project's source repository. Both are under active development and improvement and yes, it really makes a difference. Probably not as much a difference as in my initial tests (where the then-current release versions produced bad output), but you'll find that both are better experiences.)

(As before, you should pay attention to the publication date of this entry if you're coming here through a web search. It's quite possible that things will be different in a year or two. I certainly hope that Rawtherapee improves substantially over time and at least some of its issues should be relatively fixable.)

Sidebar: what happened to make me discover this

I didn't set out to try out both on a batch of real photos; instead, I set out to process a batch with Rawtherapee because I thought it was the program for me. After slogging through the whole process and getting a trio of processed photos that I wasn't really enthused with, I decided to re-run the same batch through darktable just to see. Much to my surprise I was able to do so much faster and I was uniformly much happier with the results, to the point where I immediately replaced all of the Rawtherapee versions I'd uploaded to Flickr with the darktable versions.

(The extra speed with darktable didn't come because I immediately zeroed in on the 'best' photos and only dealt with them. I reconsidered all of the batch from scratch in darktable, although I wasn't surprised to wind up with the same set of selects.)