Thanks for the advice. I’m a hobbyist photographer, been using lightroom for many years. Recently I tried out Darktable. So far I’ve managed to muddle through using base curve, tone curve, color zones, color correction etc and got acceptable results. (My lightroom skills are also just basic).
Now I’m reading up on this linear-rbg workflow and practicing using the Filmic RGB module. Haven’t yet managed to wrap my head around it. Will need a lot more practice for sure 
Another resource to understand one of the core principles of photo editing from the analog side: dodging and burning :
That makes the connection between dodging/burning and exposure compensation quite obvious.
I really miss dodge and burn tool in darktable. I know that you can imitate it with parametric and drawn masks, but it’s not the same as a brush, which you can drag discreetly, several times over some places to work on certain details. It is a very meditative activity and you feel a little bit like a painter. 
I loved it, thanks! I often hear about that process but never had the chance to see it. Now there it is.
I’m looking forward to a computer interface where I can make shadow shapes with my hands to dodge and burn like I use to under the enlarger. It’s a very intuitive way of working, once you get used to it.
Ah, I can dream…
Me too, hence Wiring darktable with Krita [nsfw]
Kinect controllers have been used for that kind of motion-controlled UI in the past. But managing drivers is a PITA for an app.
Hmmm. We could mount a video camera pointing at the computer screen, with a live video feed to the computer. The computer knows what is being shown on the screen, so can detect hands or pieces of card on wires that are blocking the screen from the camera. User feedback can come from painting a darkening “shadow” on the screen.
Back to topic, I think this raises the more general question of how to properly learn by oneself a skill set that involves theory, practice and experimentation using available resources with a critical mind to sort out BS from rigorous content. It’s pretty much Internet survival kit.
This is the difficult bit, when starting out from nothing, or even from a few years darktable use and a science background. And I fear to some extent we’re back to gurus, or at least someone who can lay out the learning path, and this was the one of the motivations behind the thread I raised (Image Processing – Recommended Reading).
You’ve done more than anyone else to try to help us get our heads around this but most (certainly beginners) would have trouble separating the intelligent and useful content you’ve generated from the “BS” that’s also out there. Personally as I’m learning more I’m starting to put together my own documentation (which may end up being something I can share with the community) but as you suggest it needs to be pretty comprehensive and every time I think about it, it grows some more.
As a rule of thumb, look at the pictures from the guy who is speaking : the results he is able to produce are the best clue about his level of craftsmanship.
Then look at his workflow : is there an apparent method/logic or if it is all random, does it looks tidy or cluttered and complicated ?
Finally, look at the way the editing is conducted : is there a clear goal at the beginning (“I want a bright commercial look” or “I want a dark and moody atmosphere”) and clear path from tool to result, or is it a succession of “let’s use that”, then “let’s use that” followed by random pulling and pushing sliders ?
Somewhere here I saw this link written by aurelienpierre posted.
Interesting to read! (and for his standard very short and easy to read 
) Just two question about this statement:
You can perform at least 80% of your processing with just 4 modules :
- exposure
- white balance
- color balance
- filmic
#1.) is exposure and basic adjustments the same?
#2.) with filmic you mean filmic rgb, right?
#1: yes and no - basic adjustment is intended to be used without further modules that affects tonal adjustments (one to rule them all) exposure is the module that can be used in combination with further modules
#2: filmic (without rgb) is only available for edits made with dt 2.6.
Yes.
Funnily enough, riding that article got me trying again, and starting off using just those modules, and I’m doing a lot better now. I wish I’d spotted it sooner.
Just another question on that.
Is there actually any difference in using the Basic module to adjust only exposure and black point, versus using the Exposure module to adjust the same things.
I have been using the Basic module simply because I followed Aurelien’s article and disabled many of the modules that were considered outdated.
Cheers
Basic adjustment is later in the pixel pipe than exposure, so several modules will get different input data if basic adjustment is used instead of exposure. This might have an impact on the result …
Btw: Aurélien isn’t really known to be a fan of basic adjustment module - so i wonder basic adjustment module survived when following his recommendations
My apologies, he does still recommend Exposure in that article and doesn’t mention Basic at all.
I think that confusion is a result of reading too many articles in the forums too. There are far too many confusing and conflicting discussions going on, which recommend a variety of modules in use.
This is of probably darktable becoming a victim of its own success, and the variety of modules available. As much as they’re great to have, are also so able to cause much confusion and argument as to how best to use them.
Some discussions have suggested that Basic is a full replacement for other modules, but you’re comments clearly show it isn’t. It’s no wonder I was having so much trouble getting images I was happy with.
Thanks.
The basic adjustments module is the result of a common mistake : in its guts, it’s exactly the same as the exposure module, colour balance (saturation and contrast options), etc. but instead of re-interfacing only their controls at the same place in GUI, it also duplicates the actual image filters and tie them together at the same place in the pipe. But exposure, contrast and saturation are better at different places of the pipe.
It’s the whole view/model/controller architecture that got mixed up (Model–view–controller - Wikipedia), but it’s only usual in opensource, aka what’s done on the GUI (controller) is not supposed to affect the pixel pipe (model) in a 1:1 fashion.
In my experience, rushed, uncareful or even bad design is not a property tied to opensource, as it is as common in closed-source/commercial software. It usually depends more on the culture/procedures of the project or company.
The inherent issue of opensource is it’s done by hobbyists. Good design needs time, a good understanding of the problem to be solved, a good knowledge of the state of the art, at least some knowledge of the audience and a couple of iterations. In practice, no hobbyist developer has time for that. Commercial software might be rushed too, but it’s done at least by people who do that all year long, chat with their peers, maybe meet their clients, and hopefully get a chance to understand what they are doing along the path.