I prefer Darktable over Rawtherapee, but both are great open-source RAW development programs. LR is obviously more fully featured but for amateur or semi-pro they are more than acceptable.
IIRC there have been a few times when CS researchers working on image algorithms published their findings with a Darktable implementation first.
This is one place where there's just no replacement to hiring engineers to sit down and work on the product for a few years. The open-source stuff is nowhere close to CaptureOne or Lightroom in terms of quality of RAW conversion or the quality of the image processing.
Does CaptureOne support a "RAW+sidecar metadata" format? It is important to me that the raws not be modified (i.e. metadata not written directly to them) and that the metadata be a per-file sidecar and not just a single giant catalog file. Lightroom seems to support this workflow but others didn't seem to (or at least didn't explicitly state they did).
That would be the ideal workflow for me in terms of generating useful catalogs and making backups smooth.
Not only that, but they also have an extensive collection of camera profiles and lens corrections for pretty much every vendor. Their support for Sony and Fuji are particularly great.
> "Open-source workflow" means you're leaving quality on the table.
That is true, but my choice of software is not uni-dimensional, I also care about trusting the software I run in my computer and their maintainability in the long term, to highlight two other important dimensions for me. When considering all this together I have little choice but to use free software exclusively.
Of course different people will assign different weights to each dimension, and that is completely fine, but let's not oversimplify the decision to just "quality".
I expect that the professionals take better pictures straight out of camera and need less adjustments while the dabblers like me need more post processing help. Now, I am struck on a loop with Topaz and LR and cannot switch out of windows (or mac).
Frankly, this is all BS. All of these softwares are capable of producing professional results. The difference is made by the user, not the software. Learn one of them; any one of them. Learn it well. The individual choice in the end is just a preference, but does not determine quality. You do. Thankfully.
(That's a bit like modern cameras. They are all great. They are all capable of professional results. The individual choice in the end doesn't matter.)
Using Darktable has gotten easier over time. But it definitely has some things that are probably not that intuitive for new users. The lack of features and opinionated nature of Lightroom might be considered a feature for some but I don't think there is anything specific that it does that doesn't have at least several alternatives in Darktable at this point. We can haggle about the quality of the algorithms on either side of course. I'm pretty happy with what Darktable does here and it seems it has developers that are really obsessive on this front. Check e.g. lead developer Aurelian Pierre's youtube channel for some in depth discussion of what Darktable does and why it works the way it works.
The transition to the so-called the scene referred work flow in Darktable in recent years has made my workflow a lot simpler. But it is also something that has a bit of a learning curve.
Scene referred means that instead of applying some one size fits all curve based on your camera model, it actually looks at your photo and applies the exposure and filmic modules which together will try to fit the curve for each photo using some heuristics and nice algorithms. The result is typically pretty good out of the box and any further tweaks tend to be straightforward.
A typical workflow for me in Darktable works like so:
- import photos and move them to the right place. I'm not big on tagging, so I tend to skip that.
- apply some initial star ratings in the lighttable section so I can focus on the ones I like the most.
- after initial screening, anything below 2 stars gets hidden and I start editing.
- I open the photos for editing one by one and let Darktable apply its defaults (and my overrides). For example, my Fuji requires +1.25 stops exposure correction as it tends to underexpose to avoid blowing highlights and I have a preset for this that auto applies. Auto applied settings and copy pasting parts of a history stack or applying them as styles is one of many workflow enhancements that allow you to be productive in Darktable. Worth reading up on if you have a lot of photos to process.
- I tweak the exposure defaults manually to set the gray point. The nice thing with this is that filmic adjusts black and white points accordingly. You can think of this as an intelligent way to set the brightness.
- tweak filmic parameters if still needed (quite often this is not needed). I do typically apply some contrast but not a lot. This is also the place to deal with highlight recovery if that is needed.
- Crop & rotate as needed. You can do this at any point of course but I like to get this out of the way early.
- decide if I want lens correction and local contrast modules turned on. Not every photo needs this. Lens correction is nice for wide angle photos. Local contrast can be nice but I try not to over use it.
- Deal with any color issues and saturation as needed. There are various modules for this, including a few recent additions that work well with the scene referred work flow. Mostly Darktable does the right things here and I don't actually need to do a lot here. But I do like the new perceptive saturation slider in the color balance module.
- Apply other modules as needed; I tend to be conservative with this. For example noise reduction is sometimes nice and profiled noise reduction does a good job out of the box but sometimes when noise is really a problem there are alternative strategies to explore.
- export everything with 2 stars or more as jpg & upload it some place for publishing
The place where I store my RAW files is not where I work on them - I'll typically copy them from a memory card to my machine, do whatever processing, then shift them away to storage, so I find the whole idea of a 'library' unecessary.