The Darktable Setup I Can't Quit

I switched to darktable five years ago and I am not going back. Not because it is perfect ( it is not ), not because it is open source and I am some ideological purist ( I am not ). I use it because it works, because it does not ask me for twelve dollars a month, and because once you learn its infuriating interface you can process RAW files faster than in anything else I have tried.

Here is my entire workflow from SD card to finished image. No fluff, no philosophy, just what I actually do.

Why Darktable and Not Lightroom

I paid for Lightroom for three years. Then Adobe moved everything to the cloud, doubled the price, and made the import dialog slower. That was it for me.

darktable is not a Lightroom clone. It is its own thing with its own logic, and that logic is weird at first. The modules are named oddly ( what is "filmic rgb"? why is "base curve" separate from "tone curve"? ). The interface has about sixty panels and half of them look like they were designed by a committee of physicists.

But here is the thing: once you internalize the pipeline order ( scene-referred processing from left to right ), it clicks. And the results are better. Not "slightly better if you pixel-peep at 400%". Actually better, especially for dynamic range recovery on underexposed shots.

Import and Cull

1. Copy RAW files from SD card to my NAS. I use a simple script for this:

#!/bin/bash
# import.sh - copy RAWs from SD to NAS with date-based folders

SD="/media/dave/SD_CARD/DCIM"
NAS="/mnt/nas/photos/raw"
DATE=$(date +%Y-%m-%d)
DEST="${NAS}/${DATE}"

mkdir -p "${DEST}"
rsync -av --progress "${SD}/" "${DEST}/"

echo "Import complete. Files at: ${DEST}"

2. Open darktable, point it at the new folder. It auto-detects new files.

3. Cull in lighttable view. I use the star system: 1 star for "maybe", reject flag for garbage, no stars for "has potential". This takes five minutes per 500-shot session. Be ruthless.

The Processing Pipeline

This is where darktable differs from everything else. The module order matters. A lot. Here is my pipeline, in order:

1. Lens correction ( auto-applied if you have the profile ) 2. Denoise ( profiled, before anything else touches the data ) 3. White balance ( always from a gray card shot if I remembered ) 4. Exposure ( push or pull, usually +0.5 to +1.5 stops ) 5. Filmic RGB ( this is the magic module, more below ) 6. Color balance RGB ( for the final creative look ) 7. Local contrast ( subtle, just to give it some teeth ) 8. Sharpen ( output-specific, usually for web )

That is it. Eight modules. Most of my edits use six of them.

Filmic RGB Is Not Scary

Everyone is scared of filmic rgb when they first open it. The graph looks like a topographic map and the labels say things like "latitude mapping" and "shadow/highlight balance". Ignore 90% of it. Here is what actually matters:

- Middle grey luminance: set this to match your scene. Outdoors is usually around 18-20. Indoor shots, lower it.

- White relative exposure: drag this until your highlights stop clipping. That is it.

- Black relative exposure: same thing for shadows. Pull until the histogram looks right.

That is three sliders. The rest is for when you want to get fancy. Filmic does what Lightroom tries to do with "Shadows", "Highlights", "Whites", and "Blacks" but in one module, with better math, and without the halos.

My Presets

I have three presets I actually use. Not fifty. Three.

- Landscape: auto exposure +0.7, filmic with latitude 12, slight clarity boost

- Street: profiled denoise, filmic contrast +1 stop, local contrast aggressive

- Portrait: soft denoise, filmic latitude 9, color balance warm shift +5

Every photo starts from one of these and gets maybe two minutes of tweaking. The idea is to have a baseline that is already 80% there, not to automate creativity out of the process.

Export Settings

Web and social media:

Format: JPEG
Quality: 92
Width: 2048px ( long edge )
Color profile: sRGB
Sharpen: yes, radius 0.8, amount 0.3

Print ( when I bother, which is rare ):

Format: TIFF 16-bit
Width: native resolution ( no upscaling )
Color profile: Adobe RGB
Sharpen: yes, radius 1.5, amount 0.5

The Things That Annoy Me

Not going to pretend it is all perfect. The things that still annoy me after five years:

- The duplicate detection is non-existent. If I import the same folder twice, I get duplicates everywhere. I wrote a script to deal with this but it should not be my problem.

- Metadata editing is clumsy. Keywording works but it is not fast. Tagging 200 shots with location data takes way longer than it should.

- The map view uses OpenStreetMap tiles which are fine until you zoom in on Montevideo and half the street names are missing. Minor, but I use the map view a lot for geotagging.

- GPU acceleration exists but it is janky on AMD. My RX 6700 works about 60% of the time. The other 40% it falls back to CPU which defeats the point.

None of these are dealbreakers. But they are real. I am not going to tell you darktable is flawless because it is not.

Why I Stay

The raw processing quality is better than Lightroom. I do not say that lightly. I have done side-by-side comparisons on the same files, same exposure adjustments, same white balance. darktable's filmic rgb module preserves more detail in both shadows and highlights than Lightroom's tone mapping. The color science is more flexible. And once you build your presets, batch processing is fast.

Also: it runs on Linux natively. No Wine hacks. No virtual machine. Just native. That matters to me more than it should.

And it is free. Not "free with a subscription" or "free until we change the terms". Actually free. The source code is there. If the maintainer gets hit by a bus tomorrow, someone else picks it up. That is more than Adobe can say.

If you shoot RAW and you have not tried darktable, give it a weekend. It will confuse you on Saturday and make sense on Sunday. That is the learning curve in a nutshell.

https://www.darktable.org/