You're tethered, directing a subject, watching the light change, and trying to catch problems in the frame before the moment is gone.
Most of that catching happens later. You shoot the set, then you sit down and review, and by then the moment you'd fix is over.
I built something for that gap.
Darktable Critique watches your tethering folder and sends each new capture to a Claude vision model for feedback, while you're still in the session. Composition, lighting, posing, technical quality, all scored and explained in the time it takes you to set up the next shot.
It's not a replacement for editing in Lightroom afterward. That's still where the real selects and retouching happen. This is for the twenty minutes before that, when you're still deciding what to shoot next.
What it does:
- Real time feedback during tethered shoots, so you're not waiting until the edit to catch a problem
- Genre aware critique for headshots, portraits, editorial, fashion, food, still life, and street work
- Four verbosity modes, from a single glanceable line during a fast session to a full breakdown for later review
- Choice of Claude models, so you can balance cost against how much nuance you need
- Session history and action points, so feedback becomes a checklist instead of a comment you forget
- Off by default. Nothing leaves your machine until you turn it on, and the plugin comes with consent language for telling clients before it does
Darktable Critique lives in a private repo right now. I'm running it on my own shoots and offering:
- guided demos
- prototype access
implementation notes for photographers curious about AI feedback in a tethered workflow
If you want to see it in action or talk about fitting something like this into your own setup, reach out for more info.
Built as part of Knightvision Labs.
