Auto develop: AI that moves the sliders
Most AI photo enhancers hand you back a new, flattened picture. Auto develop works the way a photo editor does: it looks at the photo and proposes values for LightCraft’s sliders. You see every change before it happens, and it stays an ordinary, adjustable edit.
- In LightCraft, select a photo (click it in the grid or open it with E).
- Open AI tools in the top bar and pick a goal: Balanced, Bright & airy, Moody, Punchy landscape, Soft portrait or Fix exposure. Or describe your own.
- Press Suggest settings. The panel exports a small JPEG of the photo as it looks now, reads its current slider values, and sends both to the model.
- Review the list: each slider with its current and proposed value, and a short note on why.
- Press Apply to photo. LightCraft applies all values as one step; ⌘Z or Ctrl+Z takes it back.
The model can set these 12 sliders, within LightCraft’s own ranges:
Exposure, Contrast, Highlights, Shadows, Whites, Blacks, Vibrance, Saturation, Texture, Clarity, Dehaze, Vignette.
White balance is left out on purpose: LightCraft measures it in Kelvin for raw files and on a relative scale for JPEGs, and a model can’t tell which from a preview. Use the white balance eyedropper or the Auto white balance instead.
Retouch: edits only an image model can make
Some fixes need new pixels, not slider moves. Retouch sends a photo to an image model and gets a new picture back.
- Remove an object or remove distractions such as signs, litter and people in the background, filled in to match.
- Remove the background. With an OpenAI model and “Transparent background” ticked you get a transparent PNG; other models put the subject on white.
- Replace the sky with one you describe, with the light matched.
- Restyle the photo as a painting, sketch or any look you describe.
Pick the photo with Use the selected photo, or export one from LightCraft with File › Export… (it lands in the panel and still downloads as usual), or upload any image. Each result can be added to your LightCraft library as a new photo with one click, edited again, or saved. The new photo appears under Recently Added, ready to develop like any other.
Image models redraw the whole picture, so fine detail can shift, and results are 8-bit images rather than raw files. Do your raw development first, then retouch.
Supported providers
| Provider | Auto develop model | Retouch model | Get a key |
|---|---|---|---|
| OpenAI | gpt-6.1-sol | gpt-image-2.5-flare | platform.openai.com |
| Google Gemini | gemini-3.8-flash | gemini-3.1-flash-image | aistudio.google.com |
| OpenRouter | google/gemini-3.8-flash | google/gemini-3.1-flash-image | openrouter.ai |
| Custom (OpenAI-compatible) | any vision model your API serves | any image-edit model | your own endpoint |
These are defaults; “Load models” in the panel lists what your key can use. OpenRouter gives one key access to many models, including Gemini, GPT, Claude, FLUX and Seedream. The custom option works with any service that implements OpenAI’s /chat/completions (with image input) and /images/edits and accepts browser requests.
How it connects to LightCraft
The LightCraft web build has a small automation interface, lightcraft.command(), that runs any of its commands by name. The panel uses it to read which photo is selected and its slider values, to export a preview, and to apply the sliders you accept with LightCraft’s develop.set command. Retouched images go in the way a dragged file does, through LightCraft’s normal import. Nothing in LightCraft is modified.
Privacy: what goes where
| Data | Where it goes |
|---|---|
| Your API key | Stays in your browser. Sent only to the provider, in the request header. |
| The photo preview, slider values and your goal or prompt | Sent from your browser straight to the provider you picked. |
| Your library, originals and everything else you do in LightCraft | Never leaves your device. |
| lightcraft.run servers | Serve the static app files. They receive none of the above. |
The provider’s own terms and data policy apply to what you send them.
FAQ
Do I have to pay to use the AI tools?
You pay your AI provider for what you use, at their normal API prices. lightcraft.run charges nothing and adds no markup, because the request never passes through us.
Does Auto develop change my photo’s pixels?
No. It only sets LightCraft’s own sliders, the same ones you would drag yourself, so the original stays untouched. The change is one step in the photo’s history: press ⌘Z or Ctrl+Z in LightCraft to undo it, or keep dragging the sliders from there.
How is Auto develop different from LightCraft’s Auto button?
LightCraft’s own Auto Settings (Shift+A) runs on your device, needs no key and gives a quick neutral starting point. Auto develop takes a goal in plain words, such as “moody and dark” or “keep the sky, open the shadows”, and asks a vision model of your choice.
Where is my API key stored?
In your browser only. By default it lasts for the current tab (sessionStorage). Tick “Remember my key on this device” to keep it in localStorage instead, and use “Forget all keys” in the panel settings to delete it.
What does the AI provider receive?
For Auto develop, a JPEG of the selected photo at 1024 pixels on the long side, with your current slider values and your goal. For Retouch, the image you picked, at most 2048 pixels on the long side. Exports made for the AI panel leave out GPS location.
Can the AI open my RAW files directly?
No. The panel asks LightCraft to export the photo as a JPEG with your current edits, the same way File › Export… does, and sends that. Your RAW file never leaves the browser.
Why does a request fail with a CORS or network error?
The custom provider option needs an API that accepts requests from browsers. OpenAI, Google Gemini and OpenRouter all do. Some self-hosted or regional gateways don’t; use one of the built-in providers instead.