Grandma's birthday
Ninety candles and a single photo, taken in the dark just as she blows them out. The bright version goes into the family album; the original stays on the phone.
Brighten a photo that is too dark with AI
Your too-dark evening photo comes back well exposed with natural colours, then in golden light, candles still lit and no grain. Workflow visible here, no account needed.
Real workflow: a nearly black table at the end of a birthday dinner, cake with lit candles, then the same scene well exposed with the whole room visible, and in warmer golden light.
At a glance
- You provide
- 1 photo · a text or a brief
- What the workflow produces
- 2 images in 4:3
- Canvas structure
- 6 nodes wired by 4 connections
- Models wired in
- Flux Kontext
- Cost of one full run
- about 12 credits, roughly $0.14
- Account
- The demo is visible without an account; generating requires a free account, trial credits included
The cake comes out, someone dims the lights, a phone comes out. The result: a nearly black photo with a few tiny flames in the middle and grain everywhere. And that was the moment worth keeping. This workflow takes that photo and gives you two versions you can see: one well exposed with natural colours, one warmer, like the evening felt.
Brightening without wrecking it is the hard part. Pushing exposure brings up grain and washed-out blacks. The prompt keeps the candles lit and forbids moving anything. Flux Kontext rebuilds the scene cleanly, but it invents what was in the dark, and its two versions end up fairly close.
If the photo is blurry too, run it through the photo sharpener afterwards; for an old damaged print, see photo restoration. Both versions cost twelve credits.
Actual renders from the demo
These are the files of the workflow above, exactly as the models produced them, unretouched. The starting point is shown when there is one.
Starting point
Render
Who it is for
Families rescuing a party photo, restaurants showing their dining room at night, anyone who missed a shot without flash.
Which files to provide
The sharpest dark photo you have, and what should come through: objects, faces, light sources to keep on.
How it works
- 1
Pick the photo
The sharpest of the batch, even if it is very dark. The model brightens well; it does not fix camera shake.
- 2
Name what matters
The cake, the faces, the table, the flowers: whatever should come through first. The rest follows.
- 3
Keep the mood
Lit candles, a lamp, fairy lights: say they stay glowing, or the photo turns into a daytime shot.
- 4
Compare with the original
Look at what was black and what appears. Background details are rebuilt, so check they don't get in the way.
Variations and use cases
The same canvas, tuned for different needs: each variation is two or three lines of prompt away.
The restaurant at night
The manager wants to show the evening atmosphere on her page, but her photos are black. The warm version keeps the candles and makes the plates readable.
The wedding without flash
The reception ran on after dark and the guests shot without flash. The best phone photos become good enough for the guest book. The moments are kept, and nobody has to admit they forgot the flash.
Christmas with the family
The tree is lit, the living room is dim. The photo of the Christmas Eve table comes out readable, fairy lights still twinkling, without the grain of auto mode.
The flat photographed after work
A private seller shot their flat after getting home from work and every room is dark. The natural version works as a stopgap until they reshoot in daylight.
The holiday dinner
Dinner on the terrace at sunset, one shot and almost nothing visible. Brightened, it earns its place in the summer photo book. The glasses, the plates and the faces around the table finally show up again.
Common mistakes
Starting from a blurry photo
The model brightens; it does not undo camera shake. Pick the sharpest photo in the batch, even if it is the darkest.
Forgetting the candles
Without "the flames stay lit", the model may light the room like daytime and the atmosphere is gone.
Trusting what comes out of the dark
In black areas, the model rebuilds. Books, floor or a painting may not match reality. If it has to prove anything, keep the original.
Expecting two opposite moods
In the demo, natural and warm stay close. For a real contrast, describe the light source, candles only or ceiling light.
Recommended models
Flux Kontext
Nano Banana 2.1
GPT Image 2.5 SunburstFrequently asked questions
Do the objects move?
No: in the demo, the plates, glasses, crumpled napkins, party hats and gift bags all stayed where they were.
Does it work with people in the photo?
Yes, but name them in the prompt: "keep exactly the same faces". In the dark, a rebuilt face can shift a little, so compare.
What about a gig or an outdoor party?
Yes: adapt the list of objects, stage, lights, crowd, and say which light sources should stay on.
How much do the two renders cost?
Twelve credits, six per render on Flux Kontext. A third mood on the same photo costs six more.
Why Flux Kontext?
Because it reworks the existing photo instead of making a new one: the table, the objects and where they sit stay those of your evening.
Are the details real?
The ones you could just make out, yes: cake, glasses, flowers. Anything in complete darkness is rebuilt by the model, which is why you keep the original.
Why do the two versions look alike?
Because "warm" is only a shade of light. For a bigger gap, ask for a room lit only by the candles, with corners left in shadow.
What output format?
The format of your photo, here 4:3. Nothing is cropped.
Brighten a photo that is too dark with AI
Use this workflowFree account, no card required. The workflow opens pre-filled in your canvas.