AI IMAGE EDITOR · FREE WHILE IN BETA
AI photo editor
Describe one change and keep everything else. Change a colour, swap a background, remove an object or add a caption — then keep going, round after round, without the face drifting.
Free while in beta · up to 20 edits a day · one instruction per round keeps the rest of the photo pixel-identical.
Your photo and every round of edits appear here.
How to edit a photo with words
- Choose a photo. It is reviewed once, then cached for ten minutes so later rounds start immediately.
- Write one instruction — “change the shirt to deep green” — and leave the rest alone.
- Press Edit. The result becomes the new reference for the next round.
- Repeat, or click an earlier thumbnail to branch from that version.
- Download a PNG, or a cut-out with the background removed.
One change per round is not a limitation, it is the technique. A prompt that asks for five things at once gives the model five chances to move something you wanted kept.
What this editor is good at
Colour and material swaps
Clothing, walls, upholstery, product shells. The pixels outside the change stay where they were, so a catalogue shot keeps its lighting and its shadows.
Background replacement
Studio white to a real room, a street, a beach. The subject keeps their edges and their light because only the background is being redrawn.
Object removal and addition
Take a bin out of a doorway, add a chair to an empty corner. Ask for what should be behind the object so the fill has something to build.
Text on images
Captions, labels and short headlines come back readable. Put the exact words in quotation marks and add “no other text” — that single clause removes most garbled-letter surprises.
AI photo editor questions
What makes this different from a filter or a one-shot AI image generator?
It edits the photo you already have instead of inventing a new one. Each instruction is applied to the current image and everything you did not mention is held fixed — face, hair, clothing, pose, lighting, framing. You can chain edits round after round and come back to any earlier version from the thumbnail strip.
Will my subject still look like the same person after several edits?
Yes, that is the whole point of the round-by-round workflow. The editor sends the previous result back as the reference every time and appends a fixed instruction that lists what must stay identical. The model behind it (ByteDance Seedream) is built for exactly this: its published tests run the same portrait through a hundred rounds and keep the face, accessories and background in place.
How do I get the best result?
Ask for one change per round and say what must not change. "Change the shirt to navy, keep everything else identical" beats "make it look better". If a detail keeps drifting, name it explicitly in the instruction — badge text and small accessories survive far better when the prompt calls them out.
Does it support transparent backgrounds?
The model returns a normal photo, and this page can knock out a near-white background to give you a transparent PNG in one click. True native alpha output arrives with Seedream 5.5; until then the cut-out button is the honest bridge, and it works best on a clean, evenly lit background.
Is it free, and what are the limits?
It is free while in beta: up to 20 edits a day per account, with a site-wide ceiling that keeps the provider bill bounded. No credits are deducted, and a failed edit never costs anything. Sign-in is required so the quota can be attached to an account rather than an IP address.
What am I not allowed to edit?
Photos of children, or anything sexual, violent or hateful. Every uploaded photo is reviewed before the first edit and every prompt is scanned before it reaches the model. The written rule is simple: adults only, and nothing you would not show a colleague.