Use case
Remove a person from a photo
Remove the person from a photo who should not be in it, and MightyPhoto reconstructs whatever was behind them. The same tool clears signs, power lines, cars and trash cans.
Photobomber removed
Step by step
How to do it
Upload the photo
Group shots, holiday scenery with a stray tourist, family portraits. Anything works, at up to 10 MB.
Say exactly who to remove
Identify them by position or clothing so there is no ambiguity: 'the man in the red jacket on the far left'. In a crowded photo, a precise description matters more than anything else.
Download the corrected photo
The background, shadows, and any reflections behind the person are rebuilt to match, in about ten seconds.
Describing who to remove
The tool has one box, What should disappear?, and everything depends on it. Describe the person so nobody else in the photo could match: their position, what they are wearing and what they are doing. “The man in the red jacket on the far left” works. “The guy in the back” does not, because there may be three.
In a crowd it is often easier to say who stays: “everyone except the bride and groom”. Shadows and reflections count as part of a person. If someone left a shadow on the sand or a reflection in a shop window, type “also the shadow they cast” in the Anything else? box.
Objects work the same way. Name the thing and where it is: “the trash can by the front door”, “the power lines across the sky”, “the sign on the right”.
Photos that work and photos that fight back
The AI fills the gap by continuing what surrounds it. A person against sky, sea, sand, grass or a plain wall comes out clean, because there is plenty of the same thing on every side. The harder cases are predictable. Someone standing in front of text, a tiled floor or a crowd leaves a gap full of detail that has to be invented. Someone overlapping the person you want to keep, like an arm around a shoulder, leaves a join that has to be rebuilt. Someone who fills a large part of the frame leaves a big area with nothing to copy from.
When a result is close but not right, change the wording and run it again. A more exact location or a note like “keep the railing behind him” often fixes it. Each attempt is one credit, and the original stays in your library.
Compared with doing it in Photoshop
By hand this is clone-stamp and patch work, or a content-aware fill followed by cleanup. A tourist against a plain sky takes a few minutes. A person in front of brickwork or a busy street can take an hour of careful work, plus software that you need to own and know how to use.
Here you type a sentence and get the result in about ten seconds. What you give up is pixel-level control. You cannot paint a mask or pick which part of the background gets copied. If one small area matters a great deal, such as a gap across a face, check it at full size before you print or share.
Prompts
Prompt ideas to try
Copy any of these as a starting point, or write your own.
"the person standing in the background on the right"
"the man in the red jacket on the far left"
"everyone except the bride and groom"
"the tourists behind us on the steps"
"the power lines across the sky"
Objects work the same way.
Questions about remove a person from a photo with ai
How well does the background blend after a person is removed?
The AI rebuilds what was behind them from the surrounding image. Uniform backgrounds like a wall, sand, sky, or grass come out cleanly. A person standing in front of complex detail, such as text on a sign or a patterned tile floor, is harder and may need a second attempt.
Can I remove more than one person at once?
Yes. Name each of them, or invert it and say who to keep: “everyone except the two people in the centre” is often the easier instruction.
Can I remove objects as well as people?
Yes. The same tool removes a trash can, a car, a sign or power lines. Describe the object and where it is.
Can I add someone to a photo instead?
Use Freeform Edit and describe them: build, hair, clothing and where they stand. Reproducing a specific person's exact likeness needs a reference photo of them, which is not supported yet.
