Remove Person from Photo

Keep the moment. Lose the distraction.

A stranger at the edge of a travel photo should not decide how you frame the memory. Select the people you would like to remove, then ask AI to fill the marked area from the surrounding scene. You stay in control of the selection: there is no automatic face detection or one-click promise to remove every person.

The animation above uses AI-generated photo-style source and after-frame assets. It is a prepared demonstration, not a live result from this site's provider. Opening a photo and adjusting its selection happen locally. Clicking Remove submits the working image and selection for AI processing.

Three steps, with room to review

1. Upload your photo

Choose one JPG, PNG or WebP, or drop it into the editor. Start with a reasonably sharp photo in which the background around the unwanted person is visible. Your photo opens locally. Uploading into the editor does not start an AI request.

2. Brush over the person

Cover the full outline, including small gaps between arms and legs when those areas should be reconstructed. Adjust the brush size for the current working image. Use the eraser to protect nearby faces, hands or details. Undo, redo and clear selection let you revise before submitting.

3. Generate, compare and download

Remove submits one task to the configured provider. Compare the same frame before and after; check the result at its actual resolution. Download the latest successful PNG, or keep editing it. There is no automatic retry if processing fails.

A better selection starts with the background

A modest margin around the unwanted person can give the repair room to blend. Avoid sweeping over a large area just to make the outline easier. The server rejects an empty selection or one covering more than 90% of the image. That is a safety boundary, not a recommendation to select nearly the whole frame.

Use Move mode to reposition a zoomed image, then switch back to Brush before painting. Mobile uses the same explicit modes: one finger pans in Move mode and paints in Brush mode. Pinch gestures are not implemented. You can scroll the page outside the canvas.

Travel photos and background passersby

An isolated passerby against sand, water, sky or an evenly textured wall is easier to select than a person woven through railings or foliage. Nearby background still matters: do not assume the system knows how a hidden stair, handrail or sign should continue.

Shadows and reflections

Removing a person but leaving their shadow can make an edit look unfinished. Add an unwanted shadow or reflection to the same selection when it is clearly associated with that person. Keep the main subject's shadow intact. Review the repaired light and texture rather than checking only the original outline.

Groups and overlapping subjects

People in a group may cover one another. If an unwanted person overlaps a face, sleeve or hand you want to keep, those hidden details are not recoverable facts. Try a smaller selection, inspect closely, or choose a different source photo instead of expecting an accurate reconstruction.

What AI removal can — and cannot — do

This is background reconstruction, not a hidden-layer recovery tool. A result can contain seams, repeated textures, changed writing or altered unselected details. The final composite retains the original working-image pixels outside your selection. Repairs inside the selection still need careful review. We will not describe it as lossless, guaranteed or perfectly background-preserving.

Working images keep their original frame and aspect ratio, but photos larger than the working limit are reduced to a maximum side of 1,536 pixels. The editor shows your working and download dimensions. For AI processing, the photo and selection are fitted together onto a standard canvas. After processing, the padding is removed and the edited area is resampled to the working dimensions. Pixels outside your selection come from your working photo. Unexpected canvas sizes are rejected.

Clear limits, no account

The free allowance is 3 AI attempts per network per UTC day, shared across person removal and object removal. An admitted attempt can count even if the provider later fails. Shared networks may share the allowance. Editing, comparing, restoring a saved result and downloading do not use AI attempts.

AI availability also depends on the service-wide allowance. When it is unavailable, you can keep your selection, download an existing result or use local cropping.

Your photo, your choice to submit

File decoding, brush strokes and crop adjustments happen locally. Clicking Remove sends the working image and mask through our server to a third-party AI provider. We do not establish a public image library or intentionally log photo contents. Provider retention and commercial-use terms still need verification before launch; do not submit sensitive images on an unverified assumption of deletion.

Read the privacy details, editing terms and step-by-step guide. If the distraction is a bin, cable or sign rather than a person, use the object-removal workflow.