How to Remove Unwanted Objects or People From a Photo

A stranger walks through your shot of the cathedral. A trash can sits right behind someone’s head. Power lines cut across an otherwise perfect sunset. Removing distractions like these used to require careful clone-stamping in a desktop editor. Today, AI fill tools make it a matter of brushing over the object, but the results still depend a lot on how you brush and which photos you choose.

This guide covers the main methods, when each works best, and the habits that separate a clean removal from an obvious smudge.

The three main approaches

1. Clone stamp

The clone stamp copies pixels from one area to another. You pick a source point, then paint over the object.

• Best for: repeating textures with clear structure, like brick walls, tiled floors or fences, where you need precise control.
• Weakness: slow, and easy to create visible repeated patterns.

2. Healing brush and content-aware fill

Healing tools copy texture from a source but blend the color and brightness with the destination. Content-aware fill (and similar patch tools) analyzes the surroundings automatically.

• Best for: small to medium objects on fairly uniform backgrounds such as sand, grass, sky and plain walls.
• Weakness: can pull in the wrong texture near edges, or smear structure like lines and corners.

3. AI generative fill (inpainting)

AI inpainting models predict what should be behind the object based on the whole scene, not just nearby pixels. They can continue a railing, rebuild part of a building, or extend a crowd-free pavement in the right perspective.

• Best for: larger objects, people, and complex backgrounds where copying pixels wouldn’t make sense.
• Weakness: can invent details that weren’t there, especially text, faces and fine patterns. Always review at full size.

Most phone gallery apps now include a version of this (Google Photos’ Magic Eraser, Apple Photos’ Clean Up, Samsung’s Object Eraser), and dedicated tools like DirtAI are built around the same brush-and-fill idea, with more control over the area and the result.

Step-by-step: removing a person or object

Step 1: Start with the highest-quality file

Use the original, not a screenshot or a compressed copy from a messaging app. More pixels and less compression give the model more texture to work from. If you shoot RAW, do your basic exposure and color edits first, then export a high-quality TIFF or JPEG for removal.

Step 2: Brush generously, including shadows and reflections

The most common mistake is brushing only the object. Every object casts something into the scene:

• its shadow on the ground,
• its reflection in water, glass or polished floors,
• a color spill on nearby surfaces, like the red glow of a jacket on a white wall.

If you leave the shadow behind, the removal is obvious. Brush a little past the edge of the object too, so the fill doesn’t inherit a halo of its outline.

Step 3: Work in pieces for large objects

For large removals, one giant stroke often produces a blurry or repetitive patch. Break it up:

1. Remove the parts surrounded by simple background first.
2. Then remove the parts that touch complex structure, like a doorway or a person you want to keep.
3. Re-run a fill on any area that looks wrong instead of painting over it by hand.

Step 4: Protect what you want to keep

If the object overlaps someone or something important, such as a stranger’s arm across your friend’s shoulder, brush only the unwanted part and stop short of the edge you’re keeping. It’s easier to clean a small leftover than to repair a damaged face or outline.

Step 5: Check lines, edges and perspective

Look for straight lines that should continue through the filled area: horizons, railings, curbs, window frames, tiled floors. Human eyes are very sensitive to a line that bends or breaks. If the fill gets a line wrong, try a smaller brush stroke along it, or fix it with a clone stamp afterwards.

Step 6: Zoom in and zoom out

At 100% zoom, check for smeared textures, repeated patterns, warped text and soft patches that don’t match the grain of the photo. Then view the whole image at screen size: some fills look fine up close but make the composition feel strangely empty.

Choosing photos that will work

Some removals are easy for any tool; others are hard for every tool.

Situation	Difficulty	Tip
Small object on sky, sand, grass or water	Easy	One stroke usually does it
Person in the distance on a plain street	Easy to medium	Include their shadow
Power lines across sky	Easy to medium	Brush each line in one smooth stroke
Person in front of detailed architecture	Medium to hard	Work in pieces; check straight lines
Large object covering most of the frame	Hard	The fill will be mostly invented; crop instead?
Object overlapping a face or hands	Hard	Brush only the object; refine manually
Text or signs you want to keep legible	Hard	AI may garble lettering nearby

Sometimes the honest answer is that cropping or a different frame is better than a removal. If you’re shooting now, take a few extra frames a moment apart: people move, and a combination of frames can give a clean result with no invented detail at all.

Removing crowds from travel photos

Crowds are a special case. Removing one or two people works well. Removing twenty usually produces a strangely deserted, smeary scene.

Two techniques work better:

• Multiple exposures with a tripod: shoot ten to twenty frames a few seconds apart, then use a median stack (Photoshop’s Statistics script with Median, or similar tools) to keep only what stays still. Moving people disappear because they’re in a different place in each frame.
• Selective removal: remove only the people closest to the camera or in front of the subject, and leave distant figures. A few people in the background make the scene look real.

Ethics: when removal is fine and when it isn’t

Removing a bin from a family portrait or a tourist from a vacation photo is harmless. A few situations deserve more care:

• Journalism, evidence and documentation: don’t remove anything. News organizations generally forbid adding or removing content from documentary photos.
• Property and product listings: removing clutter is fine; removing permanent features (a power pylon, a crack in a wall, a damaged part) misleads buyers.
• Historical and family photos: removing a person changes the record. If you do it, keep the original file alongside the edited one.

Removing objects from old photos

Unwanted objects aren’t limited to new pictures. In old prints, the “object” might be a photo corner mount, a date stamp, a stray finger over the lens, or someone’s handwriting across a sky. The same brush-and-fill approach works, combined with the dust and scratch cleanup covered in our guide to restoring old damaged photos.

Summary

• Pick the right tool: clone for structure, heal for small blemishes, AI fill for larger and complex removals.
• Brush generously, including shadows and reflections.
• Work large objects in pieces and check straight lines.
• Zoom in and out before you call it done.
• Keep the original, especially for anything that documents real events.

DirtAI is being built around exactly this workflow: brush over what you don’t want, let the model rebuild the background, and compare against the untouched original at any time.