A person remover does not uncover the real pixels that were hidden behind someone. It uses AI inpainting to generate a plausible continuation of the visible scene. The best results come from a complete selection, a background the model can infer, and a careful review of edges, texture, perspective, and light.
- Upload the highest-resolution original you have.
- Cover the complete person, loose hair, carried items, and contact shadow.
- Use one focused instruction, then generate one edit at a time.
- Inspect the repaired boundary at 100% zoom before downloading.
Before and after: remove one background person
In this travel-photo example, the person in the middle of the walkway pulls attention away from the foreground subject. The result continues the stone path, wall, and shoreline while keeping the woman, framing, and warm afternoon light in place.
BeforeAfter- AI generation requires an account; new accounts receive 5 credits once.
- Credits are based on rounded-up output megapixels, and each marked edit is a separate call.
- Start with one person and one mark so the result is easier to inspect and retry.
How to remove a person from a photo in four steps
1. Upload the original photo
Use the camera file when possible instead of a screenshot or a copy saved from a messaging app. More resolution gives the model more useful texture around the selected area and makes the final quality check easier.
2. Paint over the complete person
Brush over the head, body, hands, shoes, loose hair, bag, and any other detached detail that belongs to the person. Include the contact shadow when it would look unnatural on an empty path. Leave a narrow margin around the silhouette, but avoid painting over the main subject or an important straight edge.
3. Give the edit one clear instruction
Removal works best when the instruction describes the background that should continue. For this scene, a useful prompt is: “Remove the person and continue the stone path naturally.” If the first result leaves a fragment, adjust the selection before writing a longer prompt.
4. Compare before downloading
Move between the original and result several times. Check once at full size for seams and repeated texture, then check again at the final size you plan to publish. A repair can look fine as a thumbnail while bending a wall or duplicating stones when viewed at 100%.
A natural result starts with the mask
The mask tells the model which pixels may change. A tight mask that misses shoes, hair, or a bag can leave recognizable fragments. A mask that is too wide can erase useful background clues or take a bite out of the main subject. Cover the complete person plus a small, consistent margin.
The 30-second quality check
- Fragments: no hair, shoes, bags, limbs, or isolated color patches remain.
- Structure: walls, horizons, railings, and paving lines remain continuous.
- Texture: stone, water, grass, or fabric does not repeat in an obvious stamp.
- Light: brightness, color temperature, and shadow direction still match the scene.
- Main subject: face, body, clothing, and silhouette outside the mask are unchanged.
When person removal gets difficult
The person overlaps the main subject
Stop the mask at the true edge of the subject you want to keep. AI can create a plausible replacement, but it cannot know the exact face, logo, hand, or clothing detail hidden behind another person. If factual recovery matters, use another photo as a reference or choose a different crop.
Several people fill the background
Remove people one at a time, usually starting with the smallest or most isolated. InpaintingAI supports up to three marked edits, but each mark is processed as a separate generation and uses credits. Fewer edits make it easier to identify which selection caused a bent line or repeated texture.
The model replaces the person with someone else
This can happen when the selection hugs the original silhouette so closely that the model treats it as a request to redesign the person. Expand the mask enough to remove the outline cue, then describe the desired empty background rather than a new person. Retry from the untouched original instead of editing a failed result.
A shadow remains after the person is gone
Include a small contact shadow in the first selection when it clearly belongs to the removed person. For a long shadow crossing patterned ground, use a second focused pass so the model has intact texture on both sides of the shadow.
Frequently asked questions
Can I remove a person from a photo for free?
New InpaintingAI accounts receive 5 credits once. Generation uses credits based on output megapixels, so the number of free edits depends on image size and the number of marked areas.
Should I select the person and their shadow together?
Select a connected contact shadow with the person when leaving it would make the scene look wrong. Treat a long or complex shadow as a separate pass.
Can AI recover what was really behind the person?
No. It generates a plausible continuation from visible context. It cannot reveal factual pixels that the camera never captured.
How do I remove multiple people?
Work from the smallest, most isolated person toward the largest. Generate and inspect each removal before starting the next one.
Why does the repaired background look blurry or repeated?
The mask may be too large, too tight, or crossing a difficult repeated pattern. Retry from the original with a changed selection and a short background instruction.
Continue with a related removal guide
For masks, shadows, cables, product distractions, and other objects, read the broader object-removal workflow. If you want to understand the underlying edit, start with the homepage AI inpainting tool and its curated removal, repair, and replacement examples.