Remove People from Photo
Erase tourists, passers-by, and background crowds while your subject stays sharp. Free online AI background people remover — 3 removes a day without signup, 5 after you sign in.
Before / After
Erase unwanted objects and clutter
Why people in the background ruin a shot
A landmark at sunrise with one tourist in the corner. A wedding candid where the venue's catering staff happens to walk through. The family beach day framed by someone else's swimwear silhouette. Background people are the single most common complaint about travel and event photography — they are rarely anyone's fault, and you almost never notice them in the moment. But back home looking at the picture, they are all you can see.
Cropping rarely fixes it, because the crowd usually sits just inside the composition you actually want. Smartphone portrait modes help a little by blurring the background but do not erase anyone. Manual cloning in Photoshop works but takes 10 minutes a photo and a lot of patience. A purpose-built AI background people remover is faster than all of those — paint, wait, done.
MagicRemover runs SDXL inpainting under the hood for this page specifically. SDXL is generative — it does not just infer from surrounding pixels, it invents plausible continuation of the scene guided by a prompt that says "clean background, no people". That makes it especially strong on large, complex regions like tightly packed crowds or figures against busy backdrops, where the simpler inpainters would struggle. Unlike the older SD 1.5 inpainting, SDXL accepts any input aspect ratio and returns the result at the same dimensions you uploaded.
How to remove people in 3 steps
- 1
Upload your photo
Drag your picture into the upload card or tap to browse. JPG and PNG are fine; anything wider than 1536 pixels is resized automatically before it hits the AI. Travel shots, wedding shots, concert photos, street scenes — anything with unwanted people in it works here. Group shots where the subject is centre-frame and the crowd is in the background get the cleanest results.
- 2
Brush every unwanted person
Paint over each photobomber, tourist, or pedestrian — full body, head, limbs, everything you want gone. For a tight crowd, a larger brush (80–120 px) covers multiple people at once; isolate individuals with a smaller brush (40–60 px). Always add 6–10 px of margin around silhouettes so the AI has context to rebuild from. Undo and Clear are one tap away on the right panel.
- 3
Download the clean scene
Tap Remove objects. The SDXL inpainting model takes roughly 20–30 seconds to redraw the masked region with a plausible continuation of the background — sky, water, sidewalk, wall, foliage, whatever was behind the people. Hit Download to save as PNG at the same dimensions as your original upload — SDXL preserves the aspect ratio and resolution natively.
Where this tool earns its keep
Travel and landmark shots
The Eiffel Tower, the Grand Canyon, a temple at sunrise — scenes with permanent tourist traffic you cannot dodge. Brush the crowd, keep the landmark.
Street-photography cleanup
A framed composition ruined by one photobomber in the corner. Paint them, the background (road, building) fills in cleanly.
Venue and concert photos
Wide shots where the audience or venue staff clutter the frame behind the performer or guest of honour.
Family snaps with bystanders
A kids' birthday at the beach with strangers in swimwear in the background; a hiking photo with other trekkers in the distance.
Real estate and property shots
Listing photos where passers-by or neighbours wander into the frame. Clean shots rank better on listing sites.
Wedding ceremony reference photos
The official photographer handles the main shots; your candid reference pics often have catering staff or other guests in the wrong place. Clean them for a family album.
Tips for clean edits
Brush generously around heads and shadows
Human silhouettes are harder to erase cleanly than objects because the AI has to reconstruct context-sensitive backgrounds. Leave a pixel or two of margin — especially around hair, shadows cast on the ground, and reflections on glossy surfaces — so the model has enough to work with.
Works best when the background is uniform-ish
Sky, ocean, grass, pavement, brick walls, carpets, woodgrain — these all rebuild beautifully. Dense foliage, patterned fabric, stepped stairs, and anything highly structured can leave soft seams. For the latter, a quick manual touch-up in any image editor finishes the job.
Crowds are fine — just paint them as one region
You do not need to brush each person separately. For a packed sidewalk or concert crowd, paint the whole block in one stroke. The model handles large contiguous masks better than many small ones — fewer seams to blend.
Stand-alone subjects get the cleanest result
When your subject is clearly in the foreground and the people to remove are in the background, the AI has the easiest job. When the subject is intertwined with the person you are removing (shared handrail, overlapping limbs), the reconstruction can bleed into the subject. Paint carefully around the overlap zone.
The honest limits
Very large uploads get internally downscaled. SDXL handles any aspect ratio, and the result comes back at the dimensions you uploaded. Internally the model is trained around the 1-megapixel mark, so 6000 × 4000 images get processed at a lower resolution and resampled back up on the way out. Detail in the masked region can soften a touch on very large photos. For most phone uploads (under ~12 MP after the client-side 1536-px downscale MagicRemover applies) this is imperceptible.
Unmasked regions may drift slightly. SDXL inpainting is generative. Pixels outside your mask stay visually close to the original but are not byte-identical — colours and fine detail may shift a hair. The tradeoff buys you the ability to erase large, complex people-heavy regions that purely reconstructive inpainters (LaMa, ContentAware) cannot handle. For strict pixel preservation outside the mask, use the Watermark Remover or homepage Object Remover instead.
Weird edges in mixed scenes. When the person you are removing is physically intertwined with the subject (shared handrail, overlapping limbs, leaning on a wall right next to your subject), the AI can blend parts of the subject into the fill. The best defence is to paint carefully around the overlap — err on the side of including small bits of background rather than accidentally covering your subject.
Questions about removing people from photos
Other removers on MagicRemover
Free AI Object Remover
The general-purpose eraser — faster and preserves the original resolution. Good for single objects and watermarks on small regions.
Remove Watermark from Photo
LaMa-powered watermark remover, tuned for long diagonal © lines and corner stamps on stock photos.
Gemini Watermark Remover
Specialised page for the sparkle mark on Google Gemini AI-generated images.