Remove crowds from photos
Get rid of the dense crowd of strangers in front of the monument, the audience filling the stadium seats, the shoppers in the market aisle, the parade-day spectators on the sidewalk. Magic Eraser brushes out the crowd and rebuilds the ground, plaza, or backdrop underneath — even for groups of dozens to hundreds of people. Free on web, iOS, and Android.
Last updated
Essayer maintenantWhy dense crowds are different from a single stranger avec Magic Eraser
Removing a single stranger from a tourist photo is a clean problem — one subject, one background to rebuild, well-defined edges. Dense crowds are different: the subjects overlap each other, the shadows compound, the ground or floor is mostly hidden, and the eye reads the crowd as a textured pattern rather than as individual people. Manual editing here is impractical because you'd be cloning hundreds of patches and reasoning about hundreds of shadows. Magic Eraser handles the entire crowd as a region — brush the whole mass, and the AI rebuilds the plaza, stadium floor, market aisle, or sidewalk underneath using the visible margins as reference. Works on small clusters of a dozen people and on filled sports stadiums alike.
Instructions étape par étape
- 1
Upload your photo
Open Magic Eraser on the web, iOS, or Android and drop in the concert, stadium, market, parade, public-square, or event-venue photo. JPEG, PNG, HEIC, and WebP are all supported.
- 2
Brush over the crowd as a region
Paint over the entire crowd mass — don't try to outline individual people. A generous brush over the whole group gives the AI a clear region to rebuild and is faster than tracing edges. Include shadows the crowd casts on the ground or floor in your brush region; leaving shadows behind makes the cleanup obvious. For partial crowds (foreground only, leaving the back of the venue intact), brush only the foreground mass.
- 3
Tap Erase and inspect
Magic Eraser rebuilds the floor, plaza, stadium ground, or market backdrop in seconds. Scan at 100% zoom for any residual silhouettes, run a second pass on any leftover artifact, then export at full resolution for the portfolio, social post, editorial, or commercial use.
Idéal pour
- Travel photos at famous monuments (Eiffel Tower, Taj Mahal, Colosseum) where the crowd ruins the architectural shot
- Concert and music-festival photos where you want the stage, lighting rig, or venue interior visible
- Sports-stadium photos where the empty-venue aesthetic matters (architectural portfolios, real-estate listings for sports complexes)
- Market, bazaar, and souk photos where the aisle, stall design, or goods displays are the subject
- Parade and street-festival photos where the floats, performers, or route itself is the subject
- Public-square photos (Tiananmen, Trafalgar, Times Square) where the architecture is the subject
- Wedding venue exteriors during the reception when crowds fill the courtyard or terrace
- Trade-show booth photos where you want the booth design clean without other attendees
Notes importantes
Crowd removal works best when the crowd is in front of a relatively continuous background — a plaza floor, a stadium pitch, a market aisle, a building façade. Two scene types are harder. First: when the crowd is layered against deep-perspective architecture (rows of pillars receding into the distance, the interior of a long market hall), the AI may flatten the perspective when rebuilding; run smaller passes for each depth plane (foreground crowd first, then mid-ground, then back-row figures) for cleaner spatial cues. Second: when the crowd is the foreground subject of a candid composition (street photography where the crowd is the point), erasing them leaves an awkward empty scene — the right answer is usually a re-shoot at a different time rather than a crowd erase. For high-stakes commercial, editorial, and journalism use, disclose AI-cleaned crowd photos per the relevant industry standard — and be careful with news and documentary contexts where removing people materially changes the truth of the scene.
Questions fréquentes
- Is it free to remove a crowd from a photo?
- Yes. Magic Eraser's free tier covers crowd removal with daily usage limits. Upgrading to Premium ($29.99/year) removes the limits and unlocks higher-resolution exports — useful for printed editorial, architectural portfolios, and commercial licensing.
- Can I remove a stadium crowd of thousands of people?
- Yes. Magic Eraser handles dense crowds at scale — the model treats the crowd as a textured region rather than as individual subjects. Brush over the entire seating bowl and the AI rebuilds the seats underneath. For very large crowds, run two or three passes covering different sections (lower bowl, upper deck, end zones) for cleaner results than one giant masked area.
- What about the crowd's shadows on the ground?
- Include the shadows in your brush region. Leaving shadows behind is the single most common giveaway that a crowd was erased — the eye reads the shadows even when the people are gone. Brush a generous margin around each shadow so the AI has clean reference for the ground texture without the shadow contamination.
- Will the cleaned plaza, stadium, or street look obviously edited?
- On most photos, no — the AI matches the ground texture, lighting, and perspective of the visible margins. The places it can show are scenes with high-detail ground patterns (cobblestones at a specific angle, painted yard lines on a sports field, market-stall floor tiles in a regular pattern) where the rebuilt area may slightly soften the pattern; a touch-up pass with a smaller brush usually resolves this.
- Should I disclose the edit for editorial or commercial use?
- For news and documentary photography, removing crowds changes the meaning of the scene — most editorial guidelines (AP, Reuters, NPPA Code of Ethics) prohibit material alterations that would mislead the viewer. For travel, real-estate, architectural, and product photography, AI-cleaned crowd photos are widely accepted with disclosure per the relevant industry standard (FTC for ad copy, NAR Code of Ethics Article 12 for real-estate uses).