GUIDE
How to remove people from a video (automatically)
You filmed the perfect shot — and forty strangers walked through it. Until recently your options were painful: reshoot at 6 a.m., mask and track every person by hand, or crop until the problem left the frame. AI clean-plate generation replaces all of that with one upload.
The three ways to do it, compared
1. Frame-by-frame editing (After Effects, Mocha)
Track each person, mask them, content-aware fill the hole, fix the seams. Professional quality, but hours of work per shot and a paid suite. Worth it when a client is paying for pixel-perfect.
2. Mobile object erasers (per-object tap tools)
Fine for one photobomber crossing behind you. They fall apart on crowds — you tap objects one at a time, and the inpainting smears when subjects overlap or the camera moves.
3. AI clean plate (this site)
The LTX-2.3 Clean Plate model regenerates the whole shot without its moving subjects — every person at once, no masks, moving camera supported. Minutes instead of hours. The trade-off: output is an AI reconstruction, so tiny details like distant sign text can shift.
Step by step
- Open the generator and drop in your clip (MP4/MOV/WebM, ≤80 MB).
- Optionally type extra objects to remove — “bicycle, stroller” — people and vehicles are handled automatically.
- Hit Remove people. A 5-second clip takes about 2–3 minutes.
- Preview the result side by side and download the MP4.
What footage works best
The model reconstructs background it can see. Best case: people walking through a scene — streets, plazas, beaches, lobbies, landmarks — where the background is visible around them for at least part of the clip. Hardest case: one subject filling the frame the whole time (a close-up interview), because there is nothing behind them to rebuild. For that, use mask-based inpainting instead.