Remove Objects and People from a Video with AI
Name what should go: a passer-by, a parked car, a boom mic, a sticker. The AI rebuilds the background behind it across the clip and follows your camera move, so you get a clean version without masking or rotoscoping. Your original stays as it was.
Powered by leading AI video models
Edit with the leading video models that read a clip as a reference, with no API keys, no setup and no switching tools.
An AI video remover built for real footage
Removing something from a moving shot used to mean masking it out second by second and painting in what was behind it. The AI does that rebuilding for you, so the fix follows the camera instead of breaking when it moves.
Models that read video
Removal needs a model that can take your clip as a reference. Only those models appear here, so every option you pick can work from your footage.
Models that edit video
Seedance 2.5
Takes clips up to 30 seconds as references, with image and audio references in the same edit.
MiniMax H3
Edits clips up to 15 seconds and outputs at up to 2K resolution.
No masks, no rotoscoping
Type what should go in plain words. The AI finds it, removes it across the clip and fills the space with what belongs behind it.
Original video
Prompt
Edited video
One clip, several clean-ups
Clear the laundry, the toys or the whole coffee table from the same room tour. Each version saves as a new video next to your original.
Swap an object for one from a photo
Sometimes removing is only half the job. Attach a photo of the object you want and describe the swap: the AI takes the old one out and builds the new one into the shot, matching the light and the angle. Here, a plain white kettle became the copper kettle from a product photo. One generation can combine up to three clips and up to three audio tracks with your images.
What you can remove from a video
Most cleanup jobs fall into a few groups. Each of these started as an ordinary clip and a one-line prompt.
For creators, filmmakers and marketing teams
Save takes you thought you'd have to reshoot
A great take with one distraction in it doesn't have to end up in the bin.
Keep the best take
A stranger in the background or a mic in frame no longer means using a weaker take. Remove the distraction and keep the performance you liked.
Clean locations without closing them
Shoot on a busy street, in a gallery or at a viewpoint, then clear the crowd afterwards instead of waiting for an empty moment.
Ready-to-use product footage
Remove cups, cables, stickers and clutter from product clips so they're ready for ads, listings and social posts without a reshoot.
How to remove an object from a video
Three steps from the clip you have to a clean version.
1. Open your video
Open Create video and click a video you've made, or upload your clip. It opens with the clip attached as a reference, ready for the edit.
2. Name what should go
Type what to remove: "remove the trash can next to the bench," "take out the parked cars." If it helps, say what should fill the space.
3. Pick a model and generate
Choose from the available models that support videos as references and generate. The clean version arrives as a new clip, and your original stays as it was.
Remove people
Remove people from a video
Guests walking behind a first dance, visitors crossing a gallery, a cyclist cutting through a travel shot. Name who should go and the AI removes them across the clip, rebuilding the scene behind them while the people you want stay in the shot.
It works best when the people you remove are clearly separate from the ones you keep. If someone overlaps your subject for most of the clip, describe exactly who should go.
Useful for:
- Wedding and event videos
- Travel and location footage
- Interviews filmed in public places
- Real estate and venue tours.
Background cleanup
Remove clutter from the background of a video
Messy shelves, cables, laundry and boxes behind a talking head can make a good video look rushed. Describe what should go and the AI clears it while following your gestures, expressions and timing.
You can be specific ("remove the boxes and cables from the shelf") or broad ("tidy the shelf behind her"). Specific prompts give you more control over what stays.
Background cleanup helps:
- Creators recording at home
- Course and coaching videos
- Remote interviews and webinars
- Product demos filmed in real rooms.
Landscapes and locations
Remove vehicles and objects from scenic shots
A camper van at a lookout, a boat on a still lake, a crane over an old town: one object can pull attention from the whole view. Remove it and the AI fills in the landscape behind it as the camera moves.
Scenic cleanups work well for:
- Travel and tourism videos
- Drone and aerial footage
- Hotel and property marketing
- B-roll for campaigns and stock libraries.
Large objects close to the camera are the hardest to remove, because more of the scene has to be rebuilt behind them. Smaller or more distant objects usually give the cleanest result.
Example prompts to try
Say what should go, in plain words. Each tab shows one prompt and the clean clip it made. Copy it, change the details, and run it on your own footage.
Explore more AI video tools
Frequently Asked Questions
Add the clip as a video reference in Create video, type what should go, and generate. The AI removes it and rebuilds what's behind it across the clip. It's used by e-commerce teams cleaning up product clips, marketers fixing campaign footage, and photographers who now shoot video and want cleaner locations.
You don't need masking or compositing skills. If you can describe the object in a sentence, you can remove it.
Yes. Describe the person by what they look like or where they are ("the man in the red jacket walking behind her") and the AI removes them across the clip. Results are cleanest when the person you remove doesn't overlap your main subject for long.
Remove only people you don't need in the shot, and edit footage you own or have the rights to use.
Each attached clip can be up to 15 seconds, or 30 seconds on some models. For longer footage, remove the object section by section. You can attach up to three clips per generation, plus images and audio, which helps when a second angle or a product photo should guide the result.
The edit is a new video generated from your clip, so it follows your shot, motion and timing closely rather than copying every pixel. Small details like text, reflections and fine textures are worth checking before you publish.
Your original stays in your library as it was, so you can rerun the removal with a more specific prompt and compare versions.
Yes. For stills, use inpainting to erase or replace part of an image with a prompt. It works the same way as removing objects from video: describe what should go and the AI rebuilds the area.
Keep the take. Lose the distraction.
Open your clip, name what should go, and get a clean version that follows your original shot. Your original stays as it was, so you can try again with a more specific prompt.