Video Prep: crop, find cuts, cover a logo, trim and mute in your browser
Clean a clip before it goes into an AI video model: cut the black bars or crop to 9:16 or 4:3, snap the trim to the shot cuts, cover a corner logo, drop the sound – 5–15 seconds, MP4, never uploaded.
Pay per clip, from 10 credits · No watermark on paid videos
🔥 Trending AI effects
All effects →More tools
- Abrazo IA: video de un abrazo con tus fotos
- AI Crying Video Generator
- AI Halloween Costume Video
- AI Hug Video Generator
- AI Kiss Video Generator
- AI Muscle Generator
- AI Video Generator with Audio
- AI Werewolf Transformation
- AI Zombie Trend Generator
- Beso IA: video de beso con tus fotos
- Bulin 47 AI Video
- Bulin 47 IA
How it works
Step 1
Choose the clip
Pick an MP4, MOV or WebM. It opens in the player with the crop box drawn over it.
Step 2
Crop and cover
Tap “Detect black bars”, then an aspect (9:16, 4:5, 1:1, 4:3, 16:9) for a centred crop inside the picture, or type pixels per edge. A corner logo on a plain background: “Add box”, set its position and the seconds it is visible.
Step 3
Find cuts, trim, export
“Find cuts” lists every shot change; tap one to set the start, “15 s to the last cut” ends the clip on a cut. Keep “Remove sound” on, Export, Download the MP4.
Why clean a clip first
Reference-video and recast models copy everything in the clip you give them except the people: black bars, a corner logo, burnt-in captions and the original soundtrack all end up in every video made from it. Cropping the bars and the watermark, keeping a single shot of 5–15 seconds and dropping the audio gives the model a clean template. The crop is a straight rectangle – it keeps the full scene inside it rather than cutting out a person – so the look of the original stays intact.
What the export does
The clip is cut to your start and end points, cropped to the rectangle you set, scaled down only if it is taller than 1080p, and re-encoded as H.264 MP4 at 30 fps with the “faststart” flag, so it plays and uploads anywhere. With “Remove sound” on the audio track is dropped; off, it is kept as AAC. Dimensions are rounded to even numbers, which most video models require.
Finding the right 15 seconds
Reference-video models want a single shot of at most 15 seconds and a clip that starts and ends on a cut, so the motion is complete. “Find cuts” runs a scene-change scan on a small copy of the clip (a 40-second 1080p clip takes about half a minute on a phone) and lists the cut times. Tap a cut to start there, then “15 s to the last cut” to end on the last cut inside the limit – or tap another cut with “end” selected. The 4:3 centre crop keeps both people and the middle of a wide 16:9 performance clip; 9:16 is for vertical clips that are already framed for phones.
Limits and tips
Everything runs in your browser with ffmpeg.wasm, so long or very large clips take longer – a 15-second 1080p clip exports in roughly a minute on a recent phone. Keep source clips under about 60 seconds and 100 MB. If the browser cannot open the file, convert it to MP4 first; HEVC/H.265 videos from some phones do not play in every browser. Black-bar detection samples three frames and looks for near-black rows and columns; if the bars are dark grey or have a logo in them, set the crop by hand.
Frequently asked questions
Is my video uploaded?
No. The clip is read, cropped and encoded inside your browser; nothing is sent to a server.
Why 5–15 seconds?
Reference-video models that keep a clip's moves and swap the people accept clips of about 5–30 seconds and want a single shot no longer than 15 seconds. The length hint under the export button turns amber outside that range – it is only a hint, you can export any length.
Can it remove a watermark in the middle of the picture?
No. It crops a rectangle, so it only removes logos and captions at the edges. A watermark over the picture stays – use a different clip.
What format do I get?
An MP4 file with H.264 video (and AAC audio if you keep the sound), which every AI video model and social app accepts.
Does it work on a phone?
Yes, in Safari and Chrome on recent phones. The first export downloads the ~30 MB encoder; after that it is cached by the browser.
Ask AI about Video Prep: crop, find cuts, cover a logo, trim and mute in your browser
Guides
Hotel Lobby Video for Your Anniversary or Couple (in English or Spanish)
Turn the Hotel Lobby trend into a gift: make an anniversary or couple video from two photos, choose a beat, send it on WhatsApp – in English or Spanish.
Zombie Love Story AI Trend: What It Is and How to Make One
The zombie AI love story trend explained: the four story beats, how to make one from two photos, photo tips, prompt ideas and how it differs from a zombie filter.
Raindance AI Trend: What It Is and How to Make the Pier Video
The Raindance AI trend explained: the Dave & Tems pier scene, how to recreate it from two photos in one step, the longer pro workflow, photo tips and fair use.
SnapGen AI Video Generator: What It Is and Alternatives
What the SnapGen AI video generator is, how its Veo 3.1 claim compares, the different SnapGen products with similar names, and how to run Veo 3.1 clip by clip.
Dola AI Video Generator: Seedance 2.5, Limits and the US Block
What the Dola AI video generator is (ByteDance's former Cici app), which model it uses, what its free plan covers, why it isn't available in the US, and how to use Seedance 2.5 anyway.
AI Crying Video: How to Make a Photo Cry (Crying Effect Guide)
How the AI crying video effect works, how to make one from a selfie or pet photo in a few minutes, photo tips for better tears, and how to post it fairly.
Make your first AI video
Pick a trend, add your photos, pay once per video – no subscription.
Generate a video