Blur Faces in a Video for Privacy — Free, No Upload
Sharing a video that captures bystanders, kids, or sensitive screens? This guide blurs specific regions frame by frame — without sending the footage to any server.
Step-by-step
Strip metadata first
Run the file through the Metadata Cleaner to remove GPS, device model, and creation timestamps. Visual blur alone leaves location data leaking.
Mark regions to blur
Open Face Blur, draw rectangles over faces, screens, or plates. The tool tracks slow-moving subjects automatically.
Choose blur strength
Medium-strong Gaussian blur is irreversible for typical resolutions. Avoid pixelation — it can be partially reversed.
Recommended settings
| Blur type | Gaussian (irreversible at strength ≥ 30) |
|---|---|
| Tracking | Manual region per scene; auto-extrapolation between keyframes |
| Use cases | Newsroom, doxxing prevention, GDPR compliance |
Check before you publish
- Step through the clip frame by frame at every cut and every fast movement. A blur that tracks well for most of a scene and slips for four frames has published the face.
- Check reflections, name badges, licence plates, screens and tattoos. A blurred face next to a legible badge identifies the person anyway.
- Confirm you exported the blurred copy and not the source, then strip the metadata before it goes out.
Tools you may also need
FAQ
Why not pixelate?
Pixelation can sometimes be partially reversed by AI super-resolution. Gaussian blur at strength ≥ 30 is much harder to invert.
Will the blur survive social media re-encoding?
Yes — the blur is baked into the pixels before export. Subsequent compression preserves the blurred result.
Does this run in my browser?
Yes — every step in this guide uses an in-browser FFmpeg WebAssembly tool. Your video never uploads to a server and never leaves your device.