Free, offline, local face-blur app for Windows and macOS. Drag a video in, get the same video back with every face automatically blurred. No upload, no cloud, no watermark, no signup.
Automatically detect and blur faces in MP4, MOV, MKV, AVI, WEBM and more — right on your own computer. Powered by the open-source deface face-detection engine, wrapped in a friendly drag-and-drop GUI so anyone can use it.
🌐 Looking for a hosted version, batch cloud processing, or a commercial license? Visit freefaceblur.com — the Pro version of this app.
▶ Watch the 60-second promo (MP4, 1080×1920) — built with Remotion. Source in promo/.
- Privacy-friendly — videos never leave your PC. No upload, no account, no tracking.
- Free forever — open source, MIT licensed, zero cost, no watermark.
- Automatic — no clicking on faces frame-by-frame. The AI finds them for you.
- Cross-platform — runs on Windows and macOS (Intel & Apple Silicon).
- Fast — GPU acceleration on Windows (DirectML for Intel / AMD / NVIDIA); Neural Engine on Mac.
- Batch friendly — drop in 50 videos at once, get 50 blurred videos out.
- Tunable — sliders for sensitivity and coverage; choose blur, mosaic/pixelate, or solid block.
- Audio preserved — keep-audio is on by default.
Great for: YouTube creators, TikTok / Reels / Shorts, journalists, researchers, GDPR / privacy compliance, bodycam footage, protest videos, classroom recordings, dashcam clips, real-estate walkthroughs, anything where you need to anonymize people.
Need it to scale beyond a single laptop? The Pro version at freefaceblur.com does the same job in the cloud — no install, no Python, just upload-and-go.
Download Python 3.10 or newer from python.org/downloads.
Important: during install, tick "Add Python to PATH" on the first screen. If you forget, just re-run the installer and choose Modify → Add to PATH.
- Click the green Code → Download ZIP button at the top of this page
- Right-click the ZIP → Extract All
(Or git clone https://github.com/Teylersf/blurfacefree.git if you use git.)
- Double-click
install.bat— installs the face-detection engine. Takes ~1 minute. - Double-click
Blur Faces.bat— opens the drag-and-drop window.
That's it. Drag a video onto the big drop zone, click Blur faces, find the result in the output\ folder.
Get Python 3.10+ from python.org/downloads (recommended — includes Tk).
If you prefer Homebrew: brew install python-tk.
Click Code → Download ZIP above, then double-click to extract. (Or git clone if you use git.)
Open the extracted folder in Finder.
- Double-click
install.command— installs everything. Takes ~1 minute. - Double-click
Blur Faces.command— opens the app.
The first time you double-click a .command file, macOS may show "cannot be opened because it is from an unidentified developer". That's expected — it's an unsigned shell script. Two ways past it:
- Easiest: Right-click the file → Open → click Open in the dialog. macOS remembers your choice.
- Or: System Settings → Privacy & Security → scroll down → click Open Anyway next to the blocked file.
Tk on Mac doesn't honor custom button colors the same way as Windows, so the UI will look more "native" (less Gen-Z). All features work identically.
Want zero setup? The Pro version at freefaceblur.com runs in your browser — drag a video in, download a blurred copy. No Python, no install, no terminal. Coming soon.
- Drag one or more videos onto the drop zone (or click Add files...).
- Pick a mode:
- blur — smooth Gaussian blur over each face (default)
- mosaic — pixelated / mosaic censor
- solid — solid black box
- none — just draw detection ellipses (debug)
- Adjust sliders if needed:
- Detection threshold — lower = more aggressive (catches more faces, more false positives)
- Mask size — bigger = more coverage around each face
- Click Blur faces. Watch the log + progress bar.
- Open the output folder to grab your anonymized videos.
.mp4 .mov .mkv .avi .webm .m4v .mpg .mpeg .wmv .flv
Output is always .mp4 H.264 by default (works everywhere — YouTube, Premiere, Resolve, browsers).
| Option | What it does | When to change |
|---|---|---|
| Mode | Blur, mosaic, solid black, or none | Mosaic = TV-news censor look. Solid = strongest anonymization. |
| Detection threshold | How confident the AI must be before calling something a face. Default 0.2 | Lower to 0.1 if it's missing small / angled / blurry faces. Raise to 0.3+ if random objects get blurred. |
| Mask size | How much the blur extends around the detected face. Default 1.3× | Raise to 1.6–2.0 if hair / ears / chin keep peeking through. |
| Keep audio | Copy the original audio track through unchanged. On by default. | Turn off only if you want a silent output. |
| Use rectangles | Use a box mask instead of an ellipse | Boxes look more "official redaction"; ellipses look more natural. |
| Mosaic block size | Pixel size of mosaic squares | Bigger = chunkier pixels (more censored look). |
| Output folder | Where the blurred videos are written | Anywhere you like. Defaults to .\output\. |
The GUI is a wrapper around the excellent deface CLI. After running the installer you can also call it directly:
Windows (PowerShell):
deface "C:\path\to\video.mp4" --keep-audiomacOS (Terminal):
python3 -m deface ~/Movies/video.mp4 --keep-audioLive webcam demo (both OSes):
deface camFull options: deface --help.
The repo also ships pre-made shortcut scripts:
| Action | Windows | macOS |
|---|---|---|
| Drag one or more videos onto the script | blur.bat |
blur.command |
Process every video in ./input/ → ./output/ |
blur-folder.bat |
blur-folder.command |
"Python is not recognized" when I run install.bat
Python isn't on your PATH. Re-run the Python installer from python.org and tick "Add Python to PATH" on the first screen, or use the Modify button to add it after the fact.
The GUI opens but the drop zone says "Drag-drop unavailable"
The tkinterdnd2 package didn't install correctly. In PowerShell:
python -m pip install --upgrade tkinterdnd2You can still use the Add files... button in the meantime.
It misses faces (small, blurry, side-profile, sunglasses)
Lower the detection threshold to 0.10 and raise the mask size to 1.7–2.0. If you still see issues, switch to mosaic mode with a larger block size — pixelation forgives small detection errors.
Random non-face things are getting blurred
Raise the detection threshold to 0.30–0.40.
It's slow on my machine
- Make sure GPU acceleration installed (
install.batdoes this automatically — look foronnxruntime-directmlin the log). - Try lowering the resolution the detector sees: command-line option
--scale 640x360makes detection much faster with minimal accuracy loss on HD/4K videos. - Close other GPU-heavy apps.
The first run is downloading something — is that OK?
Yes. On the very first run, deface downloads a ~2 MB face-detection model (CenterFace). After that, everything is 100% offline.
macOS: "cannot be opened because it is from an unidentified developer"
That's Gatekeeper blocking an unsigned shell script. Right-click install.command (or Blur Faces.command) → Open → click Open in the dialog. macOS will remember the choice. See the First-time Mac launch section above for details.
macOS: "command not found: python3"
Python 3 isn't installed. Get it from python.org/downloads (recommended — comes with Tk preinstalled), or via Homebrew: brew install python python-tk.
I need to process hundreds of videos / want a cloud version / need a commercial license
That's exactly what the Pro version at freefaceblur.com is for — no install, no Python, no terminal, scales beyond a single laptop.
- The app launches the deface command on each video you drop in.
- deface runs the CenterFace neural network (a tiny ~2 MB ONNX model) on every frame.
- Each detected face region gets blurred / mosaiced / blacked-out.
- Frames are re-encoded with ffmpeg, audio is muxed back in, and you get a new MP4.
No data leaves your machine. No telemetry. No accounts.
| This repo (Free) | Pro — freefaceblur.com | |
|---|---|---|
| Price | Free, MIT licensed | Paid plans |
| Where it runs | Your own PC / Mac | Cloud — runs in your browser |
| Install | Python + this repo | None — just visit the site |
| Speed | Limited by your hardware | Scaled cloud GPUs |
| Batch size | Whatever your disk fits | Unlimited |
| Commercial license | Use as-is under MIT | Included |
| Support | GitHub issues | Priority email + Slack |
| Best for | Hobbyists, devs, privacy-first users | Agencies, creators, businesses |
👉 Get the Pro version at freefaceblur.com
This project is a friendly GUI wrapper. All the heavy lifting is done by these excellent open-source projects:
- deface by ORB-HD — the actual face-detection-and-anonymization engine. Please star their repo too.
- CenterFace — the underlying face-detection model.
- ONNX Runtime — fast neural-network inference.
- ffmpeg — video encoding/decoding (via imageio-ffmpeg).
- tkinterdnd2 — drag-and-drop in Tkinter.
If you use this for anything serious, please support those upstream projects.
MIT — do whatever you want with it, just don't sue me. The underlying deface is also MIT licensed.
- 🌐 Pro / cloud version: freefaceblur.com
- 📦 This repo: github.com/Teylersf/blurfacefree
- 🐛 Issues / feature requests: github.com/Teylersf/blurfacefree/issues
- ⬇️ Releases: github.com/Teylersf/blurfacefree/releases
Free face blur for video. Blur faces in video Windows. Blur faces in video Mac. Automatic face blur. Anonymize video faces. Hide faces in video free. Face blur software open source. Pixelate faces in video. Censor faces in video. GDPR video anonymization. Privacy face blur tool. Offline face blur. Face blur AI free. YouTube face blur free. TikTok face blur free. Video redaction tool free. Best free face blur software. Blur faces without watermark. Local face blur app. Drag and drop face blur. Free face blur Mac. macOS face blur. Apple Silicon face blur. Cloud face blur — see freefaceblur.com.