Blur faces in a video automatically. Detection runs at keyframes, faces are tracked between them, and you can step through the timeline to check every frame before exporting, all on your device, with nothing uploaded.
How it works
- Drop an image, video or PDF. Detection finds text or faces automatically.
- Review the regions: dismiss false positives, draw any the detector missed, pick solid fill, blur or pixelate.
- Download. Covered areas are baked into the pixels of the exported file.
Questions
- How are faces tracked between frames?
- Detection runs at keyframes and regions are interpolated between them, so a moving face stays covered without running the detector on every frame. Step keyframes with ←/→ and adjust or add a track wherever the detector missed.
- Does my file get uploaded?
- No. The models and all processing run inside your browser on your own device. You can verify this yourself: open your browser’s DevTools, watch the Network tab, and process a file: there are no uploads. After the first visit most tools also work fully offline.
- Is blur safe for sensitive information?
- Solid fill is the only fully irreversible effect, and it is the default. Blur and pixelate suit context you want obscured but not destroyed, so for account numbers and names, keep solid fill.
- Can I cover only specific words?
- Yes. After reading the text, type a keyword or pick a pattern (email, phone, card number) and only matching regions stay covered. The panel always shows how many regions are left visible.
- What happens to a redacted PDF?
- Every page is rasterized, so the output contains images only, no text layer, no fonts. Deleting a box cannot bring the words back because the words are no longer in the file.