Image Redactor — Blur / Pixelate

About Image Redactor — Blur / Pixelate

Image Redactor (Blur / Pixelate) is a free, browser-based tool for permanently hiding faces, license plates, emails, and other sensitive details in a photo before you share it. You drag rectangles over the parts you want to obscure, choose Pixelate, Blur, or solid Black, and export a flattened redacted.png. It can also auto-find faces on-device using TensorFlow.js with the BlazeFace model (self-hosted, ~465 KB), so no image data is ever uploaded — everything runs locally in your browser on a 2D canvas with no signup.

How to use Image Redactor — Blur / Pixelate

1. Drop an image onto the upload area, or click 'Choose image' (PNG, JPEG, or WebP; animated GIFs use only the first frame).
2. Pick a redaction mode with the segmented control: Pixelate (mosaic), Blur, or Black (solid fill).
3. Drag a rectangle on the canvas over each area you want to hide — every box is listed on the right and can be deleted individually.
4. Click 'Detect faces' to auto-add padded boxes over any faces the on-device model finds, then adjust as needed.
5. For Pixelate or Blur, use the strength slider (2–40) to set the mosaic block size or blur radius; Black has no strength.
6. Drag the before/after divider to confirm your regions fully cover the sensitive areas.
7. Use 'Undo last' or 'Clear all' to fix mistakes, then click 'Export PNG' to download redacted.png.

Frequently asked questions

Are my images uploaded to a server?

No. The image is decoded, redacted, and exported entirely in your browser on a local 2D canvas — it never leaves your device. Even the face-detection model is self-hosted and runs on-device, with no cloud calls.

Is it free and do I need an account?

Yes, it's completely free with no signup, account, or watermark.

Is the redaction permanent, or can it be undone?

It's permanent. The tool flattens your blur, pixelate, or black-out regions into the exported PNG's pixels, so there's no hidden original layer to recover — unlike just drawing a box on top in some editors.

What's the difference between Pixelate, Blur, and Black?

Pixelate downscales the region into a coarse mosaic (block size set by the slider), Blur applies a Gaussian-style blur at the chosen radius, and Black fills the region with solid #000. For sensitive data, Black or a strong mosaic is the safest because faint blurs can sometimes be partially reversed.

How does 'Detect faces' work and what if it misses some?

It loads a TensorFlow.js BlazeFace model on first use (cached afterward) and draws padded boxes over detected faces. If it finds none or misses a face, you'll see a notice and can still drag boxes manually; the first model load needs a working connection.

What image formats and sizes are supported?

PNG, JPEG, and WebP static images; animated GIFs are reduced to their first frame. Output is always PNG (redacted.png), and very large files or extremely high-megapixel images are rejected before processing to avoid freezing the tab.

People also search for

Image Redactor — Blur / Pixelate is also known as redact blur image, blur faces in a photo, pixelate part of an image, censor a photo, blur a license plate, black out personal info in image, hide sensitive info in a picture.