Blur Image
Blur a whole photo, or blur, pixelate or black out only the areas you select: faces, number plates, names.
Loading tool…
How blurring works, and when it is safe
A blur replaces each pixel with a weighted average of its neighbours. A Gaussian blur, which this tool uses, weights those neighbours by a bell curve, which is why it looks smooth and natural rather than smeared. The strength controls how far the averaging reaches: a small value softens, a large one destroys detail entirely. Pixelation instead replaces each block of pixels with a single colour, and a solid box replaces the area with black.
That distinction matters when the effect is used to hide something. A light blur is reversible in principle: the original detail is still statistically present, and deblurring techniques have recovered faces and text from weakly blurred images more than once. Light pixelation has the same weakness. A solid box removes the information outright, which is why it is the safe choice for card numbers, addresses and anything else that must not be read. You can apply any of the three to the whole picture or only to the areas you drag over. “Find faces” marks every face it can find for you, using a small face detector that runs on your device; remove any box with its ×, or drag to add one it missed.
Why we built this tool
Two needs meet here: the aesthetic one, where a blurred image becomes a soft background for a design or for text, and the privacy one, where a face, a number plate, a name or a screen inside a picture must become unreadable before it is published.
In the privacy case, uploading the unblurred image to a website is precisely the wrong move — the face you are trying to protect is what you just transmitted. Ours applies the blur on a canvas in your browser and exports the flattened result locally, so the original never leaves your device and the exported file contains no recoverable layer beneath the blur. No upload, no size limit, no account.
Tips and common mistakes
- To hide the people in a photo, press “Find faces”, then check every face is boxed: faces that are very small, turned away or partly hidden can be missed.
- For privacy, use a solid box, or pixelate with a large block size; light blurring and pixelation have both been reversed in published research.
- Faces, licence plates, screens, house numbers, and name badges are the usual things people forget to check.
- Choose “Only areas I select” and drag over each thing to hide; you can mark as many areas as you need, with a finger on a phone.
- Export, then reopen the exported file to confirm nothing is legible.
AI on this page
- Model: UltraFace (version RFB-320), a small face-detection model
- What it does here: Used by Find faces: it marks a box over every face it finds, and those boxes are blurred, pixelated or blacked out like areas you draw yourself.
- Download: 1.3 MB model plus the ONNX Runtime Web engine (about 2.5 MB), about 4 MB in all, downloaded once.
- Licence: MIT (model and ONNX Runtime Web).
- Privacy: Nothing is downloaded until you press Find faces. The model runs on your device; the photo is not uploaded.
- Limits and speed: Small, blurred or turned-away faces can be missed, so check every face before you share the picture and add any it missed by hand. It found all three faces in our group-photo test; one picture takes well under a second once loaded.
Quick start: using Blur Image
- Load the image.
- Choose where: the whole image, or only areas you select (then drag over each one).
- Choose the effect: blur, pixelate, or a solid black box.
- Set the strength (block size for pixelate); go heavy if the purpose is to hide identifiable detail.
- Download the flattened PNG and share only that version.
Photos also carry hidden location data — check it with Extract EXIF, and strip it with Remove EXIF.
Two reasons to blur
- Privacy: obscure faces, plates, addresses, screens, and documents before sharing a photo publicly.
- Aesthetics: soften a distracting background to draw the eye to the subject, mimicking the shallow depth of field of a good camera.
Blur for privacy — do it properly
If you are hiding sensitive information, a weak blur is dangerous. A light blur can sometimes be partially reversed or simply read through. To be safe:
- Blur strongly — err on the side of too much. Text especially needs a heavy blur to be truly unreadable.
- Cover the whole sensitive area with margin to spare, not just the exact characters.
- Flatten and re-save the image so the blur is baked into the pixels, not stored as a removable layer.
For truly sensitive data, use the Solid black box effect: a heavy blur is good, an opaque box is certain.
Blur for background — do it gently
For the aesthetic use, the opposite applies. A subtle blur on the background suggests depth without looking artificial. Keep the subject perfectly sharp and let the blur increase toward the edges for a natural effect.
Common uses
- Anonymize faces in event or street photos.
- Hide plates, addresses, and screens.
- Redact documents shared as images.
- Soften backgrounds for portraits.
Further reading
Screenshots That Leak: What to Hide Before You Post One →