Pixelate or Blur Image
Hide part of a photo — a face, a number plate, an address, a name or a signature — by pixelating or blurring it. Drag a box over what to hide, pick the effect and strength, apply to as many areas as you like, then download. Everything runs in your browser; nothing is uploaded.
Drag an image here or click to browse
JPG · PNG · WebP · GIF · BMP · AVIF — stays on your device
How to pixelate or blur part of a photo
- Add your image — drag it onto the box above or click to browse. It is decoded in your browser; nothing is uploaded.
- Draw the box over the area you want to hide — drag the middle to move it, drag a corner or edge handle to resize.
- Pick the effect (Pixelate or Blur) and drag the strength slider until the area is unreadable.
- Press Apply. The effect is baked into the image. Move the box to the next area and Apply again to hide several things.
- Download in PNG, JPEG or WebP. Use Undo to step back or Revert all to return to the original.
Pixelate vs blur — which should you use?
| Question | Answer |
|---|---|
| Which effect hides text most reliably? | Pixelate at a large block size — it destroys the shapes of letters so they cannot be read or reconstructed. |
| Which looks the most natural? | Blur — a strong blur softens an area without the blocky mosaic look, good for faces and backgrounds. |
| Can a blur be “un-blurred”? | A very light blur can sometimes be partly reversed by software. For sensitive data use a strong blur or, better, a large pixelate block so the original detail is genuinely gone. |
| Does it change the rest of the photo? | No — only the box you draw is altered. The rest of the image stays pixel-for-pixel identical. |
Rule of thumb: pixelate at a large block to reliably kill text and numbers, blur when you want a softer, more natural hide for faces and backgrounds.
What you can redact with it
| Use case | How |
|---|---|
| Blur a face | Redact a person in a photo before posting to social media or a review. |
| Hide a licence / number plate | Pixelate a car plate in a for-sale or accident photo. |
| Redact an address or ID | Blur a house number, Aadhaar/PAN, passport number or bank detail in a scan. |
| Censor a screenshot | Pixelate names, emails, API keys or chat handles before sharing a bug report. |
| Cover a signature | Blur a handwritten signature on a document image you need to send. |
| Anonymise bystanders | Pixelate people in the background of a photo for privacy. |
How the effects work
When you open an image the browser decodes it into raw pixels on a canvas. The area you select is read back with getImageData(), transformed, and written back with putImageData() — only inside your box; the rest of the photo is never touched.
Pixelation divides the selection into a grid of squares whose size is the block value. For each square the tool averages the red, green, blue and alpha of every pixel underneath, then paints the whole square that single average colour. Bigger blocks average over more pixels, so more of the original detail is thrown away — which is exactly what makes a large block a reliable censor for text.
Blur uses a separable box blur run three times, which closely approximates a smooth Gaussian blur. Each pass replaces every pixel with the average of its neighbours within the blur radius, first across rows then down columns; running it three times removes the boxy artefacts and leaves a soft, even smear. To avoid a hard seam at the edge of your selection, the tool reads a margin of surrounding pixels when it blurs, so the softened patch blends into the rest of the image. Both effects are pure pixel maths — there is no server round-trip and no external library.
Redaction done safely: what to watch for
Hiding information in an image is only as safe as the method. A few practical rules keep a redaction genuinely irreversible:
- Use enough strength. A faint blur can leave the shape of a word or a face partly legible, and specialised software can occasionally sharpen it back. When the data matters, use a large pixelate block or a high blur radius so the original detail is averaged completely away — once it is baked into the downloaded pixels, there is nothing left to recover.
- Prefer pixelate for text and numbers. Letters, digits, barcodes and QR codes have hard edges that a mild blur only softens; a big mosaic block destroys the pattern entirely.
- Cover the whole thing. Draw the box a little larger than the item — a plate number peeking out at the edge, or the last two digits of a card, defeats the point.
- Don't just draw a black box in a comment layer. Some “redactions” are only an overlay that another app can move aside. Because this tool bakes the effect into the actual pixels and re-encodes the file, the hidden data is truly gone from the download.
- Re-encoding strips EXIF/GPS. The saved file has no camera metadata or location, so you are not leaking where the photo was taken alongside what is in it.
Why the block size matters for a “mosaic” censor
The block size is the single most important setting when you are trying to make something unreadable. Think of pixelation as sampling the image at a lower resolution: a block of 4 pixels keeps a lot of structure — you can often still read large text through it — whereas a block of 30 pixels reduces a face or a line of type to a handful of flat squares that carry no recoverable detail. Because each block is the plain average of everything beneath it, the information in those pixels is mathematically collapsed to one number per channel; there is no clever filter that can invent the lost variation back.
A good workflow is to start moderate, apply, and check whether you can still guess what was there. If any character or feature is recognisable, raise the block size and apply again over the same spot — the effect stacks. For small on-screen text, 12–20 pixels is usually plenty; for a headline or a large face, push it to 30 or more. The same logic applies to blur radius: the larger the radius, the wider each pixel is spread and the less of the original survives.
Frequently asked questions
- Is my image uploaded to a server?
- No. The photo is decoded, edited and re-encoded entirely in your browser with the built-in Canvas API — it never leaves your device and is never sent anywhere. That is exactly why this tool is safe for redacting IDs, bank statements, passports, private chats and screenshots you would never want to upload to a random website.
- What is the difference between pixelate and blur?
- Pixelate (also called mosaic or censor) replaces a region with big solid squares, each square coloured the average of the pixels under it — it looks blocky and, at a large block size, completely destroys the original detail. Blur smears each pixel into its neighbours so the area becomes soft and unreadable while still looking smooth. For hiding text and numbers, pixelate at a large block is the most reliable; for faces and backgrounds a strong blur usually looks more natural.
- Can the pixelation or blur be reversed to reveal what was hidden?
- Because the edit is baked into the downloaded pixels — the original detail is averaged away and thrown out — a strong pixelate (large block) or strong blur cannot be reversed: the information simply is not in the file any more. The one caveat is a very light blur, which specialised software can sometimes partly recover. When you are hiding something genuinely sensitive, use a large pixelate block or a high blur radius so there is nothing left to recover.
- How do I hide more than one thing in the same photo?
- Apply the effect to one area, then just move or resize the box over the next area and press Apply again — each Apply is baked into the working image, so you can censor several faces, plates or lines of text one after another and then download the final result once. Use Undo to step back if you cover the wrong spot.
- Which effect and strength should I use to censor text?
- Choose Pixelate and raise the block size until the letters are unrecognisable — for small text a block of 12–20 pixels is usually enough, for larger headings go higher. Pixelation is preferred over blur for text because a blur can leave the overall word shape faintly legible, whereas a large mosaic block removes it entirely. Zoom in mentally: if you can still guess a character, increase the block size and apply again.
- Will editing the photo strip its EXIF and GPS data?
- Yes. Re-encoding through the canvas produces a clean file with no EXIF block, so the camera model, capture time and GPS location are removed automatically when you download — a second privacy win on top of the redaction itself. If you need to keep that metadata, keep your original file as well.