Blur an Image to Hide Faces and Details
How to censor faces and details with blur or pixelation
Blur or pixelate an entire image or selected regions. Mark the areas to obscure and adjust the strength while checking the preview. Small text, faces and identifying details can remain recognizable with a weak setting, so inspect the exported image at full size. For information that must be removed completely, crop it out or cover it with an opaque layer.
Cost: Free to use. File size and usage limits apply.
How to blur or pixelate part of an image
Upload one image
Click Upload or drop a file onto the preview: JPG, PNG, WebP, GIF, BMP, ICO, TIFF, TGA, AVIF, HEIC or HEIF, up to 200 MB. Images you uploaded or created in other tools are listed under Files.
Choose the whole picture or parts of it
Full image is selected by default and applies the effect everywhere. Switch to Selective to censor only chosen areas, such as a face, a license plate or a name.
Pick Blur or Pixelate and set the strength
Blur smooths the area, Pixelate turns it into square tiles. Move the Radius or Tile size slider; the preview updates as you drag.
Draw boxes over the areas to hide
In Selective mode, drag a rectangle on the preview over each area. Every box is labelled Blur or Pixelate. Clear removes all boxes at once if you want to start over.
Apply, compare and download
Click Apply Blur Effect!, then drag the compare slider to check the result against the original. Download saves the file, and the result is also added to Files. Clear result takes you back to the editor but removes your boxes.
Blur settings explained
All options sit in the Blur card. Its header shows the current strength in px.
- Full image
- The default. The whole picture is blurred or pixelated, for example for a spoiler image or a soft background behind text. Boxes drawn in Selective mode are ignored.
- Selective
- Only the rectangles you draw are changed. Boxes are always rectangular, so for a face, frame the whole head including hair and ears.
- Blur
- A Gaussian blur that melts detail into soft gradients. It looks calmer than a mosaic, but at low values the outline of a face or large lettering stays recognisable.
- Pixelate
- The classic pixel blur: a mosaic of square tiles with one colour each. In Selective mode the choice applies to the boxes you draw next, so one image can mix both effects: draw some boxes, switch the mode, then draw the rest.
- Radius
- Blur only, 1–50 px, default
12. The value is the standard deviation of the blur, and visible smearing reaches about three times as far: at12, each pixel spreads over roughly 36 px in every direction. - Tile size
- Pixelate only, 2–40 px, default
12. Roughly the edge length of each tile. Radius and Tile size share one value, so switching the effect keeps the number, and in Selective mode it applies to every box. - Clear
- Shown in Selective mode. Removes all boxes at once.
How strong a censor needs to be
Blur and pixelation alter pixels instead of painting over them. The file no longer contains the original pixels under a box, but a weak setting can leave enough shape and colour to recognise a face or guess short text, especially digits in a known font. Pixelated text from screenshots has been reconstructed this way. For passwords, account numbers or addresses, cut the area away with Crop Image whenever the layout allows it.
Judge the strength against the size of what you hide. Example: a 4032 × 3024 px phone photo in which a face is about 300 px wide. Tile size 12 leaves about 25 tiles across the face, and its features stay readable. The maximum of 40 px gives 7 to 8 tiles. As a rule of thumb, a face should be only a few tiles wide. With Blur, Radius 50 smears each pixel over about 150 px, half the width of that face.
If the maximum is not enough, scale the photo down first with Resize Image: at 1008 × 756 px the same face is about 75 px wide, and tile size 20 covers it with about 4 tiles. You can also pick the result under Files and blur it a second time; two passes at 50 act like one pass of about 71.
Check the downloaded file at 100% zoom, not only the scaled-down preview. Faces in the background and reflections in windows or glasses are easy to miss.
What happens to format, quality and metadata
The result keeps the format of the upload. This tool has no format or quality setting.
JPG
Saved as JPG again at high quality. The whole image is re-encoded, including areas outside the boxes, so fine detail can change. Compare the downloaded image and its actual file size with the source.
PNG, BMP, TIFF, TGA and ICO
Saved in the same format. Transparent areas in a PNG stay transparent; with Blur, the edges of transparent areas become soft as well.
WebP and AVIF
Still WebP and AVIF are encoded in the same format at quality 92 out of 100. Still WebP is lossy, even if the source was lossless. Animated WebP passes through a 256-colour palette, then its transformed frames are encoded losslessly; 92 sets compression effort for that animation.
HEIC and HEIF
Saved as PNG, as the note in the settings panel explains. The server reads these formats and exports the edit as PNG. The lossless file can be larger, depending on the source image. Convert Image can produce a JPG afterwards.
Animated GIF
Every frame is processed and the timing is kept. The colour palette is rebuilt; gradients and dithering can change the file size. Compare the exported bytes rather than assuming Blur or Pixelate will give the smaller file.
EXIF and GPS data
Camera data, including the GPS position of a phone photo, is not copied to the result.
Formats and limits
- Input formats
- JPG, PNG, WebP, GIF (including animated), BMP, ICO, TIFF, TGA, AVIF, HEIC, HEIF
- Output format
- Same as the input; HEIC and HEIF become PNG
- Output size
- Same px as the input in every mode
- Areas
- The full image, or any number of hand-drawn rectangles
- Strength
- Radius 1–50 px, Tile size 2–40 px
- Maximum file size
- 200 MB
- Files per job
- 1
- Processing
- On our servers with FFmpeg; WebP, AVIF, HEIC and HEIF files also pass through ImageMagick. No AI service and no automatic face detection. A job can run for up to 15 minutes.
- Watermark
- None
- Storage
- Uploads and results are deleted automatically after 48 hours for guests and after 30 days for free accounts. Paid plans can keep files longer.
Frequently asked questions
Can the tool find and censor faces automatically?
No. There is no face detection. Switch to Selective and draw a box over each face yourself; the number of boxes is not limited.
Can someone undo the blur or pixelation?
The file has no layers and does not contain the original pixels under a box. At weak settings the remaining shapes can still give a face or short text away, so use a high Radius or Tile size and crop sensitive text out.
Can I blur an animated GIF?
Yes. All frames are processed and the animation timing is kept. Boxes stay in a fixed position, so a moving subject needs a box that covers its whole path. The preview shows only the first frame, so check the animation in the result.
Why is the result larger than the original?
The result is re-encoded. HEIC and HEIF become PNG, while static WebP and AVIF are saved at quality 92, which may be higher than the source setting. A GIF's new palette and dithering also affect its size. The source and effect determine the actual bytes; unchanged dimensions do not guarantee an unchanged file size.
Is my photo uploaded, and is the location data removed?
Yes, the image is uploaded and processed on our servers with FFmpeg; it is not sent to an AI service. EXIF data such as camera model and GPS position is not copied to the result. Uploads and results are deleted automatically after the storage period listed under Formats and limits.