Blur a Face or Any Part of a Photo
Drag a rectangle over the preview to mark an area, then blur it, pixelate it or cover it. Everything outside your rectangles is left bit for bit untouched, and this page is honest about which of the three is actually safe.
Drop images here
or click to browse · or paste with Ctrl+V
JPG · PNG · WebP · GIF · BMP · AVIF
Drag a rectangle over the preview to mark an area. Drag again to add another. Regions are stored as percentages, so they land in the same place at full resolution.
Method
Presets
This matters more here than on most tools. The photo never leaves your device, so an image you are redacting because it is sensitive is not uploaded to anyone, including us.
People also use
Everything this tool does
Drag, do not guess
Draw a rectangle straight onto the preview and drag again to add more. Regions are held as percentages of the image, so what you marked at preview size lands in exactly the same place at full resolution.
A box that actually redacts
On a test region whose original content had a variance of 1,647, a solid fill brought it to exactly 0: every pixel identical, nothing of the original left. Heavy pixelation left 277 and a heavy blur left 31. Only the box reaches zero.
Honest about the weak methods
Blur and pixelation are averaging operations, and averages are data. This page says so plainly rather than letting you assume a blurred face is anonymous.
Nothing outside your regions changes
Pixels outside the rectangles are copied through rather than resampled, so the rest of the photograph is bit for bit identical to the file you started with.
The blur stays inside the box
Sampling is clamped to the region, so the redaction is built only from pixels that were going to be destroyed anyway, and nothing leaks across the boundary in either direction.
Strength scales with the region
Block size and blur radius are proportional to the area you marked. Pixelating a small region and one four times larger at the same setting left exactly 25 distinct colours in each, so a distant face is obscured as thoroughly as a close one.
Drag to compare
Before and after share one frame with a slider, so you can confirm the region landed where you meant it to.
Batch with one ZIP
The same regions are applied to every image in the batch, which suits a series of frames from the same camera position.
Six purpose-built presets
Redaction, white box, pixelate face, soft pixelate, heavy blur and gentle blur.
Nothing is uploaded
No server, no queue, nothing to delete afterwards, which matters far more here than on an ordinary filter.
Which method to use
These three are not interchangeable. One of them genuinely destroys the information underneath and two of them only make it hard to read, which matters a great deal depending on why you are doing this.
| Method | What survives | Use for | Safe to publish? |
|---|---|---|---|
| Solid box | nothing at all | Documents, IDs, real redaction | Yes |
| Pixelate, large blocks | average colour per block | Faces in casual photos | Usually |
| Pixelate, small blocks | a surprising amount | Looks only | No |
| Blur, heavy | low frequency structure | Casual anonymity | Not for anything serious |
| Blur, light | most of the shape | Aesthetic softening only | No |
| Soft edge on | partial detail at the rim | Looks only | No |
| Soft edge off | nothing extra at the rim | Any redaction | Yes |
Drag
to mark an area
Any
number of regions
0
files uploaded
0
pixels changed outside your regions
ZIP
batch download
Free
no account needed
FlipMyFormat vs other face blur tools
| Capability | FlipMyFormat | Typical free tool |
|---|---|---|
| Files stay in your browser | ||
| Drag any number of regions | ||
| Everything outside the regions untouched | ||
| Honest about which method is actually safe | ||
| Solid box redaction, not just blur | ||
| Batch ZIP download without paying | ||
| No watermark on the output |
How to hide something in a photo, properly
There are two completely different jobs that look identical on screen, and the mistake that matters is doing the first while believing you are doing the second. The first job is aesthetic: you want a face in the background of a holiday photo to not be the subject, or you want a bystander to be unidentifiable at a glance. Blur and pixelation are fine for that, they look natural, and nobody is going to attack them. The second job is redaction: there is information in this image that must not reach the person who receives the file, and the file may be published, forwarded or stored. For that job, blur and pixelation are not good enough, and people have been embarrassed by assuming otherwise for as long as both have existed. The reason is simple. Blurring and pixelating are averaging operations, and averaging discards a lot less than it appears to. Pixelation replaces a block with its average colour, which still tells you the average colour of that block; with small blocks and a known set of candidates, matching the pattern is not hard. Blur spreads information out rather than removing it, and enough of the original structure survives to be worked with. A solid box has none of these problems, because there is nothing underneath it: the pixels have been replaced with a constant, and no analysis can recover what a constant used to be. So the rule is easy. If it is for looks, use whatever looks best. If someone must genuinely not be identified, use the box, leave the soft edge off, and check the exported file rather than the preview.
- 1
Add your photo
Drag it in, click to browse, or paste with Ctrl+V. It stays on your device, which is the point when the reason you are here is that the photo is sensitive.
- 2
Drag over what you want hidden
Draw a rectangle on the preview. Drag again to add another, as many as you need. Regions are stored as percentages, so they land correctly at full resolution.
- 3
Choose the method and export
Solid box if this is a real redaction. Pixelate or blur if you only want the photo to look anonymous. Then save individually or as one ZIP.
Why a box is different in kind, not degree
Averaging keeps more than it looks like
Pixelation replaces every block with the average colour of the pixels inside it, and blur replaces every pixel with a weighted average of its neighbours. Both reduce information dramatically and neither removes it. The averages themselves are data: a pixelated face still carries the coarse arrangement of light and dark that a face has, and if the range of possible answers is small, such as a set of digits or a known list of people, comparing candidates against the same pixelation is a well-known attack. A solid fill is categorically different because it does not depend on what was there. We measure it directly. On a test region whose original content had a pixel variance of 1,647, a solid box brought that to exactly 0: every pixel is the same value and the region carries no information whatsoever. Heavy pixelation of the same region left 277 and a heavy blur left 31. Read that last pair carefully, because it is easy to misread: variance measures how much visible structure remains, not how recoverable the original is. Blur scores lower simply because it smooths detail away, while pixelation keeps a coarse grid of block averages. Neither reaches zero, and only reaching zero means the information is actually gone.
Nothing outside your rectangle is touched
Two smaller guarantees are worth knowing. First, pixels outside the regions you drew are copied through untouched rather than resampled, so the rest of the photograph is bit for bit identical to what you put in, which we verify by comparing every pixel outside the marked areas against the original. Second, the blur used inside a region samples only from within that region, clamping at its edges. That means the redaction is built purely from pixels that were already going to be destroyed, so no detail from just outside the rectangle is dragged into it, and equally nothing from inside leaks outward past the boundary you drew.
What people use it for
Faces of bystanders
Photos taken in public often contain people who did not agree to be in them, and obscuring them before publishing is basic courtesy.
Children in shared photos
Many parents want school and holiday photos shareable without their children being identifiable.
Number plates
A car in your driveway is a small identifying detail that is easy to remove before posting.
Documents and forms
Account numbers, addresses and signatures on a photographed document. Use the solid box for these, never blur.
Screenshots with personal data
Email addresses, names in a sidebar, message previews. A box is fast and complete.
Property listings
Removing family photos, personal items and neighbours' windows before an interior shot goes public.
Journalism and case studies
Protecting a source or a subject while still publishing the photograph.
Before and after posts
Sharing a client result while keeping the client unidentifiable.
Six tips worth knowing
If someone genuinely must not be identified, use the solid box. Blur and pixelation only make a photo hard to read, not impossible.
Leave the soft edge at zero for any redaction. A blended edge leaves partial detail readable at the rim, which is the one place you least want it.
Draw the rectangle larger than you think you need. Hair, ears and the line of a jaw all identify a person, not just the face.
Check the exported file, not just the preview. The preview runs on a smaller copy, so it can look more obscured than the full size result.
For a batch, the same regions apply to every image, which works when the camera did not move and not otherwise.
Metadata is a separate problem. This changes pixels only, so strip EXIF separately if the location matters.
Troubleshooting
My rectangle landed in the wrong place
Regions are percentages of the image, so they scale correctly. If one is off, use Clear all and draw it again while watching the outline as you drag.
The face is still recognisable
Either the region is too small, or you are using blur or pixelation at a low strength. Enlarge the rectangle and raise the strength, or switch to a solid box.
Can this be reversed?
A solid box cannot, because there is nothing underneath it. Blur and pixelation can sometimes be attacked, especially with small blocks and a limited set of possible answers.
Does it remove location data?
No. This tool changes pixels only. EXIF metadata including GPS coordinates is a separate thing and needs a separate step.
Frequently asked questions
How do I blur a face in a photo?
Add the photo, drag a rectangle over the face on the preview, and choose blur. Drag again for more faces. Everything outside the rectangles you draw is left exactly as it was.
Is a blurred face actually anonymous?
Not reliably. Blur spreads information out rather than removing it, and enough structure survives to be worked with. For casual purposes it is fine; if someone must not be identified, use the solid box.
Is pixelation safe?
Safer than blur with large blocks, and not safe with small ones. Pixelation replaces each block with its average colour, and that average is itself data. Where the set of possible answers is small, matching against the same pixelation is a known attack.
Why is the solid box different?
Because it does not depend on what was underneath. On a test region with an original variance of 1,647, a solid fill brought it to exactly 0, so every pixel is identical and nothing survives. Heavy pixelation left 277 and a heavy blur left 31. Note that variance measures remaining visible structure rather than recoverability, so the point is not that blur scores lower than pixelation, it is that only the box reaches zero.
Can I mark more than one area?
Yes, as many as you like. Each drag adds another rectangle, and Clear all removes them.
Will the region be in the right place at full size?
Yes. Regions are stored as percentages of the image rather than pixels, so what you marked on the preview lands in exactly the same place when the full resolution file is processed.
Should I use the soft edge?
Only for looks. A blended edge leaves partial detail readable around the rim, which is precisely where you do not want it. For any redaction, leave it at zero, and the solid box ignores it entirely.
Does anything outside my rectangle change?
No. Those pixels are copied through rather than resampled, so the rest of the photograph is bit for bit identical to the file you started with.
Does it detect faces automatically?
No. You mark the areas yourself, which is slower but predictable: automatic detection misses faces, and a missed face in a redaction is a serious failure rather than a minor one.
Is my photo uploaded?
No, and it matters more here than anywhere else on this site. Everything runs in your browser, so a photograph you are redacting because it is sensitive is not sent to us or to anyone else.
Does this remove location data from the file?
No. This changes pixels only. EXIF metadata, including GPS coordinates, survives and needs to be stripped separately if it matters.
Can I redact a document with this?
Yes, with the solid box and no soft edge. That is the only combination here suitable for information that must not be recoverable.
How large should I draw the rectangle?
Larger than the face itself. Hair, ears, jawline and clothing all help identify a person, so a tight crop around the eyes is not enough.
Can I apply the same regions to several photos?
Yes, the batch applies your regions to every image. That only makes sense when the camera did not move between frames.
Which format should I export?
Keep the original for photographs. PNG keeps the edge of a solid box exact, while JPEG softens it very slightly, though it cannot recover what is underneath.
Why does the preview look more obscured than my export?
The preview is computed on a smaller copy for speed. Always check the exported file before publishing anything sensitive.
Can the redaction be undone?
A solid box cannot be, because the original values are gone. Blur and pixelation discard a great deal but not everything, which is exactly why they are the weaker choices.
Does it add a watermark or need an account?
No to both. No sign-up, no email wall, no daily limit and no watermark on the output.