Blur Video

The blurred area does not move. Read this before you publish.

The most important thing on this page, because getting it wrong means publishing the thing you were trying to hide.

The box you draw is applied at fixed coordinates for the whole video. It does not track a face, and it does not follow anything. If the person moves, or the camera moves, or they simply turn their head, they walk out of the blurred area and are fully visible for the rest of the clip.

Tools that advertise automatic face blurring detect the face in every frame and move the blur with it. This does not do that. So:

Watch the entire clip after processing, before you publish anything. Not the first few seconds — all of it. That single habit is what separates a blur that works from a privacy failure, and it takes as long as the video.

Two ways to make a static box work:

Cover the whole range of movement. Scrub through the clip, find where the subject is at their furthest in each direction, and draw a box big enough to contain all of those positions. It blurs more of the frame than strictly necessary, and it actually works.

Or split the clip. Cut it into segments where the subject stays roughly still, blur each segment with its own box, and join them. More work, and much tighter than one enormous box across a two-minute shot.

If neither is practical — a subject moving unpredictably across the whole frame for the whole clip — then this is the wrong tool and a face-tracking editor is the right one. Better to know that now than after uploading.

Region mode blurs only the region

Worth knowing because some tools do this badly.

When you choose region mode, that area is taken out, blurred and placed back — so the rest of the frame is completely untouched, at its original sharpness. You are not looking at a softened video with a slightly softer patch; you are looking at your video with one area obscured.

Draw the box on the preview. If you forget, the tool tells you rather than silently processing the whole frame — which is the right behaviour, because a video you thought was partly blurred and is not is exactly the failure this tool exists to prevent.

Full video blur is the other mode, and it has real uses of its own: a background you want unreadable behind a talking head, a blurred clip as a backdrop for text, or obscuring an entire frame that is too complicated to fix selectively.

Use Heavy for privacy, Medium for aesthetics

The three intensities are not interchangeable, and the choice depends on whether you are hiding something or styling something.

For privacy, use Heavy. A light blur on a face is often still recognisable to someone who knows the person, and text under a light blur can sometimes be read or reconstructed. If the reason for blurring is that someone must not be identified, there is no advantage to subtlety — go as strong as the tool offers and check the result by looking at it, not by assuming.

For aesthetics, Medium is usually right. A background you want de-emphasised rather than destroyed, a soft backdrop behind titles.

Light is for a gentle softening effect and is not a privacy setting.

One check worth doing: pause on a frame where the subject is largest and closest to the camera. That is the frame where a blur is most likely to be insufficient, and it is the frame someone trying to identify them would use.

What "blurred" does and does not guarantee

An honest note, in the spirit of the rest of this.

Blurring the pixels genuinely replaces them — the original detail is not hidden underneath waiting to be recovered, because the video is re-encoded with the blurred pixels in place. There is no undo hiding in the file.

But identification does not only come from a face. A distinctive jacket, a tattoo, a gait, a vehicle, a house number, a reflection in a window, a name spoken on the audio, or the simple fact of where and when the video was taken can all identify someone with the face fully blurred. If the reason for blurring is serious, look at the whole frame and listen to the whole soundtrack, not just the face.

Why not uploading matters here specifically

More than for almost any other video task.

The reason you are blurring is usually that the footage contains someone who has not consented to being identified — a bystander, a child, a colleague, a patient, a member of the public. Uploading that footage to a free online service in order to protect their privacy is self-defeating: the unblurred original reaches a third party's server, which is precisely the outcome you were avoiding.

Here the processing happens on your own machine. Nothing is uploaded, there is no queue, no account, no watermark and no file size limit. The one cost is that the video engine loads into your browser on first use — a few tens of megabytes, cached afterwards.

Output is an MP4, and the video is re-encoded, which is unavoidable when the pixels change.

Related tools

Crop Video is often the better answer: if the thing to hide is near the edge of frame, cropping it out removes it entirely rather than obscuring it. Trim Video cuts the clip into segments, which is how you blur a moving subject properly. Pixelate Image does the equivalent job on a still photograph, with a region mode too.