How to use BlurVault
Guides for photos, folders of photos and video, which browsers handle what, and fixes for the problems people run into. Each step shows the part of the screen you'll be using.
Blur a photo
Free, no account. The whole thing takes under a minute once the detector has loaded. Watch the loop below, then follow the steps.
- 1
Open the photo
Drag it onto the editor, paste it with ⌘ V (Ctrl+V on Windows), or press Choose a photo. The browser reads the file into memory in this tab.
Want to practise first? Try the sample image. It's a drawn street scene with four people and a car.
- 2
Check the automatic marks
Faces are boxed as soon as the photo opens. The first time, the detector downloads (about 11 MB) and a progress bar shows how far along it is. Look at every box, and press D to run detection again after you've removed some.
- 3
Draw boxes over everything else
Plates, screens, name badges, house numbers and documents are never found automatically. Drag across each one to add a box. Drag inside a box to move it and drag a corner to resize it. On a phone, the same gestures work with a finger, and the corner handles get bigger.
Removed one by mistake? Press ⌘ Z or the Undo link in the status bar.
- 4
Pick the effect and export
Blur is free. Pixelate and the solid bar come with Pro; use the bar for text. Strength sets how heavy the blur is. Export PNG keeps the most detail, JPEG makes a smaller file. The export is a new file, and your original isn't changed.
Blur a folder of photos
Batch mode is in the full editor and needs Pro. It is built for event photos, site surveys and evidence sets where every face has to go.
- 1
Drop the photos and choose the settings
Open the Batch tab and drop as many photos as you like, or choose them from a folder. Pick the effect, the strength and whether to save JPEG or PNG. The settings apply to every photo in the list.
- 2
Run it, then check what was held back
Press Blur and save. Each photo is processed in turn and saved to your downloads. Stop after this one finishes the current photo and leaves the rest queued.
A photo where no face was found is not saved, because an unmarked copy could go out with faces showing. Press Open in editor next to it to check it by hand.
If the detector fails to load mid-batch
The batch stops before saving anything else. Nothing is exported without detection, so an unblurred copy can't slip into your downloads.
Blur a video
Video is a Pro feature in the full editor. Every frame is drawn, blurred and re-recorded in the browser at playback speed, so a two-minute clip takes about two minutes.
- 1
Open the clip and pin the fixed things
Open the Video tab and drop the clip. Use the timeline under the preview to find a frame where a plate, a screen or a sign is visible, then drag a box over it. Boxes you draw stay in place for the whole clip, which suits dashcam footage, screen recordings and interviews filmed on a tripod.
Leave Find faces on every frame switched on. Faces are detected again on every frame, so they are followed as they move.
- 2
Record, keep the tab in front, and check the result
Press Blur and save video. The progress bar shows the time and how many faces are in the current frame. If you switch tabs, processing pauses and resumes when you come back. Cancel throws the partial recording away.
When it finishes, the video downloads and a preview appears. Scrub through it once before you share it. Fast head turns and faces at the edge of the frame are where detection is weakest.
Output format
Chrome, Edge and Firefox save WebM (VP9 or VP8). Safari saves MP4. If you need MP4 from a WebM file, any video converter can change the container without touching the blurred frames.
Keyboard shortcuts
They work when the pointer or focus is inside the editor.
- Find faces again
- D
- Remove the selected box
- Del
- Undo (Ctrl+Z on Windows)
- ⌘Z
- Redo
- ⇧⌘Z
- Nudge the selected box (hold Shift for bigger steps)
- ←→
- Blur, pixelate, bar
- 123
- Fit the photo to the window
- F
- Zoom to 100%
- 0
- Zoom in and out
- +−
- Deselect
- Esc
Browser support
Current browser releases. “Partly” means it works with the limit noted.
| Feature | Chrome, Edge | Firefox | Safari | Phones |
|---|---|---|---|---|
| Photo editor and face detection | Works | Works | Works | Works |
| Paste a photo from the clipboard | Works | Works | WorksmacOS | PartlyUse the file picker |
| HEIC photos from an iPhone | NoConvert to JPEG | NoConvert to JPEG | Works | NoConvert to JPEG |
| Batch folders | WorksAllow multiple downloads | Works | Works | PartlyOne file at a time on some phones |
| Video export | WorksWebM | WorksWebM | PartlyMP4, recent versions | PartlyShort clips |
| Original audio in exported video | Works | Works | PartlyVaries by version | PartlyVaries |
Troubleshooting
No faces were detected
The detector is tuned for faces that are reasonably large in the frame and turned toward the camera. It misses most often on faces that fill only a sliver of the frame, heads in profile or turned away, faces behind sunglasses or masks, very dark photos, and drawings. Large photos are also scanned in overlapping tiles, which catches many small faces, but not all of them.
Zoom in with + and drag a box over anything it skipped. For a crowd in the distance, one wide box over the whole crowd is quicker than a box per person. In batch mode, photos where nothing was found are held back so you can open them in the editor.
The face detector won't load
The first time you use it, the browser downloads the detector: about 11 MB of code and a 230 kB model. BlurVault serves its own copy and falls back to the public copies on jsDelivr and Google if that fails. Corporate firewalls, strict content blockers and captive Wi-Fi portals are the usual reasons both fail.
Press Retry next to the detector status, or allow blurvault.vercel.app, cdn.jsdelivr.net and storage.googleapis.com. Drawing boxes by hand and exporting work without the detector.
Video export fails or stutters in Safari
Safari records MP4 where Chrome, Edge and Firefox record WebM, and older Safari versions can't record from a canvas at all. On iPhone and iPad, memory runs out sooner on long or 4K clips.
Update Safari, or use desktop Chrome or Firefox for long clips. Trimming the clip to the part you need before you start also helps.
The exported video has a pause or a frozen stretch
Video is processed in real time. If the tab goes to the background, BlurVault pauses the recording and resumes when you come back, so the output stays complete. A laptop going to sleep or a phone locking its screen can still interrupt it. Keep the tab in front and the device plugged in for long clips.
Only one file downloaded from a batch
Browsers ask before a site saves several files in a row. Look for a prompt or a blocked download icon in the address bar, allow it, and press Blur and save again. Photos already saved are skipped.
My iPhone photo won't open
iPhones save photos as HEIC, which only Safari can decode. Open BlurVault in Safari, or share the photo as JPEG first (in Photos: Share, then Options, then choose Most Compatible).
The text under the blur is still readable
Raise Strength, or switch to the solid bar, which paints over the area completely. For names, account numbers and documents, use the bar. A light blur over big text can sometimes still be read.
The export is smaller than my original
Free exports are capped at 1280 pixels on the long side and carry a small watermark. Pro and Lifetime export at the original size with no watermark.
Check that nothing is uploaded
Five minutes with DevTools tells you more than any privacy page can.
- 1Open DevTools with F12, or ⌥ ⌘ I on a Mac, and pick the Network tab.
- 2Drop a photo into the editor below, adjust the boxes and export it.
- 3Read the list. Sort by size, or filter for your file name. No request carries your photo or anything made from it.
- 4Still unsure? After the first photo, turn off Wi-Fi and blur another one. The editor keeps working because it never needed the connection for your files.
| Name | Type | Size | What it is |
|---|---|---|---|
| blurvault.vercel.app | document | ≈ 25 kB | This page |
| _next/static/… | script, css, font | ≈ 400 kB | The editor's code |
| vision_wasm_internal.wasm | fetch | 11.2 MB | Face detector runtime, first visit only |
| vision_wasm_internal.js | script | 320 kB | Its loader |
| blaze_face_short_range.tflite | fetch | 230 kB | Face model, first visit only |
| view | fetch | < 1 kB | Anonymous page view (Vercel Analytics) |