How to redact a license plate from dashcam video
- Step 1Load the dashcam clip — Drop the MP4, MOV, MKV, AVI or TS file. Many dashcams write MP4 or TS; all are read locally by FFmpeg.wasm with no upload. One file per run.
- Step 2Scrub to where the plate is clearest — Set the Start (s) first — the picker renders the frame at that timestamp, so jump to the moment the plate is most legible and frame your box against it. A plate is usually only readable for a short stretch, so pin Start to that.
- Step 3Draw a box over the plate — Drag a rectangle over the plate on the frame. Click-drag inside to nudge it, or drag the SE corner to resize. Make it a little larger than the plate itself to allow for vehicle and camera movement during the window.
- Step 4Set the time window the plate is on screen — Enter End (s) for the moment the plate is no longer legible or leaves frame. The blur is active only between Start and End, so the rest of the footage — road, signs, your own dashboard — stays sharp.
- Step 5Pick a strong blur — Plates are high-contrast and small, so use a high Blur strength (toward 50). Remember the radius is capped at half the smaller side of the box, so if a tight box still shows readable characters, enlarge the box rather than just raising the number.
- Step 6Export and verify legibility is gone — Process to H.264 MP4 and scrub the output frame-by-frame across the window. Confirm the registration is unreadable at every frame — including when the vehicle is closest to the camera and the plate is largest.
Plate-blur control cheatsheet
The exact controls and the values that work well for a moving license plate.
| Control | Plate-blur tip | Why |
|---|---|---|
Start (s) / End (s) | Window only the seconds the plate is legible | Keeps road, signs and dashboard sharp for the claim |
Width / Height | Draw ~20–40% larger than the plate | Covers vehicle and camera motion within the window |
Blur strength | Toward 50 | Plates are high-contrast; weak blur can leave characters readable |
| Box too small at strength 50 | Enlarge the box, don't just raise strength | Radius is capped at min(width,height)/2 |
| Plate moves a lot | Split into several passes by time range | The box is static; one box can't follow fast motion |
Output facts for evidence and claims
What the file looks like after redaction — useful when documenting the chain of custody.
| Property | Value |
|---|---|
| Processing location | Your browser (FFmpeg.wasm) — no upload |
| Redaction method | FFmpeg boxblur over a cropped region (destructive) |
| Output container / codec | MP4 / H.264 (libx264, CRF 20, preset medium) |
| Audio | Stream-copied, unchanged |
| Metadata (GPS, device) | Not stripped — use metadata-scrubber if required |
| Max file size (Free / Pro / Media / Dev) | 1 GB / 10 GB / 100 GB / 100 GB |
Cookbook
Recipes tuned to dashcam and bodycam plate redaction. Coordinates are source pixels.
Blur the plate of a car ahead for 6 seconds
The car in front shows a legible plate from 0:08 to 0:14 before turning off. Window the blur to that span so the rest of the journey footage stays sharp.
Source: dashcam_front.mp4, 1920x1080 Redact rectangle: X=880 Y=560 Width=170 Height=70 Start (s): 8 End (s): 14 Blur strength: 48 radius = min(48, floor(70/2)) = 35 Filter: [0:v]split[base][redact];[redact]crop=170:70:880:560,boxblur=35:1[blur]; [base][blur]overlay=880:560:enable='between(t,8.000,14.000)'[outv]
Tight plate box leaves characters readable — fix
A 90×30 box at strength 50 still shows the registration because the radius caps at 15. Widen and heighten the box to raise the cap.
Before: 90x30, strength 50 -> radius = floor(30/2) = 15 (readable) After: 140x70, strength 50 -> radius = floor(70/2) = 35 (illegible) Lesson: for plates, generosity on box size beats max strength.
Plate visible in two separate clips of the same incident
Front and rear dashcam files each show a plate. Redact each file separately — this tool processes one file per run.
front.mp4 -> box over plate -> front_redacted.mp4 rear.mp4 -> box over plate -> rear_redacted.mp4 Submit both redacted MP4s to the insurer.
Strip GPS metadata after blurring
Dashcam MP4s often embed location data. Blurring the plate doesn't remove it, so chain the metadata-scrubber for a clean file before sharing publicly.
Step 1: dashcam.mp4 -> redact plate -> plate_blurred.mp4 Step 2: plate_blurred.mp4 -> metadata-scrubber -> clean.mp4 Result: plate obscured AND no embedded GPS/device tags.
Keep audio for the incident account
You narrated the collision aloud. The redactor copies audio untouched, so your spoken record survives the blur for the claim.
Redact plate region only. Audio: -c:a copy (bit-identical, your narration preserved) No separate audio step needed.
Edge cases and what actually happens
Plate moves faster than one box can cover
By designOn a fast approach the plate grows and shifts across the frame. A single static box either over-blurs the surroundings or lets the plate poke out at the edges of the window. Split into shorter time ranges with a box sized for each, or draw one large box that covers the whole path.
Characters still readable after blur
Insufficient radiusIf the box is small, the boxblur radius is capped at min(width,height)/2, which may not be enough to scramble high-contrast plate characters. Enlarge the box — that raises the cap — and re-export. Always verify on the frame where the plate is largest.
Reflections or a second plate elsewhere in frame
PreservedOnly the rectangle you drew is blurred. A second vehicle's plate, or a plate reflected in glass, stays visible. Run the output through the tool again with a box over the additional plate.
GPS / location metadata survives
PreservedBlurring pixels does not touch container metadata. Dashcam files frequently embed GPS tracks and device IDs. If the clip will be public, run it through the metadata-scrubber after redaction.
Output container changed to MP4
By designDashcams that record TS or MOV will get an MP4 back. This is intentional for compatibility. If the insurer insists on the original container, transcode with the video-transcoder.
Large multi-hour dashcam file exceeds Free tier
Tier limitFree tier allows 1 GB per file; a long loop-recorded clip can be larger. Trim to just the segment showing the incident first with the lossless-trimmer, or upgrade (Pro 10 GB, Pro + Media / Developer 100 GB).
Re-encode softens the whole clip slightly
ExpectedThe full video stream is re-encoded at CRF 20, so even non-plate frames are recompressed once. CRF 20 is high quality, but if you need to minimise re-encoding, trim to the incident segment first.
Audio left intact when it shouldn't be
PreservedAudio is always stream-copied. If the clip also captured someone's name, phone number, or address spoken aloud, blurring the plate won't remove it — mute that span with audio-mute-region.
Frequently asked questions
Is blurring a license plate enough for GDPR compliance?
Obscuring the registration mark removes the most directly identifying element, and processing it locally (no upload) means the original never reaches a third party. Compliance also depends on context — other identifying details (faces, location metadata) may need handling too. Pair this tool with face-blur and the metadata-scrubber for a fuller scrub.
Will the plate be recoverable from the blurred video?
The boxblur destroys the pixels in that region — it is not an overlay. There is no hidden original layer to recover within the blurred box.
Why does a tight box over the plate still show the numbers?
The blur radius is capped at half the smaller side of the box, so a thin box can't blur very hard. Enlarge the rectangle around the plate and the cap rises, scrambling the characters. Always check the frame where the plate is biggest.
Does the dashcam clip get uploaded?
No. It is read into an in-browser FFmpeg.wasm filesystem and processed on your own machine. Nothing is sent to a server, which is what makes it safe for insurance and legal sharing.
Can I blur the plate only while the car is in shot?
Yes. Set Start (s) and End (s) to the seconds the plate is visible; the blur is gated by enable='between(t,start,end)' and the rest of the journey stays sharp.
What about the audio — does narration survive?
Audio is stream-copied unchanged, so any spoken account of the incident is preserved. If you instead need to silence a span, use audio-mute-region.
There are two plates in the same shot — can I do both?
One rectangle per pass. Process for the first plate, then run the resulting MP4 back through the tool for the second.
Does the tool track the plate automatically as the car moves?
No — the box is static for the whole window. For a moving target, draw a box large enough to cover its path, or split the redaction into multiple time-windowed passes. Automatic motion tracking is only available for faces via face-blur.
Will GPS data be removed?
No. Pixel blur doesn't touch metadata. Run the redacted file through the metadata-scrubber to strip embedded GPS and device tags before sharing publicly.
What output do I get?
An H.264 MP4 (CRF 20, preset medium) with the audio copied. Input MOV/TS/MKV files all come out as MP4.
My dashcam file is huge — will it process?
Up to 1 GB on Free, 10 GB on Pro, 100 GB on Pro + Media / Developer, with no cap on duration. For a very long loop file, trim to the incident first with the lossless-trimmer.
Can I prove exactly what was redacted?
Yes — note the box's X / Y / Width / Height (shown live as W × H @ (X,Y)) and the Start/End seconds. Those are the literal FFmpeg crop/overlay and enable values, so the redaction is fully reproducible and documentable.
Privacy first
Every JAD Video tool runs entirely in your browser via WebCodecs and FFmpeg (WebAssembly). Your video files never leave your device — verified by zero outbound network requests during processing.