RedactCam detects faces, license plates and ID documents in your photos — and faces and plates in dashcam video — then pixelates them, entirely inside your browser. Works in airplane mode. Strips GPS metadata on export. Free.
…or just drop a photo or clip onto this frame — it opens straight in the app.
The detection models download to your browser once (~32 MB) and run on your own hardware — GPU when available, CPU otherwise. Your files never travel.
Drag in one photo, fifty, or a dashcam clip. Faces, license plates and ID documents are detected automatically — on your machine, even offline.
Every detection is a box you can toggle, move, or resize. Missed something? Drag to add a region — IDs, screens, tattoos, anything.
Pixelation is written into the pixels at full resolution — not a removable layer. Re-encoding strips EXIF and GPS metadata as a side effect.
Tools that blur plates in video usually come with a sales call and a four-figure annual invoice — because they run on someone else's servers. RedactCam runs on yours, so video anonymization can just be… a button. Faces and plates are tracked across frames and pixelated; audio is kept; the result downloads as a normal MP4.
Almost every "AI" tool works by uploading your photo to a server farm. RedactCam flips that: the detection models — small neural networks, about 32 MB in total — download to your browser once and run on your own GPU. Same AI, but your files never leave your device. It may be the only AI tool you can run in airplane mode. (It detects and blurs — it never reads or keeps your documents.)
"We don't upload your files" is a promise every service makes. RedactCam is built so you don't have to take the promise on faith:
Network tab.
Process a photo — you'll see the models download to you and nothing leave.Protect sources and bystanders before material moves through email, CMSes, or cloud drives. No third party ever holds the original.
Blur plates and faces in claim photos and dashcam video before they enter your workflow — batch by batch, one zip out.
Selling a car, sharing a school event, posting a near-miss clip — pixelate the people who didn't ask to be famous.
Yes. Because processing happens on your device, each user costs us almost nothing — which is exactly why we can afford to be generous. If power features ever cost money, the core promise (local, private, verifiable) stays free.
They are opened by your browser, processed in your browser's memory, and saved back to your downloads folder. There is no server component: nothing is uploaded, logged, or stored by us — we couldn't look at your files if we wanted to.
No detector is perfect, and we'd rather tell you that on the front page than in fine print. Detection is tuned to over-blur rather than miss, and for photos you always review before export — anything missed is one drag away. For video, spot-check the result.
A modern browser — Chrome or Edge get GPU acceleration, Safari and Firefox work too. Photos: JPEG/PNG, exported at full resolution. Video: MP4/WebM in, MP4 out, audio preserved.
The default is coarse pixelation, chosen because weak gaussian blurs can sometimes be partially reversed. Pixelation at this block size destroys the information — and it's baked into the image pixels, not layered on top.