/ tools · Document Scanner
Document Scanner Beta
Point your camera at a page and it finds the edges automatically, straightens the perspective, and snaps a clean scan — like a hardware scanner in your pocket. Capture as many pages as you need, drop the bad ones, pick a filter, then download one tidy PDF. Everything happens in this tab; no photo is ever uploaded.
Beta: edge-detection and de-skew can still be unreliable depending on your device, camera and lighting. It works, but expect rough results in some conditions — improvements are on the way.
Looking for a page…
How the scan works
It finds the page, you don't. Each camera frame is edge-detected (Canny) and searched for the largest four-cornered shape — the page. A live outline tracks it; hold the phone still for about 2 seconds and the page is scanned automatically — then just place the next page. Prefer control? Untick Auto and tap the shutter to review and adjust the corners yourself.
Shot at an angle? Fixed. The four detected corners drive a perspective transform that flattens the page into a straight-on rectangle, cropping away the desk behind it — the same de-skew a real scanner does.
Scanner filters. Per page, choose Color, Grayscale, or high-contrast B&W (adaptive threshold) so text pages come out crisp and printable. Reorder or delete any page before exporting.
Stays on your device. The camera feed is processed frame-by-frame in this tab and the PDF is built locally with pdf-lib — no frame, photo, or page is ever uploaded. Needs a camera and a secure (https) page; works best on a phone. The edge-detection engine (OpenCV.js, ~11 MB) loads once when you start the camera.