ImgVX AI

Scan documents with your phone: from photo to clean PDF

Advertisement

A phone camera is a perfectly good document scanner. The difference between a photo of a page and a scan is mostly four steps: light, straighten, crop, and save at the right size.

1. Take a better photo

Everything later is easier with a good starting image, and no software fully fixes a bad one.

2. Straighten the page

A page photographed at even a slight angle comes out as a trapezoid, wider at the bottom than the top. Rotating does not fix that; it needs a perspective correction, which pulls the four corners back into a rectangle.

In the perspective corrector, drag the four handles onto the four corners of the page and it squares the image up. This one step does more for how "scanned" a document looks than anything else. If the page is merely rotated, the rotator is enough.

3. Crop and clean up

Trim the table and anything else outside the page with the cropper. If the paper looks grey or yellow, raising contrast and brightness slightly in image adjust gets it closer to white without losing faint text. Do not push it until the paper is pure white; pencil marks and light print disappear first.

4. How much resolution you need

Print and scanning are measured in pixels per inch. For documents, the useful numbers are:

PurposeResolutionA4 page in pixels
Reading on screen, email150 DPI1240 × 1754
Clear text, reprinting200 DPI1654 × 2339
Text recognition (OCR), archiving300 DPI2480 × 3508

A modern phone photo of a full page is comfortably above 300 DPI, so detail is rarely the problem. File size is. A 12-megapixel photo per page makes a 20-page PDF enormous. Resizing each page to around 1700 pixels tall is plenty for a document that will be read on a screen. Our DPI guide explains the arithmetic.

5. Make the PDF

Add the pages, in order, to the image to PDF tool. Choose the paper size the recipient uses — A4 almost everywhere, US Letter in the US and Canada — and a small margin. Each image is fitted to the page without distortion.

Everything happens in your browser. For identity documents, bank statements and medical letters, that matters: the pages are never uploaded anywhere.

If the PDF is too large for an upload form, the usual cause is page images far bigger than needed. Resize them, or compress them with the image compressor, before building the PDF.

6. Getting the text out

A PDF made from photos is still pictures of text: you cannot search it or copy from it. Optical character recognition fixes that.

Run a page through image OCR and choose the document's language. Accuracy depends almost entirely on the input. The Tesseract engine it uses works best at around 300 DPI, and its accuracy falls off sharply for text smaller than about 10 point at that resolution. Straight, evenly lit, high-contrast pages recognise far better than curled or shadowed ones — another reason steps 1 to 3 are worth the minute they take.

OCR is never perfect. Check names, numbers and anything that matters against the original before relying on it.

Going the other way

If you have a PDF and need its pages as images — to crop a signature, attach a single page, or post it somewhere that only takes pictures — the PDF to image tool exports each page as PNG or JPG.

Advertisement