Scan a document with your phone without installing anything

You need to send an ID card, an invoice or a receipt, and you have no scanner. You search the app store for “scan document” and twenty results come back, all free, all highly rated. You install one, scan, and somewhere between the watermark and the weekly subscription you find out your document has been through a server you know nothing about.

Now you can do it without installing anything: Scan to PDF.

How it works

Open the tool in your phone browser, tap Take a photo and your normal camera opens. One photo per page, reorder them if you need to, download the PDF. Nothing leaves the phone.

Using the system camera is deliberate, rather than embedding a video viewer in the page. Your phone’s camera app focuses better, meters light better and uses everything the manufacturer put in there. A video element inside a web page will never match that.

Why a photo is not a scan (and what we do about it)

Drop the photo straight into a PDF and it shows: the paper comes out grey or yellowish depending on the bulb above you, there is a shadow down one side and the text looks flat. It looks like what it is — a photo of a piece of paper.

The tool treats each image before assembling it:

• Converts to grayscale by luminance.
• Reads the actual histogram of that photo and stretches the range: the grey of the paper goes white and the ink settles into black. It is not a fixed filter, it is computed per image, which is why it works the same in evening light and under a fluorescent tube.
• Scales the image down to something sensible, because a 4000-pixel photo times ten pages is tens of megabytes for nothing.

Three treatments are available: grayscale (the one that looks like a scanner), pure black & white (the lightest file, ideal for text) and colour, for when colour genuinely matters.

Getting a good result

• Put the document on a flat surface with even light. The shadow of your own hand is enemy number one.
• Frame the whole sheet without cropping it, straight from above. Shoot at an angle and the text comes out trapezoidal.
• One photo per page, in order. If you get it wrong you can reorder or delete before generating the PDF.

What it does not do

It does not detect edges or correct perspective automatically. That is a deliberate choice: those algorithms fail precisely when you need them most — pale documents on a pale table — and end up cropping things they should not. I would rather you frame it and the result be predictable.

And the part that matters

The document you are scanning tends to be exactly the kind you would rather not hand over: an ID, a payslip, a contract. Here the photos are processed on your phone and the PDF is saved on your phone. Put it in airplane mode after the page loads and it still works.

Scan to PDF