Turn an existing photo of a document into a scan

The document is already with the landlord, in your parents' folder or in the recycling. What is left is a photo on your phone, taken at an angle on the desk, with a pen and the table edge in the picture. And the form wants a scan.

Scanner apps are little help here: most of them expect you to hold the sheet in front of the camera again. a4norm works on the photo you already have.

Is that even possible?

Yes, as long as the whole sheet is in the photo. This is what it looks like, a real shot and its result, unretouched:

Before: a phone photo off the desk. After: the A4 page from the bot

What happens along the way:

• the sheet is found in the picture and turned back into a rectangle;
• everything outside the sheet disappears: desk, pen, notebook spiral;
• uneven light is evened out and grey paper turns white;
• ink becomes neutral black, while a blue signature and a red stamp keep their colour;
• the page lands on exact A4, as a PDF of a few hundred kilobytes.

Some online tools make you drag the four corners by hand. Here the sheet is found automatically.

How to do it

1. Open the bot: @a4norm_bot. Free, no extra account, nothing to install beyond Telegram.
2. Send the photo from your gallery, ideally as a file (paperclip → File), so the original arrives without loss. A normal photo works too; then switch on HD when sending.
3. Several photos, one document? Select them together and send them as an album. You get one multi-page PDF in the order of the photos.
4. Collect the PDF. About half a minute per page; the bot writes when it is ready.
5. The form needs an image? /jpeg turns the last PDF back into pictures.

A photo that reached you through WhatsApp or another messenger is already compressed. That cannot be undone, but it is usually enough for a readable page. If someone still has the original shot, ask them to send it as a file.

When it does not work

• A corner is missing or text is cut off. What is not in the photo cannot come back.
• The photo is blurry. Blur stays blur, just on white paper.
• The flash washed out the text. That information is gone.
• No sheet can be found at all. Then nothing is straightened and the photo is simply placed on the A4 page. That is deliberate, so the bot can also put ordinary pictures into a PDF, but the result no longer looks like a scan.

So look at the PDF before you upload it. If you get the document back in your hands, a few habits when taking the photo make a clear difference.

Not a scan effect, a clean copy

Searching for "make a photo look scanned" also turns up tools that add noise, tilt and a grey cast to a file so that it appears scanned. a4norm does the opposite: it removes what makes a photo different from a scan and leaves the content as it is. The PDF shows exactly the sheet that was photographed.

It follows what it is not: not a certified copy, not an electronic signature, not text recognition. Whether an authority or an employer accepts the file is up to them.

Where does the photo go?

To the bot through Telegram, like any message. It is processed on my own server in Germany, not by a cloud provider or a third-party API, and the photos are not stored. The code is open, and there is also a Docker container that runs on your own computer with no network at all.