How to scan documents with your phone and get a PDF worth sending

The request arrives constantly — a signed form, proof of address, a passport page, a receipt for an expense claim — and it always uses the same word: scanned. Almost nobody owns a scanner any more. Almost everyone owns a phone that has a genuine document scanner built into it, and most people have never opened it, because it is not in the camera app and nothing on the home screen advertises it.

The distinction matters more than it sounds. A photograph of a page and a scan of the same page differ by more than a filename. One is a picture of a document taken at a slight angle in whatever light happened to be in the room; the other is a corrected, flattened, normalised reproduction that a human reviewer or an automated system can actually process. This guide covers the built-in scanners on both platforms, the technique choices that decide the result, and what to do with the PDF once you have it.

Why a photo of a page is not a scan

Hold a phone over a sheet of paper and you cannot get it perfectly parallel and perfectly centred. You are always slightly off-axis, which means the rectangle of the page arrives in the photo as a trapezoid — wider at the near edge, narrower at the far one. Your eye corrects for this automatically, so the photo looks fine to you. It does not look fine to the person receiving it, and it looks considerably worse to any software that has to read it.

A scanner mode fixes four things a photograph leaves broken:

• Perspective. It detects the four corners of the page and warps the image so the page is genuinely rectangular again. This is the single biggest visible difference.
• Cropping. It removes the desk, the tablecloth, your hand and the shadow at the edge of the frame, leaving only the page.
• Lighting. It normalises exposure across the page, so the corner nearest the window stops being blown out while the opposite corner stays grey.
• Contrast. It pushes the paper towards white and the ink towards black, which is what makes text sharp rather than muddy — and what makes character recognition work later.

The practical consequence: a scanned page reads cleanly at a glance and survives being printed, while a photographed page announces itself immediately as a phone snapshot. For a landlord or an admissions office that is merely unprofessional. For an insurance claim or a legal filing it can be grounds for rejection.

Scanning on an iPhone or iPad

iOS has had a real document scanner since iOS 11, tucked inside the Notes app where nobody looks for it. There is also a route through the Files app, which is the better one when you want a PDF and nothing else.

The Files app route (produces a PDF directly)

The Notes route (better for keeping the scan with other material)

Open Notes, create or open a note, tap the camera icon above the keyboard and choose Scan Documents. The capture experience is identical. The difference is where it ends up: embedded in the note rather than saved as a file. To get a PDF out, tap the scan, then Share → Save to Files. Use this route when the scan is part of a wider set of notes; use Files when you just want the document.

Scanning on an Android phone

Android's most dependable scanner is inside Google Drive, which ships on nearly every device and does not depend on the manufacturer's own camera software.

Samsung, Xiaomi, OnePlus and Pixel devices often have a scanner in their own gallery or camera app as well, and the Google Photos app can detect documents in the viewfinder. These work, but they move between versions and skins, and several of them save images rather than PDFs. Drive is the route that behaves the same on every Android phone, which is why it is the one worth remembering.

The five things that decide quality

The software handles perspective and contrast. Everything that remains is physical, and it is worth thirty seconds of setup because no amount of post-processing recovers a badly captured page.

One habit is worth more than all five: check the result at full zoom before you send it. Pinch into the smallest text on the page. If you cannot comfortably read it on your own screen, neither can the recipient, and neither can any software asked to process it. Rescanning takes fifteen seconds; being asked to resubmit takes a week.

Multi-page documents and page order

Both built-in scanners keep adding pages to the same document until you tell them to stop, which is the correct behaviour and the reason to use them rather than the camera. A six-page contract should arrive as one six-page PDF. Six separate files is six things for the recipient to lose track of, and many upload forms accept only one attachment anyway.

Scan in reading order and the problem never arises. If you do end up with pages out of sequence, or with an extra page you scanned twice, fix it with Organize PDF — it lets you reorder, rotate and delete individual pages without rebuilding the document. If you finished with several separate PDFs that should have been one, Merge PDF joins them in the order you set.

If you photographed everything rather than scanning it and now have a camera roll full of JPGs, you are not stuck: Image to PDF takes multiple images at once, lets you drag them into order, and produces a single document. The perspective correction is the part you cannot retrofit, so the result will not match a proper scan — but the structure will be right. The full walkthrough for that case is in how to convert an image to PDF.

Making the scan searchable

This is the step almost everyone skips, and it is the difference between a PDF that merely exists and one that is genuinely useful. A scan is a photograph of text, not text. You cannot select a sentence, search for a name, or copy an account number out of it. To your computer it is a picture with no more meaning than a photograph of a beach.

Optical character recognition solves this by reading the characters in the image and writing an invisible text layer underneath them. The page looks exactly the same and behaves completely differently — searchable, selectable, copyable, and readable by a screen reader. Run the finished scan through OCR PDF and it takes seconds.

Three reasons this is worth doing even when nobody asked for it: you will want to find the document again in two years and will search for a word in it rather than remember the filename; recipients frequently need to copy a reference number rather than retype it; and a growing number of institutional upload systems reject image-only PDFs outright because their own processing cannot read them.

OCR accuracy depends almost entirely on the quality of the capture, which is the reason the technique section above matters. Clean, high-contrast, properly squared text recognises at very high accuracy. A dim, angled photograph of the same page produces a text layer full of errors — worse than useless, because searches then fail silently on words that look present.

When the file is too big to upload

A phone camera sensor produces a great deal of data. A ten-page scan can easily land between 20 and 50 MB, which is past the attachment limit of most email providers and well past what many government and university upload forms accept.

The first fix is at capture time: if the document is plain black text on white paper, use the black & white or greyscale filter rather than colour. A bitonal scan of a text page can be a tenth the size of the colour version with no loss of anything you care about. Keep colour for anything where the colour carries meaning — a stamped seal, a signature in blue ink, a highlighted passage, a photograph.

For a document you have already saved, run it through Compress PDF, which re-encodes the embedded images without disturbing the page structure. Aim for under 10 MB as a general target and under 5 MB for older institutional portals, which are frequently stricter than their instructions admit. If it still will not fit, Split PDF divides it into parts that can be sent separately.

One ordering detail worth knowing: run OCR before heavy compression, not after. Aggressive compression softens the character edges that recognition depends on, so the same page scanned identically will OCR less accurately once it has been squeezed. Scan, OCR, then compress.

When a phone is not good enough

Phone scanning has become genuinely good, and for the overwhelming majority of everyday requests it is indistinguishable from a flatbed scan. There are still cases where it is the wrong tool:

• Bound books and thick documents. The page curves away at the spine and no correction algorithm reconstructs the text in that gutter. A flatbed with the book pressed flat is the only good answer.
• High-volume batches. Fifty pages one at a time is a genuinely unpleasant twenty minutes. A sheet-feed scanner does it unattended.
• Anything requiring certified reproduction. Some legal and medical processes specify scanner output at a stated resolution, and a phone capture will not satisfy the requirement regardless of how good it looks.
• Fine detail and archival work. Photographs, artwork, engineering drawings and anything where fine tonal gradation matters need real optical resolution and a controlled light source.

For a signed form, a utility bill, a receipt, a certificate or a contract — which is what the request almost always is — the phone in your pocket is the right tool, and it has been for several years.

Frequently asked questions