Scan Documents with Your Phone: From Crooked Photo to Clean PDF

Step-by-step

1. 1

   Get the lighting right before anything else

   Lighting decides more of the final quality than any software step. The best light is indirect daylight — a table near a window, but out of the direct sun patch. Turn the flash off: paper is glossy enough to bounce flash straight back as a white blob that destroys whatever text sits under it. Watch for shadows, especially your own — holding the phone over the page between the page and a ceiling light casts your silhouette across the document, and no filter fully removes a shadow gradient. If you can see a shadow edge on the paper with your eyes, the camera sees it worse. Move yourself, the page, or the light until the paper looks evenly lit edge to edge.

2. 2

   Flatten the page and pick a contrasting background

   Curled corners and fold ridges turn straight text lines into waves that perspective correction cannot fix — it corrects a tilted plane, not a curved one. Press folded documents flat under a heavy book for a few minutes, and weigh down stubborn corners with small objects kept outside the crop area. Then place the page on a surface darker than the paper — a dark desk, a navy folder, anything with clear contrast — because edge detection works by finding the boundary between page and background. White paper on a white table is the single most common cause of the scanner guessing the corners wrong.

3. 3

   Shoot straight-on and fill the frame

   Hold the phone parallel to the page, directly above its center, and fill most of the frame with the document while leaving a small border of background visible on all four sides — the algorithm needs to see the page edges to find them. A slight angle is recoverable; a steep one is not, because the far edge of the page ends up at a much lower effective resolution than the near edge, and correction stretches those few pixels into visible blur. Brace your elbows or rest them on the table to kill motion blur, and take two or three shots of important pages so you can pick the sharpest.

4. 4

   Let auto edge detection run, then verify every corner

   Load the photo into the Document Scanner and let it find the page boundary automatically. On a well-lit, well-contrasted shot it usually lands exactly on the corners — but check all four anyway, because a shadow, a table edge, or another sheet peeking into frame can pull a corner off the page. Drag any misplaced corner handle precisely onto the physical paper corner. Thirty seconds of corner-checking is the difference between a straight scan and one with a wedge of desk along one edge.

5. 5

   Apply perspective correction

   Once the corners are set, the tool warps the quadrilateral into a proper rectangle — this is the step that turns a phone photo into something that reads as a scan. Judge the result by the text: lines should run horizontally and letter shapes should look consistent across the whole page. If one region looks smeared or stretched, the usual culprit is a corner placed slightly off the true paper corner; go back and nudge it rather than accepting a warped page. Correction is geometry, so it is only as accurate as the four points you gave it.

6. 6

   Choose a filter that matches the document

   For ordinary text documents, a grayscale or high-contrast document filter is almost always the right call: it evens out lighting variations, pushes the paper toward clean white, and makes text darker and crisper — while shrinking the file. Keep color when the document contains meaning carried by color: official stamps, wet-ink signatures, highlighted passages, charts. Whatever you pick, verify at full zoom that the filter has not eaten the faintest content on the page — light pencil notes and pale seals are the first casualties of an aggressive contrast filter.

7. 7

   Export a multi-page PDF — and add OCR when it earns its cost

   Scan all pages in one session with the same lighting so the document looks uniform, check the page order, and export a single multi-page PDF rather than a pile of loose images — one file is what recipients, archives, and upload forms expect. If you will ever need to search the document's text, quote from it, or copy a reference number out of it, run the PDF through PDF OCR to add an invisible text layer. For a receipt you will attach once and never open again, skip it; for contracts, invoices, and anything you are archiving long-term, searchable text pays for itself the first time you need to find the document again.