Turning phone photos of paper into a PDF somebody can actually read

Most rejected document uploads fail for four reasons, and all four are decided before you convert anything. Light, angle, resolution, order.

Photographing paper and sending it as a PDF is now the normal way to submit a document. It is also where a specific set of failures cluster: the file is refused as illegible, the text will not extract, or the reviewer asks for it again. The conversion step is rarely at fault. What arrives at the converter is.

Light, and specifically your own shadow

The single most common defect is a shadow of the phone, or of the person holding it, falling across the page. Overhead room lighting guarantees it. Put the paper near a window with daylight coming in from the side, or turn on a desk lamp and place it off to one side rather than behind you. Avoid the camera flash: it produces a hot bright patch in the middle and dark corners, which is worse than the shadow it replaced.

Matte paper on a dark, plain surface photographs better than glossy paper on a white desk. Contrast at the edge of the page is what lets any subsequent cropping find the document.

Angle

Hold the phone parallel to the page and centred over it. A photo taken at an angle produces converging lines: text that is sharp at the near edge and progressively smaller and blurrier towards the far one. Text recognition tolerates a slight rotation well and a perspective skew badly, because the letter shapes themselves are distorted rather than merely tilted.

If your phone's document mode offers automatic edge detection and flattening, use it. It corrects the perspective properly, which no amount of later cropping can.

Enough resolution, and no more

The useful measure is not megapixels but pixels across the page. A full-frame photo of an A4 sheet from a modern phone comfortably exceeds what text recognition needs. Cropping tightly to the page before converting is therefore free: it removes the desk, and it raises the proportion of the frame spent on the text.

One PDF, in order, with the right format inside

Send one file, not eight images. A multi-page PDF in page order is what a reviewer expects, and a folder of photos named IMG_4471 through IMG_4478 is what causes documents to be reviewed in the wrong sequence. Rename the images numerically before converting if your tool sorts by filename.

Inside the PDF, format follows content. Photographs of paper are photographic images, so JPG to PDF keeps the file small and looks fine. Screenshots, or anything with crisp computer-rendered text, belong in a PNG to PDF page, because JPEG's blocking artefacts collect around sharp edges exactly where the text is.

If the original is already a Word file rather than paper, do not photograph it at all. DOCX to PDF keeps the text selectable and the layout intact, which is strictly better than a picture of the same page.

The check that catches everything

Open the finished PDF at 100% and read the smallest line of text on every page. Not skim, read. Account numbers, dates and signatures are the parts that matter and the parts that are hardest to reconstruct from a soft image. Thirty seconds of reading prevents the round trip that otherwise costs a week.

Caveats

• Requirements differ by institution. Where a specific resolution, page size or file size is published, follow that over any general advice here.
• This site produces one image per PDF page at the image's own dimensions rather than scaling to A4, so the page size of the result follows what you photographed.