Howtoscandocumentsproperlywithaphone

Light beats everything else

Daylight near a window, with the page flat and your body not casting a shadow across it, produces a better result than any app setting. Overhead room lighting puts your own shadow on the page; a flash blows out the middle of anything glossy and reflects off lamination.

If you have to use artificial light, get two sources from different angles, or move the page rather than the light.

Frame it square, fill the frame

• Shoot straight down, not at an angle. A skewed photograph stays skewed in the PDF, and cropping cannot un-skew it.
• Fill the frame with the page. Every pixel spent on your desk is a pixel not spent on the text, and it makes the file larger for no benefit.
• Rest the phone on something if you can. Camera shake is the difference between text that reads and text that has to be guessed at.

Resolution: more is not better

A modern phone photographs an A4 page at far higher resolution than any portal needs, which is why the file is eight megabytes. For a document, 150 to 300 DPI equivalent is plenty — the text is completely legible and the file is a fraction of the size.

If your phone has a document mode, it is usually doing exactly this: correcting the angle, cropping to the page, flattening the lighting and saving at a sensible resolution. It is worth using.

Turning the photos into one document

Fifteen photographs are fifteen attachments, and no form wants that. Convert them to a single PDF, one image per page, in the order the document reads.

Choose A4 if it will be printed or submitted to a portal expecting a standard page; choose fit-to-image if it is staying on screen and you would rather not have white bars. Then compress the finished PDF once, rather than shrinking each photograph first.

What a phone cannot do

It cannot produce a text layer. A photograph of a page is a picture, so the result is not searchable and text cannot be copied out of it — extracting text from it needs OCR, which is a separate operation.

For anything where the text itself matters — a document you will quote from, or a form someone will run a search against — a real scanner or an OCR step is worth the extra few minutes.