Scanned document to editable

Quality is the whole story

A 300 dpi scan of a printed original reads almost completely. A photocopy of a photocopy loses thin rules and small print, which is where footnotes, footers and table borders live. A page fed through a scanner at an angle is the worst case, because skew makes columns drift into each other and a table can come back as scrambled rows.

Stamps, signatures and handwriting

A received stamp, a wet signature or a handwritten margin note is a mark on the image, not content. Say in the prompt that they should be ignored, otherwise a stamp sitting across a heading can confuse the read of that heading. A signature is best rebuilt as a proper signature block rather than carried across as a picture.

Fix the source before you fix the output

If the scan is poor and the paper original still exists, rescan it flat at 300 dpi. Ten minutes at the scanner beats an hour repairing forty blocks, and the result is better in every case. If only a bad copy exists, expect to check tables and any small print by eye against the paper.

Rebuild, then improve

Old scanned documents usually carry decades of accumulated structure: numbering that restarts halfway, three heading styles doing the same job, a table that exists because the typewriter could not indent. Once it is editable, that is worth cleaning up. The rebuild is the moment those choices are cheapest to change.

Your scan is downsampled before anything reads it

Whatever you upload, each page is rendered at roughly 200 dots per inch and handed over as an image. A scan is already a picture, so nothing is lost in that step, but it does mean the ceiling on accuracy was set by your scanner rather than by anything here. Up to a hundred pages go through at once. The practical consequence is that the time to spend is on the capture: a flat page, even light and sharp edges change the result far more than anything you can do afterwards.

Labelled boxes survive as pairs

A scanned form or invoice is mostly labels and the things written beside them, and those come back identified: each label as printed, each value exactly as it appears, and an empty string where the box was left blank. So a stack of scanned records turns into something with reference numbers, dates and amounts you can actually pick out, rather than a wall of recognised text you would have to read to find anything in.

How it works, in three steps

1. Step 1

   Scan flat at 300 dpi

   One multi page PDF, square to the glass, no skew.

2. Step 2

   Name what to ignore

   Stamps, signatures and margin notes are marks on the image, not content.

3. Step 3

   Check tables and small print

   Those are the two things a scan loses first.

The full walkthrough with screenshots is in the guide Turn a PDF or photo into an editable document.

Limits worth knowing

• A scan is read as an image, so accuracy depends on the scan.
• Skew, stamps and photocopy loss all cost fidelity.
• Handwriting is not read as content.
• Attaching files is a paid plan feature.

See it on a finished piece

Company policy template, built out as an information security policy

15 pages

Board meeting minutes template with actions and owners

3 pages

Employee handbook

6 pages

Questions people ask

Related pages