Combining scanned pages into one clean document

Combining scans is where several small problems compound: inconsistent capture, wrong orientation, duplicate sheets, and a final file too large to email. Doing the steps in the right order avoids nearly all of it.

Capture consistently first

Most of the quality of the final document is decided before any tool is involved.

If you are using a phone, the document scanner is a better starting point than the camera app: it captures at a fixed aspect ratio and applies contrast handling suited to paper rather than to scenery. Keep the sheet flat, get even light without a hard shadow from your own hand, and hold the phone parallel to the page — a tilted capture produces keystoning that no later step fully fixes.

If you are using a flatbed or a document feeder, 300 DPI is the sensible default for text you may later want to OCR. 600 DPI doubles the linear resolution and quadruples the file size for very little recognition benefit on ordinary print. 150 DPI is fine for pages that will only ever be read on screen and never recognised.

Scan in greyscale for black-on-white documents. Colour triples the data for no gain unless the page actually has meaningful colour — a stamp, a signature in blue ink, a highlighted section.

Get the order right before assembling

Sort the source files while their names still mean something. If you captured with a phone, the filenames are timestamps and are already in capture order — which is correct if you photographed the stack in sequence, and wrong if you did not.

Rename to 01, 02, 03 with leading zeros. It takes a minute and removes the entire class of ordering mistakes described in the merge guide.

Assemble

If your captures are images, JPG to PDF lays them out one per page and lets you set the order before conversion. iPhone captures are typically HEIC rather than JPEG; HEIC to PDF handles those without a conversion step in between.

If your captures are already PDFs — one per scanner run — use Merge PDF.

Either way, confirm the sequence in the tool's list rather than assuming.

Fix orientation before, not after

A sideways page is obvious to a human and invisible to software. If one source is rotated, fix it with Rotate PDF before assembly, while you still know which file it is. Hunting for "the sideways one" inside a 60-page merged document is tedious.

Then compress once

A stack of 300 DPI scans adds up fast: twenty pages can easily reach 40 MB. Compress PDF re-encodes the embedded images at a lower quality and downsamples oversized ones.

Do this after merging, not per file. One compression pass means one quality decision, applied consistently. Compressing each source and then merging stacks the losses unevenly across the document.

Check the result at 100% zoom before deleting anything. Compression artefacts in scanned text look like softened or smeared letter edges, and they are much easier to spot on the actual pages than to predict from a percentage setting. There is more detail in reducing PDF size.

And OCR once, at the end

Running OCR on the finished document is faster than recognising each source separately and produces one searchable file. Recognition works on the page images as they exist at that point, so run it after compression if you are going to compress — but be aware that aggressive compression damages recognition accuracy. If the document matters, keep an uncompressed master, OCR that, and compress a copy for distribution.

A complete sequence

1. Capture at 300 DPI greyscale, flat and evenly lit.
2. Discard bad captures now.
3. Rename with zero-padded numbers.
4. Fix rotation on any offending source.
5. Assemble with JPG to PDF or Merge PDF.
6. Check the joins and the page order in the output.
7. OCR if the document needs to be searchable.
8. Compress a copy for sending; keep the master.

For turning a whole filing cabinet into a searchable archive rather than one document, see paper to searchable archive.