How to use
- Add the original (A) on the left and the revised PDF (B) on the right. Enter a password if a file is protected.
- Choose Text to read both documents and find what changed. The summary shows words added and removed and the changed pages; click a page button to jump to that spot.
- Pick side by side (A | B) or unified. Removed text is red with a strikethrough, added text is green and underlined, and long unchanged stretches fold to two lines of context.
- To check layout or image changes, switch to Visual and look at the same page number from both files overlaid.
- Save the result as an HTML report or TXT (each line prefixed with
-or+).
Text compare or visual compare
| Text | Visual | |
|---|---|---|
| What it looks at | The text stored in the PDF | Pixels of the rendered page |
| Best for | Changed wording, amounts and dates in contracts and quotes | Stamps, images, table shape, layout |
| When pages shift | Follows, because the whole document is compared | Useless | it compares the same page numbers only |
| Scanned PDF (no text) | Nothing to compare | Works |
| Result | Word counts, changed pages, side-by-side or unified view, report | Percentage of different pixels, red highlights, overlay |
Visual compare renders both pages at 1,000 px wide and marks a pixel as different when the sum of its red, green and blue differences is above 48. A page that exists in only one file is compared with a blank page.
Options and how to read the result
- Ignore whitespace and line breaks (on by default): where PDF text extraction breaks lines depends on layout, so the same sentence can look different. Runs of spaces and line breaks count as one space. Turn it off to see line-break differences as changes.
- Ignore case: treats
Net 30 daysandnet 30 daysas equal. With whitespace ignored too,Net 30 days\nPayment dueagainstnet 30 days Payment dueshows 0 differences. - Compare by character: word mode is the default. Turn this on to catch a single changed letter.
- Word counts follow the pieces the diff engine produces. Changing
총 금액은 1,000,000원이며 납기는 10월 31일입니다.to총 금액은 1,200,000원이며 납기는 11월 15일입니다.splits numbers at commas and counts 3 added and 3 removed (000→200,10→11,31→15). In character mode you see only the changed digits. - Each PDF is compared up to 500 pages and 2,000,000 characters; beyond that only the beginning is used and you are told. A comparison that takes more than 10 seconds is stopped and explained.
Reports and caveats
The HTML report is a single static file. Every piece of document text is escaped, so a line such as <script> inside a PDF is shown as text and never runs. The TXT report shows a changed line as a - old line and a + new line.
- Text compare reads the text data in the PDF. Tables and multi-column layouts may be read in a different order than they look.
- Scanned PDFs have no text, so use Visual compare.
- This is a tool for finding candidate changes. For an important contract, check the changed pages yourself.
To pull out each file’s text on its own, use PDF to Text. For plain text, try Text Diff; for structured data, JSON Diff; and to just read a file, PDF Viewer.
Your files are processed only in your browser and are never sent to a server.