Compare PDFs

Each page of A and B is rendered to a bitmap at the chosen DPI. The tool walks pixel-by-pixel across both renders; any pixel whose RGB distance exceeds the threshold is flagged red in a diff image. Pages that don't overlap (if page counts differ) are flagged in the summary and skipped.

Example

Compare: contract-v4.pdf (12 pages) vs contract-v5.pdf (12 pages)
DPI:     100
Output:  3 pages differ, 9 identical.
         Page 7: 4.2% of pixels changed
         Page 8: 0.8% of pixels changed
         Page 11: 11.6% of pixels changed

Common errors and pitfalls

Most issues come from how PDFs are rendered, not from actual content differences. The summary line shows what the tool found.

• "Identical" PDFs reporting differences. Two PDFs can render the same page with different anti-aliasing or font hinting — visually identical, pixel-different. Bump the threshold slider to 12–16 to filter out sub-visible diffs.
• Out-of-memory on large documents at high DPI. Each rendered page at 300 DPI on a letter-size page is ~2500×3300 = 33 MB of canvas data, times three (A, B, diff), times every page. Keep DPI at 100 for anything over 50 pages.
• Rotation or scale differences. If one PDF has pages rotated or at a different page size, the tool compares by pixel alignment and everything looks different. Normalise page geometry first with the PDF Rotate or PDF Organizer tool.
• Encrypted PDF. Password-protected PDFs are refused. Open the file in any PDF viewer, remove the password via File > Save As, and re-drop the unprotected copy.
• Different page counts. If A has 12 pages and B has 10, the tool compares the first 10 pages and notes the extras in the summary. Extra pages are not lost, they simply have nothing to compare against.
• Wrong file dropped. "Not a PDF — drop a .pdf file." A .docx or .pages file is rejected. Convert to PDF first — the tool only accepts real PDFs.

Is this a text diff or a visual diff?

A visual diff. Each page of each PDF is rendered to a bitmap at the chosen DPI, and every pixel that differs between the two is highlighted in red. This catches layout changes, image swaps, font substitutions, and content edits — anything visible. For a text-only diff, extract text from each PDF with /pdf-to-text and paste into /diff-checker.

What DPI should I use?

100 DPI is the default — fast and catches most meaningful differences. Bump to 150 for finer changes (1-pixel line shifts). 200+ mostly wastes time unless you're comparing highly detailed technical drawings.

What happens if the two PDFs have different page counts?

The tool compares the overlapping page range and flags the extra pages in a summary line ("A has 12 pages, B has 10 — comparing pages 1–10"). Extra pages aren't lost, they just aren't compared against anything.

Why is my "identical" page showing differences?

PDFs can encode the same visible page in different ways — font embedding, image compression, or anti-aliasing hints — that produce pixel-level differences at rendering time even when the content looks identical. Bump the similarity threshold slider up or rely on the percentage rather than the red pixels.

Can I compare encrypted or password-protected PDFs?

No. Open the file in any PDF viewer, remove the password via File > Save As, and re-drop the unprotected copy. The tool refuses encrypted files to avoid silently rendering only the parts that don't need the password.

Do you save the PDFs I compare here?

No. We don't save the PDFs you drop, the rendered images, or the diff result. Everything is discarded the moment you close or refresh the tab. You can verify in your browser's developer tools.