PDF Compare
Spot visible differences between two PDFs. It runs in your browser.
Compare rendered pages side by side and highlight the pixels that differ.
Files stay in your browser.
Comparison results
Rendered pixels only. This does not compare meaning, hidden objects or document structure.
Choose an original and a revised PDF to begin.
Magenta marks differing pixels; pale gray provides original-page context. An overlay blends both pages equally. Previews fit this panel; open the downloaded PNG to inspect full resolution.
Download results
The ZIP holds every pair’s difference image and the JSON report.
How to compare
- Choose both PDFs and enter passwords if needed.
- Compare all pages by number, or map selected pairs. Adjust DPI, tolerance and revised rotation.
- Compare, inspect each pair and download the PNG differences or JSON report. ZIP includes every selected pair’s difference and the report.
What is compared
Pages are rendered on white at the same DPI, with saved rotation, crop bounds, default visible layers and saved annotation appearances. Different sizes align at the top left without stretching; areas present on only one page count as differences.
Color, font or rendering variations can produce pixel differences without a change in meaning. Additional rotation turns the rendered revised image; font antialiasing differences can remain after alignment. Zero differences means equal rendered pixels at these settings, not identical PDFs.
Frequently asked questions
Are my files uploaded?
Rendering, comparison and report preparation run in browser workers. File contents, names and passwords are not uploaded, logged or stored. Only this site’s code and renderer assets load.
Cancel stops processing. Reset releases file and download references. Original PDFs stay unchanged.
Can this find inserted pages or text changes automatically?
Same-number matching pairs page 1 with 1, page 2 with 2, and so on; extra pages are unmatched. Custom pairs let you align a shifted section or select only relevant pages. For an insertion use, for example, 1:1, -:2, 2:3. There is no automatic content alignment, OCR or text diff. Scanned pages are compared as rendered images.
What is excluded?
Attachments, metadata, hidden layers, tags, links, scripts and unpainted objects are not compared. Forms use their saved appearances; interactive controls, comment threads, multimedia and XFA-only forms may not render as in a specialist viewer. Both files can look the same while differing in these features.
How should I choose resolution and tolerance?
At 144 DPI a 72-point edge becomes 144 pixels. Higher DPI reveals finer detail and uses more memory. A pixel differs if any RGB channel changes by more than the tolerance (0–255). Antialiasing and color conversion can affect results. Page-size comparison follows the rendered pixel dimensions, rounded up, so subpixel size changes may not be detected.
What if processing fails?
Check passwords and page numbers. Damaged PDFs may need to be exported again. Large documents or high DPI can exhaust your browser’s memory; lower DPI or compare fewer custom pairs. Cancel remains available during processing. The tool requires workers, OffscreenCanvas and browser image decoding. ZIP has the existing classic ZIP format limits; individual downloads remain available if archive preparation fails.