Compare PDF versions
Compare two versions of a PDF page by page and see which pixels and which text changed, with the differences highlighted.
- Runs in your browser
- No upload
- Check before download
Local file
Local files
Files and content stay on this device. Aggregate action counts are sent only with your separate consent. Plain unencrypted PDFs only: 10 MB/file, 20 pages, 10000 objects, no object streams, signatures, forms or complex navigation. Text and vectors remain. Page numbering supports unrotated pages. Visual pixel and extracted text comparison; geometry changes are reported separately. Text reading order may differ. No signature or semantic/legal verification. Two PDFs, 10 MB each, 20 pages each, 16 million pixels each.
Settings
Result
Local file
Compare PDF versions
- Prepare the input
Choose two supported PDFs in before and after order, and set the visual tolerance. The tool compares raster pixels and extracted text, and reports changes in page geometry separately. Keep both originals so you can interpret the result.
- Try the worked example
If the page before says Alpha and the page after says Beta, the text report should show the change. Identical page pixels don't mean the metadata is identical, and a changed page size counts as a separate geometric difference.
- Check and save
Check the marked pages and the downloadable text report against both files. A difference in rendering or text order needs a person to interpret it. This is not a certificate that the files are equivalent in signature, meaning, legal terms or accessibility.
Questions
Does a visual match prove the documents are identical?
No. The comparison covers only the pixels, the extracted text and the page geometry. An unchanged picture doesn't tell you anything about hidden objects, metadata or legal meaning.
Why is the text order surprising?
PDF text extraction follows how the content is stored, which may differ from the order you read it in. Review the source pages before you treat text that seems moved or reordered as a real editorial change.
What are the input limits?
Two plain, supported PDFs, each up to 10 MiB and 20 pages, with 16 million pixels per input. Encrypted, signed or unsupported structures are refused, and raising the tolerance doesn't make them valid.