Compress PDF Free Online — No Upload
The document is opened and edited in this tab. It stays local — a contract or a scan is never uploaded to be processed.
Loading the tool…
How it works
What compressing a PDF can and cannot change
A PDF's bulk is almost always its image XObjects — each one a stream of pixel data with a /Filter saying how it is encoded. A phone scan is typically a 3000-pixel-wide JPEG per page, stored as /DCTDecode at high quality; a screenshot is usually /FlateDecode, which is lossless. This tool decodes each image in the browser, scales it down to no more pixels than the largest page can show at the preset's resolution, and re-encodes it as JPEG. The new image replaces the old one only when it is at least 10% smaller. Transparency lives in a separate /SMask object, which is kept as it was. Three lossless passes follow: objects nothing in the document refers to any more, which incremental saves leave behind, are removed; streams stored with no compression at all are deflated; and the remaining objects are packed into compressed object streams. Fonts, text and vector content streams are not touched, which is why text-only files hardly change. Tools that report 80% savings on a text PDF are usually rasterising every page, which wrecks the text layer.
- Image resolution
- Light caps images at about 200 pixels per inch of the largest page, Balanced at 150 and Strong at 100. An image already below the cap keeps its size and is only re-encoded.
- JPEG quality
- 80% for Light, 70% for Balanced and 55% for Strong. Each re-encode loses some detail, so an image is only replaced when the saving is at least 10%.
- Lossless images
- Screenshots, charts and logos stored with Flate stay lossless under Light and Balanced. Only Strong converts them to JPEG, where sharp edges and flat colour soften.
- Images left alone
- CMYK and ICC-profiled colour, palette images, masks, JPEG 2000, JBIG2 and CCITT fax scans. The browser cannot round-trip them faithfully, and the report counts each kind.
- Unused objects
- Superseded images, fonts and form appearances left behind by incremental saves are found by walking the document from its catalog, then deleted. Bookmarks and form fields stay.
- Text and fonts
- Unchanged. Text stays selectable and searchable, links keep working, and embedded fonts are not subset or removed.
How to use it
How to compress a PDF online
- 01
Open the PDF
Read from disk with the File API, up to 100 MB. Nothing is sent anywhere.
- 02
Pick Light, Balanced or Strong
Balanced suits most scans. Use Strong for an attachment limit, and Light when the images need to stay print-sharp.
- 03
Compress and check the report
You see the before and after sizes, how many images were recompressed or skipped and why, and how many unused objects were removed.
- 04
Download the smaller copy
Offered only when it is meaningfully smaller than the original. Your original file stays on disk unchanged.
Where it earns its keep
Where shrinking a PDF actually pays off
- Getting a scanned contract or a bundle of phone-photographed receipts under an email attachment limit.
- Meeting an upload cap on a visa, job or university application portal.
- Shrinking a slide deck exported with full-resolution photos before sharing it.
- Clearing the leftovers of many incremental saves from a form that has been edited for years.
Questions
Compress PDF, answered
Why did my PDF barely get smaller?
Because its size is not in images. A text-only or vector PDF is mostly fonts and drawing instructions, which are already compressed and which this tool does not rewrite, so it typically drops by single-digit percentages. A PDF whose images are CMYK, JPEG 2000 or black-and-white fax scans also changes little, because those images are left alone. The report shows which case applies.
Is my PDF uploaded to be compressed?
No. pdf-lib parses and rewrites the document, and the images are decoded and re-encoded by the browser's own canvas, all inside this tab. Open your network inspector while it runs: no request carries the file.
Will the text still be selectable?
Yes. Only image streams are re-encoded. Text, fonts, links, bookmarks and form fields are copied through as they were, so the result can still be searched and read by a screen reader.
What happens to a digitally signed PDF?
You are asked first. Any change to a signed file invalidates its signature, and compressing rewrites the whole file. If the signature matters, keep the original and compress a copy only for sharing.
Is a PDF/A file still PDF/A afterwards?
It is saved the conservative way, without object streams, which PDF/A-1 forbids, and its XMP metadata is left uncompressed. Conformance is not re-validated, so if an archive requires PDF/A, check the output with a validator before submitting it.
Can it compress a password-protected PDF?
No. An encrypted document's streams cannot be read without the password, so the file is refused with a clear message rather than producing something broken. Remove the password in the application that set it, then compress.
How this compares
- Compress PDF compared with iLovePDF — where your data goes, the limits, and where iLovePDF does more.