Give a size limit and get a PDF under it, with the text untouched.
Or choose a file. It is opened and rewritten on your device and never uploaded. Text is never touched — only the images inside.
Working on your device…
Your PDF is opened and rewritten in this browser. It is never uploaded, and nothing about it is stored.
"Compress PDF" is almost always "the images inside this PDF are enormous". A phone photo or a 600 dpi scan dropped into a document arrives at several megabytes and is then drawn into a box a few inches wide, carrying five to ten times the pixels it can possibly show. Everything else in the file — the text, the fonts, the vector artwork — is already small.
So this re-encodes the pictures and leaves the rest alone. Give it a size limit and it works down through progressively smaller image settings until the document fits, stopping at the first one that does. Your text still selects, searches and copies afterwards, because it was never touched.
Job applications, court filings, visa uploads and university forms all cap attachments. Put the cap in and get a document under it.
Most mail servers stop somewhere between 10 and 25 MB. A scanned contract with photographs in it clears that easily, and almost all of the weight is the scans.
A ten-page scan can run to 40 MB because the scanner saved every page at print resolution. Nothing on screen changes when that comes down.
Tools that rasterise every page hit any size target easily and destroy the text doing it. This one refuses to — if the target needs that, it says so instead.
Drag a file onto the page or choose one. It is opened on your device and never sent anywhere.
You see the original size, the target, and what the document came out at. Text, fonts and vector artwork are never touched — only the images inside are re-encoded.
If the target cannot be reached without destroying the document, you are told how small it can honestly go rather than handed a rasterised page.
Yes. Only image streams are rewritten; text, fonts and vector artwork are copied through untouched. A page that could be searched and copied before still can be. This is the main difference from tools that flatten each page to an image.
You get it back unchanged. Re-encoding every photograph in a document that already meets the requirement would be a second generation of quality loss bought for nothing.
It tells you the smallest the document goes without rasterising the text, and suggests splitting the file. It will not quietly flatten your pages to hit a number.
Several kinds, deliberately. Bitonal scans (CCITT and JBIG2) are already the right format for one-bit text and would get larger and greyer as JPEGs. JPEG 2000 cannot be decoded reliably in a browser. Anything under about 12 KB is left because there is nothing to win.
No. The document is opened and rewritten in your browser using a PDF library that runs on this page. Nothing is transmitted, which also means password-protected files have to be unlocked in your own reader first.