How this PDF compressor works
Drop a PDF, decide what happens to the pages, download the result. Two JavaScript PDF libraries do the work inside the tab, so your payslip never leaves your laptop.
Keep text reloads the document and writes it back with a tighter internal layout: objects packed into compressed streams, cross-reference table collapsed. Pages, fonts and text come through untouched. Across nine documents (research papers, a slide deck, NVIDIA's CUPTI manual, a conference talk) it saved between nothing and 29%, and three came back byte for byte. A PDF that was already written well gains nothing.
Page as photo draws each page at 96 DPI, encodes that as a JPEG, and pastes it into a fresh PDF at the original page size. A 22 MB, 82-page report came out at 7.3 MB; a 6.3 MB slide deck at 3.0 MB. Afterwards the text is a picture: no selecting, no copying, nothing for Ctrl+F or a screen reader to find.
Which setting to pick
- A scan or a photo of a document. Its text was never selectable, so flattening costs you nothing. Go straight to Page as photo.
- A slide deck full of charts. Vector artwork often costs more bytes than a picture of it. Page as photo halved a 6.3 MB deck Keep text couldn't touch.
- A contract you might search later. Keep text. If that isn't enough, drop the pages you don't need instead.
What this tool cannot do
Keep text leaves the images inside your PDF alone. Browser PDF libraries offer no way to swap an embedded image for a re-encoded one, the trick desktop compressors use to halve a scan without touching the text. Flattening is the substitute here, and it takes the text down with the images. Fonts don't get subset either.
When a setting produces something bigger than the file you dropped in, you get your own file back and the size row says so. Page as photo does that on most prose: five of those nine came back untouched, because a JPEG of a page of text costs more than the text.
Frequently asked questions
Does compressing lose quality?
Keep text loses nothing. Page as photo re-encodes at 96 DPI, which leaves body text readable but soft beside the original. Hairlines and footnote-sized type suffer first.
Why did my file come back the same size?
Either it was already packed tightly, or your setting produced something bigger and you got the original back. Try the other one.
How long does a big document take?
That 82-page, 22 MB report took 1.3 seconds on Keep text and 5.9 on Page as photo, on a desktop. Flattening holds every page's JPEG until it writes the file, so a long document on a phone can run the tab out of memory.
Can I compress a password-protected PDF?
If it asks for a password to open, no. If it opens fine but blocks copying and printing, Keep text refuses it and Page as photo handles it.
Is my document uploaded anywhere?
No. The page downloads its PDF libraries from this site, and a JPEG 2000 or JBIG2 scan pulls one more decoder file. Your document stays in the tab.
Will the page count or page size change?
The page count never does, and Keep text leaves the dimensions alone. Page as photo bakes rotation in, so a page stored sideways comes back the right way up with its width and height swapped.