Compress PDF Without Losing Quality Free Online

Images account for nearly all oversized PDFs. A scanner set to 600 DPI captures four times the data of one set to 300, and neither is visible on screen, so a twenty page scan can arrive at 80 MB when 8 MB would look identical. Phone cameras do the same thing, producing a twelve megapixel photograph of a page that only ever needed to be legible.

Embedded fonts contribute a fixed overhead that surprises people on short documents. A two page letter using four font families can carry several hundred kilobytes of typeface data, which is why very short PDFs sometimes seem disproportionately large and why compressing them achieves almost nothing.

Then there is accumulated debris. Documents edited repeatedly, or exported by software that keeps revision data, hold objects nothing references any more. This is the part compression removes cleanly with no quality cost at all, and it is why even a text document usually gives back something.

The four levels map onto how the document will actually be used, so decide that first. Low is for a file that has to clear an email limit and will be read on screen. Medium suits most general sharing and is the sensible default when you are unsure. High keeps enough resolution for an office printer. Extreme barely reduces anything and exists for documents heading to a professional print shop.

Match the level to the destination rather than to the number you are hoping for. Compressing a document to Low because 3 MB sounds better than 9 MB is a poor trade if the result is going to be printed, since the resolution loss is permanent and the file cannot be restored. Keep the original until you have looked at the output.

Check the images rather than the text when you review it. Text stays crisp at every level because it is stored as glyph instructions rather than pixels, so the only place quality shows is in photographs, logos, and scans. Zoom to a hundred percent on the most detailed image in the document and judge from there.

Compress last. Almost every other operation either adds content or copies it, so anything you do afterwards partly undoes the savings. Merge the files, remove the pages you do not want, add the numbering, and compress the finished document once at the end.

OCR is the important exception to think about, and it also goes before compression, for a different reason. The recognition engine reads the page image to identify characters, so handing it a downscaled image permanently produces a worse text layer. Run OCR on the original, then compress, and the invisible text layer survives untouched because text is not what compression removes.

If the file is still too large after all that, stop adjusting the quality setting. A document dominated by pages nobody needs shrinks far more from removing those pages than from any compression level, and a file above 50 MB has to be split before it can be uploaded at all.