Compress a PDF

How to use it#

1. Pick a compression level. Balanced suits most files, and Lossless keeps every picture exactly as it is.
2. Keep pictures in colour, or choose Greyscale to turn photos and scans grey for a smaller file. Lossless keeps colours as they are, so Greyscale is greyed out there.
3. Drop one or more PDFs onto the box above, or click it to choose them.
4. Check each file's new size in its row. Changing the level or the colour compresses every file again straight away.
5. Download a file from its row, or Download all for a ZIP of every file. The originals are left as they were.

Why the text stays selectable#

A PDF is a set of instructions: draw these letters in this font here, put this photo there. Almost all the weight of a large one sits in the photos and scans, and the text is tiny by comparison. So this tool leaves the text alone and works on the pictures only. At every level but Lossless it resamples each one to a sensible resolution and saves it again as a JPEG.

Turning every page into one big picture would also shrink a file, but at a real cost. You could no longer select or search the words, a screen reader would have nothing to read, links would stop working, and small print would blur. This tool never touches the text, so none of that happens.

A few savings lose nothing at all, and they are made at every level. Some programs save parts of a PDF without any compression, and those parts are compressed. A file that has been edited many times can carry old data that nothing in it uses any more, which is dropped. The objects that describe the pages are packed together and compressed as well. When the first two saved a noticeable amount, the result has a line saying how much.

Compress a PDF without losing quality#

Choose Lossless and no picture changes at all. Every photo and scan is kept byte for byte, at the resolution and quality it already had, and the text, fonts, and drawings are copied across as they were. The file gets smaller only where that costs nothing. Parts stored without any compression are compressed, data left behind by earlier edits is dropped, and the objects that describe the pages are packed together and compressed. A picture stored with no compression at all is compressed too, which keeps every pixel as it was.

How much that saves depends on where the weight of the file is. A PDF from a program that skips compression, or one edited many times over, can shrink by half or more. A scan or a file full of photos is mostly pictures that are compressed already, so Lossless can take out only a few percent, and the result says what share of the file the pictures are. For a file like that, the only way to get much smaller is to let the pictures give up some detail, and Light does that with the least visible change.

Which files shrink, and which barely move#

Scanned documents and files full of photos shrink the most, often to a fraction of their size. A scanner or a phone camera usually saves pages at far more detail than a screen or a printer can show, and that surplus is what gets taken out.

A PDF exported from Word or Google Docs with no pictures in it usually has little to give. Its text and fonts are already stored compactly, so it may only come down by a few percent. The exception is a file saved without compression, which can shrink a long way with no pictures in it at all. When nothing gets smaller, the tool says so rather than handing back a copy that is no better than the original.

What each level costs you#

Lossless keeps every picture as it is and makes only the savings that lose nothing. Minimal keeps pictures at up to 300 dots per inch at high JPEG quality, a change few people could spot even in print, and suits a file that needs to be smaller without looking any different. Light keeps them at up to 220, measured against the page, which still prints cleanly. Balanced keeps them at up to 150, which looks sharp on any screen and is right for email and upload forms. Strong goes down to 100 and compresses harder, for when a file has to fit under a strict size limit.

The cost shows up in the pictures only. At Strong, a photo looks soft if you zoom in on it, and fine print inside a scanned page can lose its edges. Try Balanced first, check the result, and move to Strong only if the file is still too big. To get under a particular limit, such as 1 MB, the compress to size tool tries the levels for you and keeps the lightest that fits.

Greyscale: smaller pictures, same detail#

A colour picture stores three values for every pixel, and a grey one stores one. With Greyscale chosen, every photo and scan is written again in a single grey channel, so the file holds less for the same sharpness. It works at every level except Lossless, which changes no picture. Colour photos gain the most, often coming out a third smaller or more, and so do colour scans of an ID or a certificate, where the colour tells the reader nothing. A scan of black text on white paper has little colour in it to begin with, so it shrinks by only a few percent more.

Only pictures turn grey. Text, lines, tables, and shapes the PDF draws itself are not pictures, so a coloured heading, a blue link, or a logo drawn as shapes keeps its colour. A scanned document is nothing but pictures, one per page, so every page of a scan comes out grey.

A picture turns grey only when that makes it smaller, which is almost always. Pictures in print colours (CMYK) and JPEG 2000 pictures stay as they were, and the result says how many pictures turned grey and how many stayed in colour.