Compress a PDF without losing quality: what is true

Lossless compression is real, and small

Lossless means the file gets smaller and nothing you can see changes, not even 1 pixel. In a PDF there are a handful of ways to do that, and our compressor does all of them on every run:

• Raw streams get compressed. Some writers store page content or fonts as plain bytes. Deflating them (the same method a ZIP file uses) is fully reversible.
• Unused objects go. Files edited many times carry old versions nobody can reach. They are swept out.
• Duplicate pictures are merged. A logo stored once per page becomes 1 picture used on every page.
• Page thumbnails go. Small preview images some older writers embed; no current reader uses them.
• The structure is packed. The file is written with object streams, a compact way of storing the small objects every PDF has.

These steps are worth having, but they rarely move a big file much. Our 3-page text file goes from 2 KB to 1 KB. Our 500-page text file stays at 117 KB on every setting: it has no pictures and its structure is already tidy, so the tool hands it back unchanged and says so. If your file is big, it is big because of pictures, and pictures do not shrink much losslessly.

The big savings are lossy, by design

Almost all the weight in a big PDF is pictures: scans, photos, design artwork. To make those substantially smaller, a compressor has to change the pixels in 2 ways:

1. Fewer pixels. A picture placed on the page at more than the target resolution is shrunk to it: 300 dpi for Print, 150 for Web, 110 for Email. Halving the dpi leaves a quarter of the pixels.
2. Coarser encoding. Photographs are re-encoded as JPEG, which saves space by discarding detail the eye notices least. Lower quality numbers discard more.

What that does to a real page, a generated test scan (1 A4 page at 300 dpi in colour, 3.3 MB):

Preset	What happens to the picture	Result	Saved
Print	stays 300 dpi, JPEG quality 85	2.9 MB	10 %
Web	150 dpi, JPEG quality 75	328 KB	90 %
Email	110 dpi, JPEG quality 65	130 KB	96 %

What “no quality loss” means on other compressor pages

Many online compressors promise “no quality loss” and then cut a file by 80 percent. Both cannot be true in the strict sense. What the promise usually means is “you will not notice at normal viewing size”, which is a fair claim for many files and not the same thing as lossless. The pixels changed; the question is whether it matters for your use.

So treat the promise as a claim to test, not a fact. The test takes 30 seconds and is below.

Where loss stays invisible, and where it shows first

Invisible, almost always:

• Real text. In a PDF made from a word processor, text is stored as characters with a font, not as a picture. Our compressor never touches it, so it stays sharp, selectable and searchable at any zoom.
• Vector drawings. Charts, lines and shapes drawn by the PDF itself have no pixels to lose.
• Flat pictures. A screenshot, logo or diagram with 64 colours or fewer is stored losslessly, because JPEG would smear its edges. It is still shrunk if it sits on the page at more than the preset’s resolution.
• Photos at page width. A photo shown 10 cm wide on screen needs far fewer pixels than a camera captured.

Visible first, when you zoom:

• Fine print in scans. In a scan, text is part of the page picture. Footnotes, small stamps and pencil soften first.
• Thin lines in scanned or photographed drawings. Plans and line art saved as a photo pick up faint halos around edges at low quality.
• Detailed photos enlarged. A product shot someone will zoom to 300 percent shows the lower resolution.

Check your own file in 30 seconds

1. Compress the file with the preset you think fits.
2. Open the original and the compressed copy side by side.
3. Zoom both to 200 percent on the smallest text on the page, then on a photo or scanned area.
4. If you cannot tell them apart at the zoom your reader will use, send the compressed copy. If you can, go back one step: Web instead of Email, or Print instead of Web.

When a portal or mail server names a size, type it as a size target instead of guessing a preset. The tool tries its settings from 300 dpi down and keeps the gentlest one that fits, so you lose no more than the limit forces; a file already under the target comes back untouched.

The result screen helps you aim: it lists the 5 pages whose pictures changed most, with their bytes before and after, so you know which pages to look at.

The honest part: our tool is not lossless

Our compressor does not have a lossless-only switch. Every preset may re-encode photographs. The gentlest is Print: it keeps pictures at 300 dpi and replaces a picture only when the re-encoded version is at least 10 percent smaller. On our 10-page scan, already at 300 dpi, Print returns the file unchanged and says so. On the generated test page it re-encoded the picture at quality 85 and saved 10 percent; that is a small, real loss.

What it promises instead:

• Text, fonts and drawings are never re-encoded.
• The result says what it did: how many pictures were re-encoded, how many were shrunk and to what dpi, how many were kept lossless.
• A result that is not smaller is never delivered: you get the original back with a sentence explaining why.
• Your original is never overwritten; the compressed copy is a new file.

If you need a strictly lossless result, for an archive or a legal record, take the small saving and stop: do not compress at all, or use a command-line tool such as qpdf that rewrites a file’s structure without touching its pictures. And if the file is a print master, keep it big; compress only the copy you send.

For the lever that actually decides how much you lose, read Reduce PDF DPI. For realistic floors by file type, see How small can a PDF get.