---
title: Compress a PDF without turning the text into a picture
source: https://arazioformac.com/docs/compress-a-pdf
---

# Compress a PDF

Pick the PDF up, hold <kbd>⌥</kbd>, let go toward **Compress**.

**The one-gesture way**

1. Pick the file up in Finder, as if you were moving it.
2. Hold `⌥`. A ring of tools appears around the cursor.
3. Let go toward **Compress**. The new file lands beside the original.

Six tools work on PDF files — hold ⌥ and they appear around the cursor:
Compress, Extract Pages, Merge, Rotate, Split, Strip Metadata.

Six actions work on any file at all — hold ⌃, including on files Arazio cannot
convert: AirDrop, Shelf, Copy, Move, Copy Path, Terminal.

That is the whole thing. It works on one file or four hundred, and
nothing is uploaded — the compression happens on your Mac.

## What makes this different

Most PDF compressors rasterise. Every page is rendered to an image, the images
are JPEG-compressed, and a new PDF is built out of the pictures. The file gets
dramatically smaller, which is why the number looks so impressive.

What you lose: selecting, copying and searching text; screen readers, so the
document stops being accessible; crisp text when zoomed; and any chance of
getting it back.

**Arazio does not do that.** It subsets the embedded fonts — cutting each one
down to the characters the document actually uses — and shares streams that
appear twice. The words stay words.

## And then it checks

Subsetting is not universally safe. A PDF refers to glyphs by number, and in
some documents the character codes reach glyphs in a way that a subsetter can
get wrong. When that happens the file still opens, still reports all its text
to a search, and draws the wrong letters.

So Arazio renders both documents and compares them pixel by pixel. If anything
moved, it throws its own work away and gives you back the file you started
with.

This is why compressing can never quietly ruin a document, and also why it
takes a second or two on a long one.

## How much smaller

It depends entirely on what is inside.

**A text document from Word, Pages or Excel** — often 20–30% smaller. One
31-page report here went down 27% with every page rendering identically; it had
no images at all, and was carrying ten typefaces with 4,551 glyphs for a
document that used 85 distinct characters, including Arial embedded twice.

**A scan or a photo album** — every page is already a picture, so the saving
comes from re-encoding those images, and how much depends on how they were made.

**An already-tight PDF** — nothing. If a document was made properly, with
subset fonts and compressed images, there is nothing left to take. Arazio says
so in those words and returns the original rather than a copy that is somehow
larger.

## Checking it worked

Open the result and try to select a line of text.

If you can highlight individual words, the text survived. If the whole page
highlights as one block, or nothing highlights, it was rasterised — by whatever
produced it. Do that check before deleting an original.

## Why Preview often makes it bigger

macOS has **File → Export → Quartz Filter → Reduce File Size** in Preview. It
is free and worth trying on a scan, where it downsamples images to 72 dpi.

On a text document it frequently produces a larger file. It has nothing to take
from a page of words, and the re-encode adds overhead. That is the tool meeting
a file it has no answer for, not you doing it wrong.