How to Compress PDF Online Free

A large PDF is slow to email, awkward to upload, and heavy on storage. Compressing it re-encodes the file's contents so the same pages take up less space, without changing what the document says or how its pages are laid out.

The important thing to understand is that a PDF is a container of objects: text, fonts, images, and metadata.^[2] Compression works on those objects, and where it saves space depends heavily on what your PDF is made of. This guide explains the internal structure that makes those savings possible, where the biggest wins come from, what you give up if you push too hard, and how to do it locally in your browser.

What compressing a PDF actually does

Compressing a PDF does not throw away pages or reflow the layout. It re-encodes the objects inside the file so they occupy fewer bytes. The main levers are re-compressing embedded images at a higher JPEG compression ratio, downsampling those images to a lower resolution, stripping out metadata, subsetting or flattening fonts so only the glyphs you use are stored, and removing redundant or orphaned objects left behind by the software that created the file.

None of this changes the page structure or the underlying text layer. A paragraph of selectable text stays selectable and searchable. What changes is the fidelity of the images and the amount of overhead the file carries around.

Inside a PDF: objects, streams, and filters

PDF was developed by Adobe and later standardized as an open format, ISO 32000, so its internal structure is well documented and stable.^[1] A PDF file is essentially a collection of numbered objects referenced through a cross-reference table. Some objects are small, like a dictionary describing a page; others are large streams that hold the actual content: the drawing instructions for a page, an embedded font program, or the raw bytes of an image.

Every stream in a PDF can carry a filter that says how its bytes are compressed. Text and vector drawing instructions are typically compressed with FlateDecode, which is the same DEFLATE algorithm used in zip files and PNG, applied losslessly. Photographic images are usually stored with DCTDecode, which is JPEG compression, while bilevel scans of black-and-white text often use fax-style CCITT or JBIG2 encoding. When a tool compresses a PDF, it is choosing better filters and better filter settings for these streams, and pruning objects that are no longer needed.

This structure is why the composition of a document matters so much. A PDF whose bulk is text and vector graphics is already compressed losslessly and has little slack, whereas a PDF whose bulk is high-resolution images is carrying most of its weight in streams that can be re-compressed far more aggressively.

Image downsampling and re-encoding

For most heavy PDFs, images are where the file size lives, and two operations shrink them. The first is downsampling, which reduces an image's pixel resolution to match how it will actually be viewed. A photo scanned or placed at 600 DPI holds four times as many pixels as the same photo at 300 DPI, and for on-screen reading even 150 DPI is usually indistinguishable. Because file size scales with the number of pixels, dropping the resolution to what the use case needs is the single largest lever available.

The second is re-encoding, raising the JPEG compression ratio on the images so each one stores fewer bytes. This is the same lossy quantization that governs a standalone JPG: it discards fine image detail the eye is least likely to notice. Downsampling and re-encoding compound, so a photo-rich brochure can often shrink to a fraction of its original size with the images still looking perfectly acceptable on screen. The trade, as always with lossy image compression, is that the discarded detail does not come back.

Font subsetting and stream compression

Fonts are a quieter but real source of bloat. When a PDF embeds a font to guarantee it displays identically everywhere, it may embed the entire typeface, thousands of glyphs, even though the document only uses a few dozen characters. Font subsetting keeps only the glyphs the document actually uses and discards the rest, which can save a surprising amount in files that embed several full font families. The document still renders exactly as before, because every character it displays is still present.

Alongside that, a compressor re-runs stream compression across the whole file: it applies FlateDecode to any content streams that were left uncompressed, consolidates many small objects into compressed object streams, rebuilds a compact cross-reference table, and removes orphaned objects such as deleted-but-retained page content or duplicate images. None of these steps touch the visible content; they simply remove overhead. On a text-heavy PDF that was exported carelessly, this housekeeping alone can trim a meaningful slice with zero loss of quality.

Object streams and cross-reference cleanup

Modern PDF versions added two structural features that a good compressor exploits, and understanding them explains why a carelessly exported file can be much larger than it needs to be. The first is the object stream. In an older PDF, every object, even a tiny dictionary, sits in the file separately with its own uncompressed header. A PDF that defines thousands of small objects therefore wastes a great deal of space on overhead. An object stream bundles many of these small objects together into a single stream and then compresses the whole bundle with FlateDecode, so the overhead disappears and the small objects ride along under one efficient compression pass. Re-writing a file to use object streams is a purely lossless win.

The second is the cross-reference table, the index a viewer uses to jump to any object by number. In older files this table is stored as verbose plain text, one line per object. Newer PDFs replace it with a compact binary cross-reference stream that stores the same index in a fraction of the bytes and can itself be compressed. Editing software also has a habit of leaving old, superseded versions of objects behind through a mechanism called incremental updates, where changes are appended rather than rewritten, so a file that has been edited many times can accumulate layers of dead data. Rebuilding the file linearizes it: the compressor writes out only the objects that are actually reachable, packs them into object streams, and emits a single compact cross-reference stream, discarding every orphaned remnant along the way.^[1]

What shrinks the most

Not every PDF has the same room to shrink. A text-only PDF, such as a plain report exported from a word processor, is already small because text and fonts are compact and losslessly packed. Trying to compress it further yields very little. The big wins come from image-heavy documents and scanned pages, where high-resolution photos or page scans dominate the file size.

The single most effective lever is image resolution. Downsampling embedded images to 150 DPI is plenty for on-screen reading, while 300 DPI keeps the file crisp for printing. Raising the JPEG compression on those images compounds the saving.

The quality trade-off

Compression that touches images is not free. Aggressive downsampling degrades photo detail, and on scanned documents it can make text noticeably blurry or harder to read, since a scan of text is really an image and downsampling it softens the letterforms. The trick is to match the resolution to how the PDF will be used: screen viewing tolerates lower DPI, printing needs more. For scanned text specifically, a bilevel (black-and-white) encoding often keeps letters crisp at a small size better than a downsampled color image would.

Compress a PDF step by step

1. Open the converter

   Open the FileFormer document converter in your browser. It runs entirely on your device, so your PDF is never uploaded to a server.

2. Add your PDF

   Drag the PDF into the tool or click to pick it from your computer. Large, image-heavy files are exactly where compression helps most.

3. Choose a compression level

   Set your target resolution: 150 DPI for screen sharing, or 300 DPI if the file needs to print cleanly. The tool re-encodes the embedded images and trims metadata to match.

4. Download the compressed PDF

   Export and save the smaller file. Because everything happened on-device, nothing left your computer during the process.

Lossless cleanup versus lossy image reduction

It helps to separate PDF compression into two very different categories, because they carry completely different risks. Lossless cleanup covers everything that removes overhead without touching the visible content: re-running FlateDecode on uncompressed streams, subsetting fonts, consolidating objects, rebuilding the cross-reference table, and stripping metadata such as editing history, thumbnails, and author fields. These operations are always safe to apply, because the rendered document is byte-for-byte identical in appearance afterward. If a document is mostly text, this category is the only compression that will help it, and it will help only modestly.

The lossy image reduction category is where the large savings and the real risk both live. Downsampling and re-encoding photos genuinely discard detail, and once the file is saved that detail is gone. The sensible strategy is to always let a compressor apply the lossless cleanup fully, then apply lossy image reduction only as hard as the intended use allows: gently for anything that may be printed, more aggressively for a throwaway email attachment. Because the two categories are independent, you can get every lossless byte for free and then decide separately how much image quality you are willing to trade.

Real-world scenarios and settings

The right setting depends entirely on what the document is made of and where it is headed. A few common cases:

The pattern is that screen-bound files tolerate aggressive image reduction while anything destined for print or long-term keeping should be handled gently. Matching the resolution to the real use is what separates a file that is both small and usable from one that is small but disappointing.

Common mistakes when compressing a PDF

A few habits quietly ruin results. The first is compressing the same PDF repeatedly, which stacks lossy image passes the same way re-saving a JPG does; each round softens the images further while saving less and less. Always keep and re-compress from the original, not from a copy you already squeezed. The second is over-downsampling scanned text, where pushing the resolution too low turns crisp letters into a blur that is legible on screen but useless in print. Scanned documents reward a bilevel or higher-DPI setting more than they reward brute-force downsampling.

The third mistake is expecting compression to fix the wrong problem. If a PDF is huge because it embeds a hundred full-page photographs at print resolution, no lossless trick will shrink it meaningfully; the honest fix is downsampling those images. Conversely, if a PDF is huge despite being mostly text, the cause is usually un-subsetted fonts or uncompressed streams, and the fix is lossless cleanup, not image reduction. Diagnosing which kind of content dominates the file tells you which lever will actually move the number.