PDF Compression: How It Works & How to Reduce PDF Size by 80%

Why PDF File Size Matters

Large PDF files create friction in nearly every workflow where documents are shared or stored. Email attachment limits (typically 25 MB for Gmail, 20 MB for Outlook) reject oversized PDFs. Cloud storage quotas fill up faster. Website uploads time out. And mobile devices struggle to render documents with hundreds of megabytes of embedded content.

Understanding why PDFs become large — and how compression algorithms reduce their size — gives you the knowledge to optimize your documents effectively.

What Makes PDFs Large? Anatomy of a Bloated PDF

A PDF file is essentially a container format that packages multiple types of content together. Here's what typically contributes to file size:

Embedded Images (Usually the #1 Culprit)

Images are almost always the largest component of a PDF. A single high-resolution photograph can consume 5-20 MB of space, and documents with many images (product catalogs, photo reports, scanned documents) can easily reach hundreds of megabytes.

The image format matters enormously:

• Uncompressed bitmaps: Maximum quality, maximum file size
• JPEG/DCT compression: Lossy compression that works well for photographs
• JPEG2000: More efficient than JPEG but less universally supported
• CCITT/Fax compression: Optimized for black-and-white scanned documents
• Flate (ZIP) compression: Lossless compression used for line art and screenshots

Embedded Fonts

Each font used in a PDF can add 50-500 KB to the file size. Some PDFs embed full font files even when only a few characters are used. Font subsetting — embedding only the glyphs actually used in the document — can dramatically reduce this overhead.

Duplicate Resources

Poorly generated PDFs may contain duplicate copies of the same image, font, or graphic resource. Each page might embed its own copy of the company logo instead of referencing a single shared resource.

Metadata and Structure Overhead

PDF metadata (document properties, XMP data, embedded thumbnails, form field definitions) and structural elements (cross-reference tables, object streams) contribute to file size, though typically less than images and fonts.

Legacy Content and Incremental Saves

When PDFs are edited with tools that use incremental saving, the original content is preserved and new content is appended. After multiple edits, a document may contain outdated objects that are no longer referenced but still contribute to file size.

How PDF Compression Works: The Technical Details

Image Downsampling

The most impactful compression technique is reducing the resolution of embedded images. Most PDFs created from scanners or design software embed images at 300 DPI (dots per inch) or higher. For on-screen viewing, 150 DPI is more than sufficient. For web sharing, even 96 DPI can look acceptable.

The downsampling process works by:

• Detecting all raster images in the PDF
• Comparing each image's pixel dimensions with the chosen level's size cap (the longest edge, in pixels)
• Resampling images that exceed the cap with high-quality smoothing
• Re-encoding the downsampled images as JPEG, keeping the original whenever the new version would not be smaller

Note that BloomPDF caps pixel dimensions rather than targeting a DPI value: the effective DPI of an image depends on how large it is placed on the page, so a 1600 px image is about 200 DPI across an 8-inch-wide page but only about 100 DPI across a 16-inch poster.

Stream Compression

PDF objects are stored in "streams" — sequences of bytes that can be independently compressed. BloomPDF applies Flate (zlib) compression to all uncompressed streams, which typically reduces text-heavy content by 60-70%.

Object Deduplication

The compression engine scans for identical objects (images, fonts, content streams) and replaces duplicates with references to a single shared instance. This is particularly effective for documents generated by merging multiple files that share common elements.

Font Subsetting

Full font files embedded in the PDF are replaced with subsets containing only the characters actually used in the document. For a typical business document using a few hundred unique characters, this can reduce font data by 80-95%.

Metadata Stripping

Optional removal of non-essential metadata including:

• Embedded thumbnails (which PDFs generated by Adobe products often include)
• XMP metadata blocks
• Document history and incremental save data
• Unused named destinations and bookmarks

Compression Levels Explained

BloomPDF offers three compression presets (plus a custom target size, which steps through stronger settings until the file fits):

Recommended (Balanced)

• Images: longest edge capped at 1600 px
• JPEG quality: 72%
• Lossless images (Flate) converted to JPEG when that is smaller
• Metadata removed; object streams enabled
• Best for: General office documents, reports, and presentations

Extreme (Maximum Compression)

• Images: longest edge capped at 1100 px
• JPEG quality: 55%
• Metadata, embedded thumbnails and application data removed
• Best for: Web uploads, email attachments, and archival

Low (Quality Preservation)

• Images: only those over 3000 px (longest edge) are downsampled
• JPEG quality: 85%
• Lossless images kept lossless; metadata kept
• Best for: Print-ready documents where image quality is critical

Actual savings depend entirely on the content: image-heavy PDFs often shrink by 60-95%, while text-only PDFs may barely change.

Real-World Compression Examples

How to Compress PDFs with BloomPDF

• Open the Compress PDF tool at bloompdf.app/tools/compress-pdf
• Upload your PDF by dragging and dropping or clicking to browse
• Select your compression level (Recommended, Extreme, or Low)
• Click "Compress PDF" — processing happens entirely in your browser
• Download your optimized file and compare the size reduction

Tips for Minimizing PDF Size at the Source

Prevention is better than cure. Here are ways to create smaller PDFs from the start:

• Scan at appropriate resolution: 200 DPI is sufficient for most text documents. Only use 300+ DPI for documents with fine detail.
• Use vector graphics instead of rasterized images where possible (charts, diagrams, logos)
• Choose "Save As" over "Save" to eliminate incremental save bloat
• Export from source applications with web-optimized settings
• Avoid embedding unnecessary fonts — use standard fonts when possible

Conclusion

PDF compression is both an art and a science. By understanding what makes PDFs large and how compression algorithms target each component, you can make informed decisions about the right balance between file size and quality for your specific use case. With BloomPDF's client-side compression, you get professional-grade optimization without compromising your document privacy.