#1PDF compression: quality vs. file size

PDFs get large fast, and the trade-off is usually straightforward: reduce image resolution and encoding enough to keep downloads small without ruining text or charts.

The real levers are the practical ones: image downsampling, encoding choices, font subsetting, and the cleanup steps that reduce size without making the final PDF harder to read.

#21. How PDF Compression Works Inside the Document Stream

To optimize a PDF effectively, you must understand how space is allocated inside a PDF container file. A standard PDF consists of four distinct internal object types:

┌─────────────────────────────────────────────────────────────┐
│                    PDF FILE CONTAINER                       │
├─────────────────────────────────────────────────────────────┤
│ 1. Vector Geometry & Text Streams  (~1-5% of file size)     │
│ 2. Embedded Font Files             (~5-15% of file size)    │
│ 3. Structural Metadata & Objects   (~1-5% of file size)     │
│ 4. Embedded Raster Images          (~80-95% of file size)   │
└─────────────────────────────────────────────────────────────┘

Because text strings, page layout instructions, and vector paths take up negligible disk space (usually a few kilobytes per page), raster images (bitmaps) account for 80% to 95% of the total file weight in almost every large PDF document.

#3The Three PDF Compression Mechanisms

Efficient PDF compression applies three targeted operations:

1. Raster Image Downsampling: Reducing the resolution (DPI/PPI) of embedded images to match the target display medium.
2. Stream Encoding: Re-compressing uncompressed or inefficiently encoded image streams using modern compression algorithms (Flate, JPEG 2000, JBIG2).
3. Document Pruning: Removing redundant structural objects, embedded thumbnails, duplicate fonts, unreferenced resources, and metadata history.

#22. Image Downsampling: Choosing the Right DPI

Downsampling decreases the total number of pixels in an embedded image by resampling the pixel grid. The resolution of images in a PDF is measured in DPI (Dots Per Inch) or PPI (Pixels Per Inch).

#3Target DPI Matrix by Use Case

#3Downsampling Algorithms Explained

When reducing an image from 300 DPI to 150 DPI, the compression engine must decide how to calculate the color of the new, smaller pixel grid:

• Bicubic Downsampling (Best Quality): Uses a weighted average of surrounding pixels to compute new pixel values. Produces the smooth transitions and sharpest edges for photos and complex graphics.
• Bilinear Downsampling (Faster): Calculates pixel averages across a linear grid. Faster processing but can produce slight blurring on high-contrast text edges.
• Subsampling (Fastest / Lowest Quality): Selects a sample pixel from the center of the grid group and discards the rest. Fast but produces jagged aliasing artifacts on diagonal lines.

Recommendation: Always specify Bicubic Downsampling when compressing PDFs containing photographs or complex marketing visuals.

#23. Lossy vs. Lossless Image Encoding Algorithms

Inside a PDF, different types of images require different compression algorithms to balance file size reduction against visual fidelity:

#3Lossy Compression (JPEG / JPEG 2000)

Lossy compression permanently discards subtle color data that human eyes are less sensitive to, achieving compression ratios of 10:1 to 50:1.

• Best Used For: Continuous-tone photographs, complex gradients, background imagery.
• Warning Signs of Over-Compression: "JPEG Artifacts", blocky grid patterns and ringing halos around text or high-contrast edges inside images.

#3Lossless Compression (Flate / ZIP / JBIG2)

Lossless compression reduces file size using mathematical pattern encoding without altering a single pixel of image data.

• Flate (ZIP): Used for color screenshots, flat vector graphics, logos, and line drawings. Prevents blurry text edges.
• JBIG2: Specialized lossless/lossy compression algorithm specifically designed for monochrome (black and white) scanned text documents. JBIG2 can compress scanned text pages up to 500% more effectively than standard Flate compression.

#24. Font Subsetting: Eliminating Hidden Font Weight

When you author a document using custom typography (e.g., OpenType or TrueType fonts like Inter, Roboto, or Helvetica), the PDF authoring software embeds the font file into the document so it renders accurately on computers that don't have that font installed.

A complete OpenType font family with full international character sets (Cyrillic, Greek, CJK) can add 2MB to 10MB of weight to a PDF document.

#3How Font Subsetting Works

Font Subsetting analyzes the document and creates a customized, stripped-down font file containing only the specific characters used in the text.

Full Font File (Included by default):
[A B C D E F G H I J K L M N O P Q R S T U V W X Y Z a b c d e ... + 5000 glyphs] = 4.2 MB

Subsetting (Only characters used in document):
[A B C E H I L O P R S T W a b c d e i l n o r s t v y] = 38 KB

Enabling font subsetting reduces font overhead by up to 98% without changing the visual appearance of a single word.

#25. Vector Path Simplification and Flattening

Complex architectural drawings, CAD exports, or vector illustrations often contain millions of individual vector anchor points and path segments. While vector elements scale infinitely, extremely dense vector paths can make a PDF slow to render on mobile devices and unnecessarily large on disk.

#3Vector Optimization Steps

1. Path Decimation: Reduce redundant vector anchor points in graphics tools before PDF export.
2. Transparency Flattening: Convert complex overlapping vector transparency blend modes into single flattened objects.
3. Clip Path Reduction: Crop hidden graphics outside the page bounding box (/MediaBox) rather than clipping them dynamically during render time.

#3PDF/A Archival Standards and Compression Constraints

If your organization requires PDF/A compliance (PDF/A-1b, PDF/A-2b) for long-term legal or government archiving, specific compression techniques are restricted by the standard:

• Font Embedding Required: PDF/A strictly forbids font subsetting or un-embedded system fonts, all font glyphs and metrics must be fully embedded, increasing file size.
• No External References: External URL links or dynamic content bindings are prohibited.
• Device-Independent Color: Colors must be converted to sRGB or ICC color profiles, which adds ICC profile dictionary overhead.

When compressing PDF/A documents, expect file sizes to be 20% to 40% larger than standard web-optimized PDFs due to mandatory archival compliance compliance structures.

#26. Document Structure Pruning & Linearization

Beyond images and fonts, unoptimized PDFs accumulate structural "junk data" over multiple edit cycles:

#3Content Pruning Checklist

• Remove Embedded Thumbnails: Legacy PDF creators embedded small JPEG previews of every page. Modern PDF viewers render previews dynamically on CPU; embedded thumbnails are pure wasted space.
• Strip Unreferenced Objects: Deleted pages, abandoned form fields, and overwritten annotations often leave orphaned object bytes inside the file stream.
• Remove Duplicate Fonts: Merging multiple PDFs often results in the same font being embedded 5 separate times under different object IDs.
• Strip Metadata History: Remove revision logs, edit timestamps, and author details (see our PDF Privacy Guide).

#3Fast Web View (Linearization)

Standard PDFs store their object index (xref table) at the very end of the file. This means a web browser must download 100% of a 50MB PDF before it can render Page 1.

Linearization (Fast Web View) reorganizes the internal PDF structure so that Page 1 objects and the primary index are placed at the beginning of the byte stream. This allows web browsers to display Page 1 instantly while the rest of the document downloads progressively in the background.

#26. Compression Security & Data Privacy

Compressing sensitive company documents (financial audits, legal contracts, HR records, medical files) using cloud-based PDF compression websites creates significant security risks.

#3The Cloud Compression Risk

When you drag a file into a free online PDF compressor:

1. Your entire document travels across the internet to a third-party server.
2. The server processes the file in memory or temporary storage.
3. If the service is unencrypted, logged, breached, or uses third-party APIs, confidential information is exposed.

#3Client-Side Local Compression

Modern browser technologies (WebAssembly and PDF WASM libraries) enable complete PDF compression 100% locally inside your browser's memory.

Using the AllDevToolsHub PDF Compressor:

• PDF parsing, downsampling, and re-encoding happen on your local CPU.
• Zero bytes are transmitted to any remote server.
• Processing speeds are bound by your local hardware, not your upload bandwidth.
• You can verify zero network requests via browser DevTools (Network tab).

#27. Command-Line PDF Compression for Automated Pipelines

For server backend infrastructure, CI/CD document processing, or batch command-line workflows, developers can automate PDF compression using open-source utilities locally:

#3Using Ghostscript for Automated Compression

# High Quality (300 DPI / Ebook preset)
gs -sDEVICE=pdfwrite -dCompatibilityLevel=1.4 -dPDFSETTINGS=/ebook \
   -dNOPAUSE -dQUIET -dBATCH -sOutputFile=compressed_ebook.pdf input.pdf

# Screen Quality (72-150 DPI / Screen preset)
gs -sDEVICE=pdfwrite -dCompatibilityLevel=1.4 -dPDFSETTINGS=/screen \
   -dNOPAUSE -dQUIET -dBATCH -sOutputFile=compressed_screen.pdf input.pdf

#3Ghostscript Preset Summary

• /screen: Selects low-resolution images (72 DPI) for minimum file size.
• /ebook: Selects medium-resolution images (150 DPI) for web and email.
• /printer: Selects high-resolution images (300 DPI) for desktop printing.
• /prepress: Preserves high-resolution images and CMYK color spaces for commercial printing.

#28. Recommended Compression Profiles

Use these guidelines when configuring compression settings:

### Profile 1: Web & Email Distribution (Recommended Default)
- Image Resolution: 150 DPI
- Image Compression: JPEG (Bicubic, Medium-High Quality / 75%)
- Font Subsetting: Enabled
- Structural Pruning: Strip thumbnails, metadata, and unused objects
- Linearization: Enabled

### Profile 2: High-Quality Archival / Print Preview
- Image Resolution: 200 – 300 DPI
- Image Compression: JPEG (High Quality / 90%) or Flate (Lossless)
- Font Subsetting: Enabled
- Metadata: Preserved (if required for archival indexing)

### Profile 3: Maximum Size Reduction (Mobile / Low Data)
- Image Resolution: 96 DPI
- Image Compression: JPEG (Low Quality / 50%)
- Monochrome Images: JBIG2
- Font Subsetting: Enabled
- Structural Pruning: Aggressive (Strip all non-visual objects)

#3Real-world compression results

We tested these three profiles against a 12-page scanned contract (original: 8.4 MB) to show the actual trade-offs:

The 150 DPI profile hits the sweet spot for most use cases: 87% size reduction with no visible quality loss on text-heavy documents. Drop to 96 DPI only when bandwidth is the constraint and the document is primarily text.

#2Summary

Optimizing PDF files requires balancing visual quality against network efficiency:

1. Target 150 DPI for web and email distribution, the optimal balance between visual crispness and file size.
2. Apply Bicubic Downsampling to preserve clean photographic and text edges inside embedded graphics.
3. Enable Font Subsetting to eliminate megabytes of unused font character glyphs.
4. Prune Structural Junk: Strip thumbnails, duplicate objects, and old revision histories.
5. Compress Locally: Use browser-based WASM compression to keep your confidential documents private.

Compress your documents securely at the AllDevToolsHub PDF Suite.

#2Related Tools

• PDF Compressor, Compress PDF files locally in your browser with adjustable DPI settings
• PDF Metadata Cleaner, Remove hidden author info, revision history, and EXIF tags
• PDF Merger & Splitter, Combine or extract PDF pages locally

#2Related Articles

• PDF Privacy: How to Scrub Metadata and Prevent Tracking
• The Ultimate PDF Toolkit: Merge, Split & Compress
• Mobile PDF Workflow: Scan, Sign & Edit

#2Frequently Asked Questions

Q: Why does my PDF get larger after running compression?

A: If a PDF contains already highly optimized, low-resolution JPEG images, attempting to re-compress them can introduce additional encoding headers or convert efficient streams into less efficient representations. Additionally, if the compression tool fails to subset fonts or adds uncompressed thumbnail streams, the total file size can increase. Always check whether your PDF is already compressed before applying aggressive re-compression.

Q: What is the difference between DPI and PPI in PDF compression?

A: DPI (Dots Per Inch) refers to physical ink dots on printed paper, while PPI (Pixels Per Inch) refers to digital pixel density on a display screen. In the context of digital PDF software, the terms are used interchangeably to describe image sampling density.

Built and reviewed by the AllDevToolsHub Engineering Team, with a focus on tools that run locally first.

Written by Rahul Jalavadiya, founder of AllDevToolsHub. All tools run locally in your browser.

#2Sources / Further reading

• ISO - PDF 2.0 Specification (ISO 32000-2:2020)
• Adobe - PDF Reference, Sixth Edition
• Ghostscript - PDF compression documentation