PDF Compression: How to Reduce File Size Without Losing Quality

March 29, 2026 · 11 min read

Why PDFs Get Large

A PDF is a container format that can hold text, images, fonts, vector graphics, JavaScript, multimedia, and metadata. File size bloat usually comes from a few specific sources:

Use our PDF Info tool to analyze what is consuming space in your file.

Compression Methods

Image Downsampling

The most effective compression technique. Downsampling reduces image resolution — a 300 DPI image downsampled to 150 DPI becomes roughly 1/4 the pixel count.

• Bicubic downsampling — Best quality, averages pixel neighborhoods
• Average downsampling — Faster, slightly lower quality
• Subsampling — Fastest, picks nearest pixel (can look blocky)

Image Recompression

Font Subsetting

A full font file contains thousands of glyphs. If your document uses only 50 characters, subsetting removes the unused glyphs. A 2 MB font might shrink to 30 KB after subsetting.

Content Stream Compression

PDF content streams (text positioning, vector drawing commands) can be compressed with Flate/Deflate. Uncompressed streams are surprisingly common in PDFs generated by older software.

Object Deduplication

When the same image appears on multiple pages, it should be stored once and referenced by each page. Some PDF generators embed a separate copy per page — deduplication fixes this.

Lossy vs Lossless Compression

For most use cases, a combination works best: lossless techniques first (metadata removal, stream compression, deduplication), then lossy image compression if more reduction is needed.

Recommended Settings by Use Case

Try our PDF Compressor — it processes files entirely in your browser with no upload required.

Ghostscript Commands

Ghostscript is the most powerful free tool for PDF compression. The -dPDFSETTINGS flag provides preset compression levels:

# Screen quality (72 DPI, smallest)
gs -sDEVICE=pdfwrite -dCompatibilityLevel=1.4 \
   -dPDFSETTINGS=/screen -dNOPAUSE -dBATCH \
   -sOutputFile=output.pdf input.pdf

# Ebook quality (150 DPI, good balance)
gs -sDEVICE=pdfwrite -dPDFSETTINGS=/ebook \
   -dNOPAUSE -dBATCH -sOutputFile=output.pdf input.pdf

# Printer quality (300 DPI)
gs -sDEVICE=pdfwrite -dPDFSETTINGS=/printer \
   -dNOPAUSE -dBATCH -sOutputFile=output.pdf input.pdf

# Custom: 150 DPI with JPEG quality 80
gs -sDEVICE=pdfwrite -dCompatibilityLevel=1.4 \
   -dDownsampleColorImages=true -dColorImageResolution=150 \
   -dJPEGQ=80 -dNOPAUSE -dBATCH \
   -sOutputFile=output.pdf input.pdf

Python Libraries

pikepdf (recommended)

import pikepdf

# Open and save with optimization
pdf = pikepdf.open('input.pdf')
pdf.save('output.pdf',
    compress_streams=True,
    object_stream_mode=pikepdf.ObjectStreamMode.generate,
    normalize_content=True,
    linearize=True  # optimize for web viewing
)

PyPDF2

from PyPDF2 import PdfReader, PdfWriter

reader = PdfReader('input.pdf')
writer = PdfWriter()

for page in reader.pages:
    page.compress_content_streams()
    writer.add_page(page)

writer.add_metadata(reader.metadata)

with open('output.pdf', 'wb') as f:
    writer.write(f)

Compression Comparison Table

Practical Tips

1. Check image resolution first — Use PDF Info to see embedded image DPI. A 600 DPI scan viewed on screen wastes 90% of its pixels.
2. Convert RGB to sRGB — CMYK images are 33% larger than RGB equivalents. Convert to sRGB unless printing professionally.
3. Remove hidden layers — CAD exports and Illustrator files often contain hidden layers that add significant size.
4. Flatten transparency — Complex transparency effects increase file size and rendering time.
5. Strip metadata — Remove XMP metadata, document history, and page thumbnails.
6. Linearize for web — "Fast web view" reorganizes the PDF so the first page loads before the entire file downloads.
7. Audit fonts — Check for fully embedded fonts that could be subsetted, or system fonts that don't need embedding at all.