How to reduce PDF file size

Reducing PDF file size sounds like one of those tasks that has a single button labelled “make smaller.” In reality it's a stack of trade-offs: image resolution, JPEG quality, font subsetting, metadata, stream packing. Each lever has a cost. This guide walks through what each one does, when to pull which, and what you'll lose at every tier — using Flint's Compress PDF tool as the worked example.

Where the bytes actually go

Before you compress anything, it's worth knowing what you're compressing. A PDF is a container holding a few distinct kinds of payload, in rough order of how much space they usually consume:

• Raster images. Photographs, scanned pages, embedded charts saved as PNGs. Usually 70-95 percent of the file size in any image-bearing PDF. This is where the wins come from.
• Embedded fonts. Full font files weighing 200-800 KB each. A PDF with five custom fonts can carry 2-3 MB of font data alone. Subsetting prunes this.
• Vector content. Paths, gradients, masks. Compresses well already; rarely the problem.
• Object streams and structural data. The plumbing — cross-reference tables, content streams, page objects. Older PDFs leave this loose; modern ones pack it.
• Metadata and incrementally-saved cruft. XMP metadata, version history from incremental saves, embedded thumbnails, unused objects. Small individually, can add up.

The four levers that actually move the needle

1. Image downsampling (DPI)

Images carry resolution. A photograph embedded at 600 DPI looks identical on a screen to the same image at 150 DPI — the viewer hardware can't even display the extra pixels. Downsampling reduces the stored pixel count to a target density. Common targets:

• 72-96 DPI — screen reading only. Tiny files, visibly soft if printed.
• 150 DPI — comfortable on-screen, acceptable home print. The sweet spot for most office documents.
• 300 DPI — professional print standard. No visible quality loss in print or on screen.
• 600 DPI+ — large-format or fine-detail print. Massive files. Overkill for almost anything that isn't a poster.

2. JPEG quality factor

Embedded photographs are usually re-encoded as JPEGs. JPEG's quality factor ranges from 1 (terrible, tiny) to 100 (visually lossless, large). The interesting range for document use is 65-90. Below 65, you start seeing compression artefacts on flat colour areas. Above 90, you're mostly paying for bytes nobody can see.

3. Font subsetting

Embedded fonts ship with every glyph the font designer drew — usually thousands of them. A subsetted font keeps only the glyphs the document actually uses. For a typical English-text document using a 600 KB font, subsetting drops it to 30-80 KB. Multiply across several embedded fonts and you can recover a megabyte without touching image quality.

4. Object stream packing and metadata stripping

Re-saving the PDF with packed object streams, removing unused objects, and stripping XMP metadata and document-history entries. The smallest of the four wins, but it's free — no quality compromise at all.

What Flint's presets actually do

Flint exposes three quality tiers rather than every raw knob. Each maps to a defensible combination of the levers above.

High quality

Light touch. Images downsampled only if above 300 DPI. JPEG quality around 85. Fonts subsetted, metadata stripped, streams re-packed. Expect 10-25 percent reduction. Use when print fidelity matters or the file is going to a designer.

Medium quality

The sensible default. Images downsampled to 150 DPI, JPEG quality around 75, everything else cleaned up. Expect 40-60 percent reduction on a typical image-bearing document. Indistinguishable from the original at normal viewing size. This is what we recommend unless you have a reason not to.

Low quality

Aggressive. Images downsampled to 96 DPI, JPEG quality around 65. Visible softening on photographs if you zoom in. Text and vectors untouched. Expect 60-80 percent reduction. Use when the email gateway is the hard constraint and quality is negotiable.

How to reduce file size in Flint

After compression — sensible next moves

• Sign before sending. Compression first, then Sign PDF. The signed document is based on the small one, so the recipient gets one tidy file.
• Merge related documents. Merge PDF after compression keeps the final pack lean. Compressing the merged result also works, but tends to give slightly worse ratios than compressing each piece first.
• Redact sensitive content. Redact PDF first (compression doesn't remove data, redaction does), then compress.
• Encrypt the result. Password Protect PDF adds AES-256 encryption — useful for anything containing personal or financial data.

Other ways to reduce PDF file size

Adobe Acrobat Pro

Acrobat's Optimize PDF dialog is the gold standard for granular control — separate sliders for colour images, greyscale images, monochrome images, plus font and metadata toggles. Costs about $20/month. Worth it if you live in Acrobat all day; overkill for occasional use.

Ghostscript

Command-line workflow: gs -dPDFSETTINGS=/ebook for screen-sized output, /screen for the smallest, /printer or /prepress for higher quality. Scriptable, free, not user-friendly. Useful if you're batching dozens of files in a pipeline.

macOS Preview's Quartz Filter

Free, one-click, built into the OS. Catch: Apple's default Reduce File Size filter is aggressive and applies a single target across all images regardless of content. Can leave text looking blurry on text-heavy documents.

Flint

Three sensible presets, content-aware processing, no install, no monthly subscription required to use the tool itself. The result lands in the editor so you can continue working — sign, merge, redact, lock. Bookmark Compress PDF if you do this more than once a month.

Tips for reliable compression results

• Don't double-compress. Running compression twice on the same file gives marginal extra savings and stacks JPEG artefacts. Go back to the original and pick a more aggressive tier instead.
• Compress after edits, not before. Adding text, signatures, or annotations grows the file slightly. Compress as the final step in your pipeline.
• Scanned documents are the biggest wins. A 200 MB scanned contract typically drops to 10-20 MB at medium without anyone noticing.
• If you need OCR, run it first. Compressed scanned PDFs OCR less reliably than the originals. Run OCR on the source, then compress the OCR'd version.
• Watch for already-compressed files. If the source PDF is already optimised, every tier gives single-digit savings. There's no magic — you can't compress what's already compressed.

Reducing PDF file size: technical FAQ

What DPI does each preset target?

High keeps images above 300 DPI as-is and only resamples anything higher. Medium targets 150 DPI. Low targets 96 DPI. Text resolution isn't affected at any tier — text is rendered from vector glyphs.

Does compression touch the text layer?

No. Selectable text remains selectable, searchable text remains searchable. Compression operates on the image and font payload, not the text content stream.

Will it break form fields, signatures, or annotations?

Form fields and form data carry across cleanly. Existing signature fields stay valid because we preserve the signed byte range. Annotations are preserved as PDF annotation objects, not flattened into the page.

What about PDF/A and other archival formats?

Compression produces a standard PDF. If you specifically need PDF/A output for archival, do that conversion as a separate step after compression.

Can I batch-compress files?

Not yet — one file at a time today, batch is on the roadmap. For now, process files sequentially; each pass takes seconds.

What's the max file size?

250 MB on Pro.

Ready to reduce yours?

Drop your file into Flint's Compress PDF tool and pick a tier. Medium covers most needs, low when size is the only thing that matters, high when fidelity is. From there, the rest of Flint's toolkit is one click away.