Compress PDF — Free Online, Quality Presets

PDF compression is one of those operations that's easy to underestimate until the moment you really need it — usually when an attachment refuses to send because it exceeds the recipient's mail server limit, or when a shared drive starts complaining about space, or when someone on a slow connection gives up waiting for a file to download. The reasons PDFs grow large are stable: scanned documents at high DPI carry significant pixel data, decks built around screenshots inherit the screenshot file sizes, reports with embedded charts and images accumulate weight, and any document that's been through several rounds of editing tends to carry version history that bloats the file. Compression knocks all of that down without changing what the document looks like to a reader.

The four-tier preset model is genuinely useful because it maps to actual delivery contexts rather than abstract quality numbers. The screen preset (72 DPI internally) is for documents that will only ever be viewed on a screen — typical for an email attachment that the recipient will open once, glance at, and probably never print. The ebook preset (150 DPI) is the balanced default — good for shared drives, document portals, and any case where the file might occasionally be printed but won't be the master copy. The printer preset (300 DPI) preserves enough detail for office printing and copying, which is the right choice for documents that might land on someone's desk as a physical paper. The prepress preset is for files heading to a professional print shop where every pixel matters and shrinking the file is a secondary concern.

The largest single source of bloat in most PDFs is embedded images, and the compression algorithm spends most of its effort there. A 5MB photograph embedded at full resolution in a PDF turns into a much smaller equivalent at the resampled DPI of the chosen preset, with the tradeoff being detail when the file is zoomed in. For documents where the photographs need to remain clear at zoom levels above the page size, the higher-DPI presets are the right choice; for documents where the photographs are illustrative rather than central, the screen or ebook preset is sufficient and the size savings are large. Knowing which photos need preservation is usually obvious from the document's purpose.

Text content survives compression well across all preset levels because text is stored as glyph references rather than as rasterised pixels. A PDF with mostly text and few images compresses surprisingly little — the floor is set by the text and font data, which is already small, and the gains from compression apply mostly to the images that aren't there. This is why pure-text PDFs (court documents, contracts, books) often compress only modestly, while image-heavy PDFs (illustrated reports, photo albums, presentation decks with embedded screenshots) can shrink to a quarter or less of their original size at aggressive presets.

Scanned PDFs are the special case worth understanding. A scan is essentially a series of large images, one per page, embedded in a PDF wrapper. The compression algorithm treats those images the same way it treats any other embedded image — resampling at the chosen DPI and re-encoding. The savings are typically dramatic: a 40MB scan at 300 DPI compressed at the screen preset can come down to 5MB or less, while still being legible on screen. For scanned documents where you've also run OCR (adding a searchable text layer), the OCR text layer survives compression and the resulting file remains both small and searchable, which is the best of both worlds for archiving.

Compression is fundamentally lossy when image resampling is involved, which is worth being explicit about. The original PDF is not recoverable from the compressed output; the tool here doesn't keep a copy and the resampling discards detail that can't be reconstructed. The right pattern for documents that might need both the small and the large versions is keeping the original in archive storage and producing the compressed copy on demand for distribution. The compressed copy can be regenerated from the archive if the recipient's needs change; the original is the canonical record. Treating the compressed file as a derivative rather than as the master is the disciplined approach.

Email attachment caps deserve a moment because they're the single most common reason this tool gets used. Gmail caps inline attachments at 25MB, Outlook 365 typically at 25MB, internal corporate mail servers at varying levels (10MB to 50MB depending on policy). When a PDF lands above the cap, the file silently fails to send or gets quarantined as too large; recipients sometimes don't even see the bounce notification. Compressing to a level that reliably passes through these caps — typically the ebook preset for image-heavy documents, the screen preset for scans — is the small step that prevents the recurring 'did you get my email' follow-ups that otherwise accumulate.

Shared drive and document management quotas are the other practical motivator. Box, SharePoint, Google Drive, and similar systems impose either per-user storage caps or per-organisation quotas; uploading uncompressed PDFs accumulates against those quotas faster than necessary. A team of twenty people each saving 100 uncompressed scans per month consumes terabytes per year; the same team compressing those scans before upload consumes a fraction of that, often with no visible difference to anyone who actually looks at the files. The cumulative storage savings translate directly into either lower fees or longer retention windows under a fixed budget.

There's a final operational note worth making for sensitive documents. Compression preserves the visible content but the resampling process introduces small differences from the original that are theoretically detectable. For most everyday cases this doesn't matter, but for documents where bitwise fidelity to the source is part of the value (forensic evidence, audit-grade financial records, certain regulatory submissions), compressing destroys that property. The right answer for those cases is keeping the uncompressed original alongside the compressed distribution copy, with appropriate metadata noting which is which.

There's a comparative note worth including about how this stacks against the alternatives. Adobe Acrobat Pro has a Reduce File Size feature that does similar work and produces comparable results, but it requires a paid subscription and is overkill for someone who only needs to compress a few files per month. Native macOS Preview can compress PDFs through Quartz filters but the quality at small sizes is noticeably worse than what Ghostscript produces here. Specialised desktop compressors (Smallpdf desktop, ILovePDF desktop) often paywall their compression features. The web tool here covers the common cases without requiring software installation or recurring fees, which suits the typical pattern of using compression occasionally rather than as a daily workflow.

A small detail about how compression interacts with already-compressed PDFs. If a PDF has already been compressed once at the screen preset, running it through compression again at the same preset doesn't make it meaningfully smaller — the operation is roughly idempotent at any given preset level. Going from screen preset to a more aggressive custom setting can extract additional savings, but the diminishing returns come quickly. The practical implication is that double-compressing the same file isn't useful; if the first compression didn't get the file small enough, the right move is choosing a more aggressive preset on the original rather than re-running on the already-compressed output.

Operationally the tool takes a single drop. Upload the PDF, choose a preset, click compress, download the result. The size estimate appears before the final download so you can preview the trade-off and adjust the preset if the result is too aggressive or too conservative. Files are processed in temporary storage, links expire quickly, no signup is required, no watermark is added, no per-day quota counts down in the background. Multiple PDFs can run through one after another, useful when compressing a backlog of scans for archive cleanup rather than a single one-off file. Most files process in a few seconds; very large multi-hundred-page sources take proportionally longer but still complete in a single pass.