PDF Compression Explained: Lossless, Images, DPI and Quality Settings

Lossless and lossy describe different decisions

Lossless PDF optimization changes representation without intentionally discarding page information. Examples include efficient stream encoding or object-stream serialization. Results vary because many files already use these techniques.

Lossy optimization changes image samples or color information. It can deliver larger reductions for scans and photographs, but the outcome depends on resolution, encoding, scene detail, and how large the image appears on the page.

Effective DPI matters more than a raw pixel count

Dots per inch becomes meaningful when pixel dimensions are considered against the image’s displayed physical size. A 2400-pixel image filling an eight-inch page is roughly 300 pixels per inch; the same image displayed in a two-inch box has far more detail than most screens or office printers need.

A good engine evaluates each image in context and avoids upscaling. It should also distinguish photographs from line art and scanned text because aggressive JPEG compression can create halos around characters and sharp edges.

• ✓Preserve original samples in lossless mode.
• ✓Do not recompress efficient images automatically.
• ✓Treat soft masks and transparency carefully.
• ✓Inspect barcodes and QR codes after any lossy pass.

Quality sliders are not universal measurements

A JPEG quality value is an encoder instruction, not a cross-product fidelity score. The same number can produce different results with different encoders or content. PNG and monochrome scan formats use different mechanisms entirely.

For that reason, a product should name its actual operations and validate candidates. It should not claim a mathematically precise quality percentage unless that measurement is defined, reproducible, and meaningful for the document.

Structure can be more valuable than visual similarity

Text search, copy and paste, form fields, annotations, bookmarks, attachments, reading order, and accessibility tags are functional parts of a document. Two pages may look identical in a screenshot while one has lost all of those capabilities.

Compression acceptance should therefore combine structural checks with rendered comparison. When only partial validation is available, the result should say so clearly and keep the original available.