Compress PDF

Compress PDF makes a file smaller by re-encoding the images inside it at a lower quality and, if they're larger than needed, scaling them down — the images in a typical scanned or photo-heavy PDF account for nearly all of its size, so shrinking those is what actually moves the number. Text and vector graphics (real typed text, drawn lines and shapes) are left completely untouched: they aren't rasterised into pictures, which is how some compressors get an impressive-looking size reduction at the cost of making the document's text no longer searchable, selectable, or accessible to a screen reader. That trade-off is deliberate — a PDF that's smaller but broken for reading and copying isn't actually an improvement.

Two ways to compress are offered, because "make it smaller" means different things depending on why you're doing it. Choosing a level picks one of three fixed presets — Light, Balanced, or Strong — each pairing a JPEG quality setting with a maximum image dimension, applied once in a single pass. Aiming for a size — useful when a portal or application has a hard cap like "under 2MB" — instead runs a bisection search across quality settings, narrowing in step by step on the highest quality that still lands under your target, and tells you plainly if nothing it tried got there (some scans genuinely can't be compressed into a very small target without becoming illegible). Not every image is touched: ones the tool can't safely re-encode without risking a corrupted page are left exactly as they were, which is a deliberate conservative choice over guessing.

That conservatism extends to how images are evaluated before touching them at all: the tool checks each image's encoding filter, colour space, bit depth, and whether it carries a mask, and skips re-encoding any image it isn't confident it understands correctly. Getting a compression estimate wrong just costs a bit of extra file size; guessing wrong about how to re-encode an image can corrupt the page, which is why the tool errs toward leaving something alone over risking that. This is also why a text-only PDF barely shrinks when you compress it — there's almost nothing but text and vector content in it, and that's exactly the part this tool doesn't touch.