When to use it
PDFs fall roughly into two kinds: vector text documents exported by layout software (contracts, papers, manuals), and image-based PDFs produced by scanning. The former are often only 20–80 KB per page of pure text; a single uncompressed colour A4 scan can reach around 20 MB. Anything absurdly large is almost always the second kind.
The most common case is hitting an attachment limit: mail servers often block at 20–25 MB, and government, recruitment and school systems often cap at 5 MB or 10 MB. Divide total size by page count to judge: only tens of KB per page means it is already tightly compressed and lossless mode will not help; hundreds of KB to several MB per page means rasterising has real room.
The other case is a file carrying things you should not send out: title, author, company name, producing software, and the hidden thumbnail Office leaves behind. They change nothing on screen but add size and leak information, so removing them is pure gain.
How to choose the settings
- Where lossless optimisation saves space: It repacks scattered small objects into object streams, rebuilds the cross-reference table, and clears the document information dictionary and thumbnails. The saving is usually only a few percent — clearing metadata will not make a file dramatically smaller.
- Sharpness (DPI): Pixel count grows with the square of DPI, so 150 to 300 is 4× the pixels. Choose 96–120 for on-screen reading; 120–150 for email or WeChat; 200 when printing is needed; and 300 only for small text, thin table rules or submission requirements. Above 300 there is no real gain for office documents, while size multiplies.
- JPEG quality: 60–75 is a comfortable range for on-screen reading. Below 55, visible blocky noise starts appearing at text edges; at 30–50 basically only the photo background is still viewable. To shrink the file further, drop the DPI one step before considering lower quality — lowering DPI fades the whole page evenly, while lowering quality produces local blocks, and the latter stands out more.
- A combination you can use as is: 120 DPI with quality 70 for on-screen reading; 150 DPI with quality 72 for email attachments; 200 DPI with quality 80 when printing is needed; for black-and-white scans tick greyscale first, which usually saves another chunk.
- Colour and encryption: Converting to greyscale saves 20%–40% of the size, but colour logos and charts turn grey, so it only suits pure black-and-white documents. Password-protected files must be unlocked before compressing, otherwise you may get blank pages.
Compared with other approaches
Compared with Ghostscript or Acrobat's Reduce File Size, our advantage is that you install nothing and upload nothing; we are weaker on granularity: they can downsample images inside a page individually, subset fonts and convert colour spaces, whereas we only have whole-page re-rendering and structural optimisation. When the bulk is fonts or vector art, they win.
Compared with online compression sites, the difference is mainly where the file goes. Here everything from reading to output happens locally in your browser — the PDF is never uploaded to any server, so there is no queue, no per-file size limit and no copy left on someone else's machine. For contracts and medical records that must not be sent out, that is a hard requirement. The trade-off is that your own computer does the work: a 300 DPI file of several dozen pages takes a while on an older machine.
One more thing worth stating plainly: small, searchable and sharp when printed cannot all be had for free. A text PDF is often only tens of KB per page, and rasterising it to 150 DPI images can actually make it larger while the text layer, links, bookmarks and form fields all disappear. For contracts, papers and official documents, use lossless mode only.
Additional Q&A
Why does lossless mode sometimes produce a larger file?
Because the output is re-serialised: the compression implementation and dictionary encoding differ from the original, so the same content can gain a few hundred bytes to a few KB. If the original already used object streams, repacking has nothing left to save.
What is the difference between 150 DPI and 300 DPI, and which should you choose?
The difference is pixel count — exactly 4× — and file size usually grows 2–4× with it. At 150 DPI you already see no jaggies on screen; 300 DPI only matters for printing small text or for submission requirements. If unsure, start with 150 DPI and quality 72 and decide from the result.
The scan is still large or blurry after compressing — what now?
Work out the cause first. Blurriness usually means the DPI was set too low — blur from downsampling a scan is irreversible, so raise it to 200 or 300 and the quality to 80 and look again. If it is still large, the original may already be compressed as black-and-white bilevel (only tens of KB per page); such files are already small and rasterising cannot help.
Can the text be recovered after rasterising?
It cannot be recovered. Once the text layer is replaced by an image it is irreversible, so always keep the original file. If the other party requires a searchable PDF, you can only use lossless mode, or do OCR separately yourself; this tool does not provide that step.