Compress a PDF to an exact size: how it actually works
How does a tool hit "under 100KB" exactly? A plain-language explanation of target-size PDF compression — sampling, estimation, the reachable floor — and why honest tools tell you when it is impossible.
Quality presets guess; target-size works backwards
A normal compressor asks "how much quality do you want to lose?" and the output size is whatever it is. A target-size compressor inverts the question: given your byte limit, it computes which image resolution and encoding quality will land just under it.
Paplume does this by sampling a few pages at a probe setting, measuring the bytes they produce, extrapolating to the whole document, then tuning scale and quality until the projection fits your target.
The reachable floor — and why honesty matters here
Every PDF has a physical floor: the text layer, fonts and structure cannot be compressed away. A 30-page scan cannot become 100KB and stay readable. Tools that pretend otherwise output an unreadable file and count it as success.
Paplume computes an estimated floor before processing and shows it up front. If your target is below the floor it says so — and the estimate consumes no quota. If compression would not shrink the file at all, it tells you and consumes no quota either.
What stays sharp and what does not
Real text and vector graphics are never rasterized — they stay crisp at any compression level. Images are downsampled and re-encoded, so photos soften as the target shrinks. In a scanned PDF the "text" is itself an image, which is why heavy targets visibly soften scans.
Step by step
- 1
Open the target-size compressor
Free during Beta, no sign-up; runs locally in your browser.
- 2
Enter your exact limit
e.g. 100KB, 200KB or 1MB — whatever the portal demands.
- 3
Read the estimate
The reachable floor shows before processing. Below the floor, the tool says so instead of producing garbage.
- 4
Run and verify
Compress, check the real output size on the result card, and download. Failures refund your quota.
FAQ
Why is the result a bit under my target rather than exact?
The tool aims safely under your limit so the upload never bounces. Landing a few KB below the cap is intentional.
Is the estimate ever wrong?
It is an extrapolation from sampled pages, so unusual documents can drift from it. The result card always shows the true final size, and the estimate itself costs no quota.
Does this upload my file?
No — sampling, estimation and compression all run locally in your browser. Verify in DevTools → Network.