How to Compress a PDF Without Losing Quality

Why some PDFs are huge, what compression changes inside the file, and how to pick settings that shrink the size while keeping text sharp and photos clear.

NewKitApp19 de outubro de 2026 9 min de leitura

Nossos guias estão disponíveis em inglês e indonésio. Os artigos abaixo abrem em inglês.

You've finished the document. It looks perfect. Then the email bounces with "attachment too large", or the job portal rejects it because it's over 2 MB. Compressing PDFs is one of the most searched document tasks for exactly this reason, and it's also the one where people most often wreck their files: blurry text, smeared photos, signatures that turn into grey smudges.

It doesn't have to go that way. Once you know where the bytes in a PDF come from, you can shrink most files by 70 to 95 percent and keep them looking the same at normal viewing size. This guide explains what makes PDFs big, what compression changes, which settings to use for which job, and when to stop compressing and do something else.

Where the size comes from

A PDF holds three kinds of content, and they weigh very different amounts.

Text and vector drawings are instructions: "draw the letters of this word in this font at this position", "draw a line from here to there". A full page of text needs a few kilobytes of instructions. Charts and diagrams exported from Excel or PowerPoint as vectors are similarly light.

Fonts are embedded so the document looks the same on every computer. A full font file can weigh 100 to 500 KB, but most PDF makers embed only the characters the document uses (a "subset"), so a typical report carries 20 to 80 KB of font data per typeface. That's paid once for the whole document, not per page.

Images are where the weight lives. A single photo from a modern phone holds 12 to 50 million pixels. Even after JPEG compression, it weighs 3 to 8 MB. A scanned page at 300 DPI is an image of about 9 million pixels. Put twenty photos or scans in a document and you have a 50 MB PDF, while the text around them weighs less than a single image.

So the first question to ask about a large PDF is: how many images does it contain, and how big are they? If the answer is "lots, and large", compression will work wonders. If the file is mostly text, compression will barely move it, and you need a different fix.

What a compressor changes

There are two broad approaches, and the difference matters.

Approach 1: recompress the images, keep everything else

The compressor finds every image inside the PDF, decodes it, and saves it again at a lower size. Two settings control the result:

Resolution. An image stored at 4000 pixels wide but displayed in a 10 cm box has far more pixels than any screen or printer can show. Scaling it down to, say, 1600 pixels removes data nobody would see.
JPEG quality. JPEG throws away fine detail the eye notices least. At quality 85 you rarely see a difference; at 70 you need to look closely; at 50 you start to see blockiness around sharp edges.

Text, fonts and vector graphics stay untouched. The text remains selectable and searchable, and it stays perfectly sharp because it was never an image.

This is how the NewKitApp compressor works at its three standard levels:

Level	Longest side kept	JPEG quality	Good for
Light	2400 px	82%	printing, archiving
Balanced	1600 px	68%	email, upload portals
Strong	1100 px	50%	the smallest file, screen only

The tool only replaces an image if the new version is at least 5 percent smaller, and it never gives you a file larger than the one you started with.

Approach 2: turn every page into a picture

Some compressors render each whole page, text included, into one image and build a new PDF from those images. This shrinks scanned documents dramatically, because a scan is already one image per page. For a normal document, though, it's destructive: the text becomes a picture, you can no longer select or search it, and screen readers can't read it aloud.

NewKitApp offers this as an option called Maximum compression, meant for scans. Leave it off for anything with real text.

Choosing settings for the job

Email attachments

Most email services cap attachments at 20 to 25 MB, and many company servers cap them at 10 MB. Use Balanced. Photos stay clear on screen and the document usually lands well under the limit. If it doesn't, read the section on splitting below.

Job and government portals

Portals often set limits between 1 and 5 MB. Start with Balanced. If you're still over, try Strong and check the result at 100 percent zoom. CVs and cover letters are usually text, so they're already small; the heavy parts are scanned certificates and ID photos.

Printing

Use Light, or don't compress at all. Printers need more resolution than screens. A photo that looks fine on a laptop at 1100 pixels wide will print soft on an A4 page.

Archiving

For long-term storage, keep the original and a compressed copy if space allows. Compression is lossy: you can't get the removed detail back.

Scanned documents

Scans are the one case where Maximum compression makes sense. A 40-page scan at 300 DPI can drop from 60 MB to around 5 MB. The text in a scan was never selectable, so you lose nothing on that front. If the scan is black-and-white text, also consider Grayscale PDF first, because a gray image needs a third of the data of a colour one.

Step by step

Open Compress PDF and drop your file on the page.
Pick Balanced.
Click Compress PDF and watch the progress bar.
Look at the result: the tool tells you the new size and the percentage saved.
Open the downloaded file and zoom to 100 percent on a page with photos. Check that they look acceptable for how the document will be used.
If it's still too big, run the original (not the compressed copy) again with Strong.

That last point matters. Compressing an already compressed file again loses detail twice for a small extra saving. Always go back to the original when you change settings.

Why some PDFs refuse to shrink

You run the compressor and get a 3 percent saving. A few common reasons:

The file is mostly text. A 4 MB PDF of text is usually carrying embedded fonts, possibly several complete ones. The compressor can't remove fonts without breaking the text. Check with your PDF reader's document properties (Ctrl+D in many readers): the Fonts tab lists what's embedded. If you see a dozen fonts, the file was exported with "embed all fonts". Re-exporting from the original application with subsetting enabled helps more than any compressor.

The images are already compressed. PDFs exported from Word or Google Docs with "optimize for web" or similar settings already have downsampled images.

The images are special types. Images with transparency, CMYK images from print workflows and certain mask types are skipped by careful compressors, because recompressing them would change colours or break transparency. Brochures and print-ready files often contain these.

The bulk is something other than images. Some PDFs carry attachments (a source spreadsheet, for instance) or large amounts of metadata. Sanitize PDF removes embedded files and metadata and can free surprising amounts of space.

Alternatives to compression

When compression alone isn't enough, three other moves help.

Split the document

If a portal allows several uploads, split the file into parts. Split PDF with the "Every N pages" mode cuts a file into equal parts; increase N until each part fits under the limit.

Remove what isn't needed

A 30-page scan with blank backs is really a 15-page document carrying 15 pages of scanner noise. Delete PDF Pages removes them. Appendices nobody asked for can go too.

Fix the source

If you made the PDF yourself, the cheapest fix is often upstream:

Resize photos before you put them in a Word document. Word's "Compress Pictures" option (on the Picture Format tab) set to 150 ppi removes most of the excess.
When exporting to PDF from Word, choose "Standard" or "Minimum size" rather than "High quality print" if the document is for screen.
When scanning, choose 200 DPI rather than 600 for text documents, and grayscale rather than colour if colour adds nothing.
When turning phone photos into a PDF with JPG to PDF, compress afterwards. Phone photos are much larger than they need to be for a document.

How much quality do you lose?

At Balanced settings, viewed on a laptop or phone at normal size, most people can't tell compressed and original photos apart. Differences appear when you zoom in on fine detail: hair, foliage, small text inside a photo (a street sign, a label). JPEG artefacts show up first as slight blockiness around high-contrast edges.

At Strong settings, photos look softer, and small text inside photos may become hard to read. That's acceptable for a portfolio thumbnail page, not for a scan of a receipt where someone needs to read the numbers.

Two kinds of content deserve extra care:

Signatures and stamps on scanned pages. They're often thin lines that JPEG compression smears. Check them at 100 percent zoom.
Screenshots with small text. JPEG is a poor format for them. If your PDF is full of screenshots, Light is a safer level than Strong.

Privacy: where your file goes when you compress it

Most "compress PDF online" services upload your file to their server, compress it there and let you download the result. The file stays on their machines for a while. For a holiday brochure that's fine; for a payslip, a medical report or a passport scan, you're handing a copy of sensitive data to a company you know little about.

NewKitApp's compressor runs in your browser. The file is read from your disk, the images are recompressed using your browser's own JPEG encoder, and the result is saved back to your disk. You can confirm this by opening your browser's developer tools on the Network tab: no request carries your document. For large files there's a side benefit: you don't wait for a 50 MB upload and a 5 MB download.

A worked example

Here's a real case to make the numbers concrete. A property agent sends a 14-page brochure for a flat: two pages of text, a floor plan drawn as vectors, and twenty-two photos straight from a camera. The PDF weighs 38 MB, and the agency's email server rejects anything over 15 MB.

Running it through Balanced compression gives these results:

The twenty-two photos, originally 4000 by 3000 pixels each, are scaled to 1600 by 1200 and re-encoded at 68 percent quality. Each drops from about 1.6 MB to about 140 KB.
The floor plan, a vector drawing, isn't touched. It stays sharp at any zoom.
The text, set in two embedded fonts, isn't touched either.
The final file weighs 3.4 MB, a saving of 91 percent.

On a laptop screen, the photos look identical to the originals. Zoomed to 200 percent on a kitchen shot, you can see slight softness in the tile grout. Nobody reading a property brochure will zoom that far, so Balanced is the right choice. If the same brochure were going to a print shop for glossy A4 copies, Light would be the better pick, giving a file of about 9 MB with more detail to spare.

The lesson: decide where the document is going before you choose a level. Screen and print have different needs, and a setting that's perfect for one is wrong for the other.

A practical routine

When a PDF is too big, run through this list:

Check what's inside. Lots of photos or scans? Compression will work. Mostly text? Look at fonts and attachments instead.
Compress with Balanced. Check the result at 100 percent zoom.
Still too big? Go back to the original and try Strong.
A scan? Try Grayscale first, then Maximum compression.
Still too big? Delete unneeded pages or split the file.
Before sending, open the final file and look at the pages that matter most: signatures, numbers, photos of documents.

Compression is a trade: bytes for detail you hopefully don't need. Pick the level that matches how the document will be used, always start from the original, and look at the result before you send it. Do that and nobody receiving your file will notice it was ever compressed.