Compress a PDF Without Turning It Into a Picture of Itself

How it works

1. Choose the PDF that will not send, or drop the whole folder of them, up to twenty at once. They are opened on this device and never uploaded.
2. Read what is actually inside: pages, pictures, and how much of the file the pictures are. A folder gives you the list instead, with what each one weighs. Then pick how small you need it.
3. Look at the before and after on the biggest picture, then shrink and download. One file comes back as a file, a folder comes back as one zip with the same names inside. If nothing gets smaller, nothing downloads and your originals are left alone.

What you get

• Says what is inside before it touches anything: pages, pictures, and how many megabytes those pictures are
• A before and after of the biggest picture in the file, both at the same size on screen
• Three settings written in plain English instead of a slider from 1 to 100
• Text stays text, so the document is still selectable, still searchable and still copies out
• Leaves logos, signature images and anything it cannot shrink honestly exactly as it found them
• Downloads nothing at all when the file will not actually get smaller, rather than handing you the same file back
• Takes a whole folder, up to twenty files at once, and hands them back as one zip with the same names inside
• Puts the files that could not shrink into that zip untouched, so what you get back is the folder again and not a set of gaps to fill in

The file is 38 MB and their mail server stops at 10

This is the most ordinary failure in contractor paperwork and it always happens at the worst time. The progress bar creeps to the end and the message bounces: attachment too large. Gmail draws the line at 25 MB, Outlook at 20, and plenty of general contractors and property managers run mail servers that stop at 10. A photo report, a scanned and signed contract, a submittal with cut sheets in it, any of these clears all three limits without trying.

Almost always the size is pictures and nothing else. A modern phone camera writes a 4 MB photo. Twenty of them in a report is 80 MB before a single word of text is added. A desktop scanner set to 600 dpi color writes about 8 MB a page for a document that is black ink on white paper. The words in that document, if they were words instead of a picture of words, would be a few kilobytes for the whole file.

So the size is not really the document. It is resolution that nobody is ever going to look at. A photo of a cracked slab is exactly as useful at 1800 pixels across as it is at 4032, because nobody is printing it on a billboard, and the difference between those two numbers is the difference between a file that sends and a file that does not.

That is the whole job here. Bring the pictures down to a size a human eye cannot tell apart on a screen or on paper, leave everything else exactly as it is, and hand the file back.

What it changes, and what it refuses to touch

It re-encodes the JPEG pictures inside the file. Each one is decoded, drawn at a smaller size if it is larger than the setting you picked, saved back as a JPEG at that setting, and dropped into the file in place of the old one. Everything about the page stays where it was: the picture still sits in the same box on the same page at the same size on paper, it is just made of fewer pixels.

It does not touch the text. This is the part that matters and it is where most compressors quietly cheat. The fast way to make any PDF small is to draw every page as one flat picture and throw the real content away. The number on screen looks great and the document is ruined: nothing in it can be selected, copied, searched for, or read out by anything. A contract that cannot be searched is a contract nobody can find the retainage clause in. Vector lines on a drawing become fuzzy edges the moment anyone zooms. That trade is never made here, and never made silently.

It steps around anything it cannot shrink honestly. Pictures under about twelve kilobytes are left alone, because those are logos, signature images and letterhead marks, and re-encoding one of those saves nothing and can visibly wreck it. Images that are not JPEGs are left alone, because making those smaller means guessing at what they are, and a guess here is somebody's drawing. Images that act as a mask for another image are left alone, because changing one of those changes the picture on top of it.

And it checks its own work per picture. If a picture comes back from the re-encode bigger than it went in, which happens when it was already saved harder than the setting you chose, the original is kept. A compressor that grows the file is worse than no compressor at all.

The three settings, and what each one costs

Email it, about 1200 pixels on the long side. This is the hardest squeeze and it is the right choice when the only thing that matters is that the message sends. A photo at 1200 pixels fills a laptop screen and looks completely normal on a phone. Printed at full page size you would start to see it, which is why it is not the default.

Print it, about 1800 pixels on the long side, and the default. A photo at this size prints cleanly at the size photos are actually printed in a report, four or six to a page, and it still shrinks a phone photo by most of its weight. If you do not have a reason to choose otherwise, this is the one.

Keep the detail, about 2400 pixels on the long side. Use this when somebody is going to zoom in: a photo of a serial number on a condenser, a close up of a crack you are arguing about, a scanned drawing with small dimensions on it. It shrinks the least and it keeps the most.

None of these three upsize anything. A picture already smaller than the setting is left at the size it is and only re-encoded, and if that does not make it smaller it is not touched at all.

Look at it before you commit

Every compressor on the internet shows you a percentage and asks you to trust it. A percentage tells you nothing about whether the writing on the scanned page is still readable, and by the time you find out, the version you sent is already in somebody's inbox.

So the biggest picture in your file is shown twice, before and after, as the same patch of the same picture at the same size on screen. That last part is the part everybody else skips. A whole page shrunk down to thumbnail size always looks fine, because shrinking a picture is what hides the damage in the first place. A crop at the size you would actually view it is the only comparison that means anything.

Change the setting and both sides update, with the file size of that one picture under each. If Email it looks fine to you, use Email it. If the writing goes soft, you can see that it went soft before you send anything.

It is rarely one file, it is the folder

Nobody keeps one oversized PDF on its own. It is a week of photo reports off the same phone, a stack of signed waivers a subcontractor scanned at 600 dpi, every piece of a closeout packet sitting in one folder and every piece of it too heavy to send. Doing that one file at a time is the same three clicks twenty times over, and it is the reason most people give up and send a download link instead.

So hand over the folder. Up to twenty files go in at once and you get a line for each, with what it weighs now. They are squeezed one after another on this device, exactly the way a single file is, and they come back as one zip called pdfs-smaller.zip with your own names inside it.

One of them is shown in the before and after, the first one with a picture in it, named so you know which one you are looking at. That is on purpose. A sample of the real thing at the size you would view it tells you whether the setting is right; a percentage across twenty files tells you nothing you can check.

The files that could not shrink go into that zip untouched, byte for byte as you handed them over. That is deliberate too. A zip with holes in it is a folder somebody now has to rebuild by hand, working out which three of the twenty are missing and where the originals went. The zip you get back is the folder again, ready to attach.

The only thing left out is a file that will not open at all, a password protected one for instance, and it is named on its own line so you know which original to hold on to. And if not one file in the batch came out smaller, nothing downloads at all, the same rule as for a single file.

When this will not help, and the page says so plainly

A PDF with no pictures in it will not shrink here, and it is told to you in those words before you press anything. A generated change order or a contract written out as text is already tiny, usually under a hundred kilobytes for a document of real length, because text in a PDF is just text. There is nothing in it to squeeze.

A scan saved in an older format will not shrink either. Some office copiers write fax style or flate encoded images rather than JPEGs, and those are left alone on purpose rather than converted, because converting them means deciding what they should look like. The page tells you that is what it found instead of running and reporting a one percent saving.

A file that somebody has already compressed will not go much further. If it came off a file sharing site or through another tool, the pictures in it may already be at or below the setting you picked, and every one of them will be kept as it is. When nothing in the file actually got smaller, nothing downloads at all. You are not handed a copy with a new name and the same size.

And the one thing no browser can do: a password protected PDF cannot be opened by anything without the password, including this. Open it in a reader, save an unprotected copy, and start from that.

What to do before you send it

Build the whole thing first, then shrink it once at the end. If the report is photos plus an invoice plus a signed waiver, turn the photos into a captioned report, put the three files together, and compress the finished packet. Shrinking each piece separately and then merging gives you the same result with more steps.

Number the pages before you send it so a question about it can be pointed at a page instead of described. This is worth doing on anything over about five pages and it costs one click.

Check what is in it. A photo report assembled in a hurry has a way of carrying a picture nobody meant to send, and a scanned packet often carries a subcontractor's pricing on a page you skipped past. Black out what should not travel before it leaves, because after it leaves is too late.

If it still will not send after all of that, split it in two and send two messages. That is not elegant, but a photo report is a stack of pictures, and past a certain number of them no amount of compression gets a hundred photographs under ten megabytes while they are still worth looking at.

Questions people ask

Related