How to get the images out of a PDF

“Get the images out of a PDF” sounds like one job and is two, and picking the wrong one is why people end up with 400 tiny fragments, or with a blurry picture of a page they wanted at full size.

The two jobs are rendering and extracting. They answer different questions, they fail in different ways, and once you know which you are doing the rest is straightforward.

Rendering: a picture of the page

Rendering means running the page. A PDF page is a set of drawing instructions rather than a picture, so something has to execute them: place this text in this font here, fill this shape, draw this image into this box. What comes out is one picture per page, with everything on it, at a resolution you choose.

This is what most people actually want. If the plan is to put a page into a slide deck, attach it to a message, post it, print it, or crop something out of it, you want a picture of the page.

Because the page is instructions rather than pixels, you choose the resolution, and that choice is the only quality decision in the whole operation. An A4 page is 8.27 by 11.69 inches, so:

• 72 dpi gives 595 by 842 pixels, which is a thumbnail.
• 150 dpi gives 1,240 by 1,753, which is sharp on any screen.
• 300 dpi gives 2,480 by 3,508, which will print properly.
• 600 dpi gives 4,961 by 7,016, which is for cropping into a detail.

There is a limit to what rendering high buys you. Text and vector shapes are instructions, so they are drawn as finely as you ask and stay perfectly crisp at any resolution. Images that were placed in the document are not: if page four holds a 200 dpi scan, rendering at 600 dpi gives you a large picture of a 200 dpi scan. What dpi actually means has the arithmetic, including the trap of asking for more than the source has.

For format, use PNG for pages of text, tables or line art, because it stores every pixel exactly and text stays sharp. Use JPG for pages that are photographs, because the files are a fraction of the size and JPEG is good at photographs. This is the same decision as in does converting to PDF lose quality, turned around.

Extracting: the original pictures inside

Extracting means going into the file and pulling out the image objects it contains, at whatever size and compression they were stored with. No text, no layout, no page: just the pictures somebody placed.

This is the right job when you want the photograph itself. Somebody sends a brochure and you want the product shot. A report contains a chart you want at its own resolution. A PDF of a catalogue has forty pictures in it and you want all forty as files.

Three things surprise people, and they are all worth knowing before you start.

What is in the file is often smaller than the original. Word, PowerPoint, InDesign and most “export to PDF” dialogues downsample placed images by default, usually to 150 or 220 dpi. So the photograph inside the PDF may already be a reduced copy, and no extractor can recover what was thrown away when the document was made. If the extracted picture is disappointing, this is almost always why.

Not everything that looks like a picture is one. A chart made in Excel, a logo placed as a vector, a map drawn in Illustrator: those are shapes and text, not images, so there is nothing to extract. Rendering the page and cropping is the only way to get them as a picture.

One picture can be stored as several. Some producers cut large images into strips or tiles to keep memory down, and some store a photograph as one object with its transparency as a second, separate mask. An extractor faithfully hands you the pieces, which is why people sometimes end up with hundreds of fragments from a document with twelve pictures in it.

How to do each, with what you already have

Copy and paste, first. In Acrobat Reader, Preview on a Mac and several other readers, you can right-click a picture on the page and choose Copy Image, then paste it wherever you want it. This is extraction, it is free, it takes two seconds, and for one picture it is usually the whole answer.

Preview on a Mac will often let you simply drag an image off a page onto the desktop. It also renders: File, then Export, with a format and a resolution, gives you the page as a picture.

Acrobat Reader on any system will export pages as images through Convert or Export, though the free version limits what it will do. Its Snapshot tool, under Edit, copies a rectangle of the page at screen resolution, which is quick and is only as good as your zoom level at the time.

Your browser renders a PDF too. Opening a PDF in Chrome, Edge or Firefox and printing it to an image is possible but fiddly, and printing to PDF just gives you a PDF again.

A browser tool does the rendering job properly: pick the pages, pick the resolution in dots per inch, choose PNG or JPG, and download one file or a ZIP. The good ones do the work on your own device, which matters more here than almost anywhere else. The PDFs people extract images from are catalogues, reports, contracts, statements and scans of identity documents, and the free extractor sites upload all of it. Is it safe to upload photos online is about photographs and applies with more force to documents.

Which one do you need?

A short test.

Do you want the words and the layout visible in the result? Then you want rendering, because extraction throws them away.

Do you want the picture at the size the photographer took it, without the page around it? Then you want extraction, remembering that the document may hold a reduced copy.

Do you want to crop something out of a page? Render the page at a high resolution and crop. That works whether the thing you want is an image, a chart or a paragraph.

Do you want every picture from a 200 page catalogue? Extraction, and expect to sort through fragments.

Is the PDF a scan? Then every page is a single image, so the two jobs converge. Render at the resolution the scan was made at, or extract and get the same thing.

Why a picture comes out with a black or white box round it

This one catches people who extract rather than render, and the explanation is in how a PDF stores transparency.

A picture with a transparent background is not stored as one thing. It is stored as the colours, plus a separate greyscale image called a soft mask that says how opaque each pixel is. An extractor that hands you the objects as it finds them gives you two files: the colours, with whatever was in the transparent areas showing (often black, sometimes garbage), and a black and white silhouette beside it.

Nothing is wrong with the document. The two halves simply have to be put back together, which a rendering does for you and a raw extraction does not. If you have a colour image and a puzzling greyscale one of the same dimensions, that is what you are looking at: open both in an editor and use the second as the first one’s alpha channel, or render the page instead and save yourself the trouble.

The same thing explains why a logo extracted from a PDF sometimes arrives on a black square when it looked fine on the page.

One more option people forget

If you need one picture from one page and it is going into a message or a slide, a screenshot is a perfectly respectable answer. Zoom the PDF to a large size on screen first, so you are capturing plenty of pixels, then use the built-in capture: Shift, Command and 4 on a Mac, or the Windows key with Shift and S on Windows.

It is rendering, done by your PDF reader at whatever zoom you chose, and for a single picture it is faster than any tool. It stops being the right answer when you need many pages, a specific resolution, or a result that is not limited by the size of your screen.

The short version

Rendering gives you the page as it looks, at a resolution you choose, and works on everything. Extraction gives you the original pictures at the size they were stored, which is often smaller than you hoped and sometimes in pieces. Most people asking this question want rendering, at 150 dpi for a screen or 300 for print, as PNG for text and JPG for photographs. And whichever you want, the document is usually the kind of thing that should not be uploaded to a stranger’s server to get it.

The steps, in short

1. Decide which job you have. If you want the page as it looks, including text and layout, you want rendering. If you want the photograph somebody placed on page four, at its original size, you want extraction.
2. To render pages, choose a resolution. 150 dots per inch for reading on a screen, 300 for printing. An A4 page at 150 dpi comes out about 1,240 by 1,753 pixels.
3. To extract on a Mac, use Preview. Open the PDF in Preview, choose Tools then Show Inspector, or simply drag an image off the page onto the desktop, which works in many PDFs.
4. To extract on any system, try copy and paste first. In most readers you can right-click an image on the page and choose Copy Image, then paste it into anything. It is quick and it gives you the original if the reader supports it.

Questions

How do I extract images from a PDF for free?

Try right-clicking the picture in your PDF reader and choosing Copy Image, which works in Acrobat Reader and Preview and costs nothing. If you need whole pages rather than placed pictures, render them to PNG or JPG in your browser instead.

Why is the extracted image lower quality than I expected?

Usually because the person who made the PDF downsampled it. Word, PowerPoint and most export dialogues reduce placed pictures to 150 or 220 dpi by default, so what is in the file is already smaller than the original they started from. Nothing can recover that.

What is the difference between rendering and extracting?

Rendering draws the page, so you get everything on it as one picture at whatever resolution you ask for. Extracting takes out image objects the document already contains, at their own resolution, without the text or the layout around them.

Can I get an image out of a PDF at full resolution?

Extraction gives you the image exactly as it was stored, which is as full as the document goes. Rendering can give you any resolution you like, but rendering a 200 dpi scan at 600 dpi does not add detail; it just makes a bigger picture of the same thing.

Why does one picture come out as several pieces?

Some producers slice a large image into strips or tiles, and some build a picture from an image plus a separate mask. Extraction hands you the pieces as they are stored, which is one of the reasons rendering the page is often the more useful answer.

Can I extract the text as well?

Only if there is text in the file. A born-digital PDF has real text you can select and copy. A scanned one is a picture of a page, so getting words out of it needs OCR, which reads the shapes and guesses.

Is it safe to use an online PDF image extractor?

Most of them upload your document, and a PDF is usually a far more sensitive thing than a photo: contracts, statements, letters, scans of identity documents. A tool that works inside your browser does not send the file anywhere, and you can check that by going offline and using it anyway.