Extract Images From a PDF at Full Resolution

The usual way to get a picture out of a PDF is a screenshot. Zoom in, drag a box, paste. It works, and it throws away most of the image.

I tested it. I put a 3000 x 2000 pixel photo into a PDF at 3 x 2 inches, the size a photo usually takes in a report. Then I rendered the page at 150 DPI, which is what a typical "PDF to image" conversion does. The photo came out at 450 x 300 pixels: 135,000 pixels out of the 6,000,000 that were in the file. One forty-fourth. Extracting it instead returned a file byte for byte identical to the JPEG I'd put in.

Quick answer: To extract images from a PDF at full resolution, pull out the embedded image objects instead of rendering or screenshotting the page. Extract Images does this in your browser: drop the PDF, filter out icons, download the originals as JPEG, JPEG 2000 or PNG in a ZIP. Photos come out as the exact bytes the author embedded, at their native resolution.

Rendering a Page vs Extracting an Image

A PDF page is a set of drawing instructions. A photo on it is a separate object, an image with its own pixel size, plus an instruction that says "draw that image in this 3 x 2 inch box". The page doesn't care whether the image is 300 pixels wide or 3,000. It scales it to the box.

So there are two different things you can get out of a PDF:

• The page as it looks. A screenshot, or a PDF to JPG conversion, paints the whole page at some resolution and saves the result. Every image on the page is resampled to fit that resolution, and the output is re-compressed.
• The image as it was embedded. Extraction reads the image object itself and saves it. No resampling, no re-compression, no surrounding page.

In my test the embedded image was effectively 1000 pixels per inch, and pdfimages -list reports exactly that. To match it by rendering you'd need to render the page at 1000 DPI, an 8500 x 11000 pixel image, then crop it, and you'd still have re-compressed it.

This works in the other direction too. If an author dropped a 400-pixel web image onto a full page, extraction gives you those 400 pixels and no more. It can't add detail that was never there. What it guarantees is that you get everything that was.

What Comes Out, and in What Format

PDFs store images in a handful of compressed forms. A good extractor keeps the original bytes wherever the format allows and only converts when it has to. Here's what Extract Images does with each:

Most photos in PDFs are JPEGs, and those are the ones where the difference matters most: a screenshot re-compresses a JPEG on top of its existing compression, and each round adds artifacts. A PNG from a Flate-compressed image is lossless, so it's the same pixels as the original even though the file is new.

Two details people run into:

• A logo on every page comes out once. The tool recognises when 30 pages all point to the same image object and saves it a single time.
• Tiny images are noise. Bullets, icons and decorative rules are often images too. The width, height and file-size filters hide them, so a 40-page brochure doesn't hand you 200 files of which 180 are 12 x 12 pixel dots.

When Extraction Gets You the Wrong Thing

Extraction returns the image object, and sometimes the image object is not what you see on the page:

• Cropped photos come out uncropped. A PDF can draw an image through a clipping path, showing only part of it. You get the whole original, including the part the designer cut off. Usually a bonus; occasionally a surprise.
• Text or graphics on top of a photo aren't part of it. A headline over a hero image, an arrow pointing at something, a watermark: those are separate page content. The extracted photo is clean underneath.
• Charts and diagrams are often not images at all. A chart exported from Excel or a vector logo is drawn with lines and shapes. There's no image object to extract. The same goes for Adobe's own Export all images, which says "You can export raster images, but not vector objects."
• Some images are stored in pieces. A few PDF producers split a large picture into strips or tiles. You get the pieces, not the assembled picture.
• Tiny inline images are skipped. PDFs can also embed small images directly in the page's drawing instructions. The tool doesn't read those; they're almost always icons anyway.
• Transparent JPEGs lose their transparency. A PDF can pair a JPEG with a separate transparency mask. JPEG has no alpha channel, so the extracted original is the JPEG alone, without the mask.

When you want what the page looks like (the photo with the headline, the chart, the whole layout), render the page instead. PDF to JPG or PDF to PNG do that, and the guide to converting PDF to JPG covers picking a DPI that doesn't blur. To grab a region of a page, crop the PDF first and then render it.

Scanned PDFs Are One Big Image per Page

A scanned document is a stack of photos of paper, one per page. Extraction on a scan gives you each page's image at the resolution the scanner captured it, which is the cleanest copy you can get. Rendering a scan re-samples it to whatever DPI you pick, which is either upscaling the same blur or throwing pixels away.

If what you actually want from a scan is the text, extraction is the wrong tool and so is rendering. Run OCR PDF to get searchable text.

Extract Images From a PDF, Step by Step

1. Open Extract Images and drop in your PDF. Up to 20 at once.
2. Look at the grid. Under each image: its file name (which includes the page), pixel size, file size and format, marked "converted" when it isn't the original bytes.
3. Set filters if there's clutter. A minimum width of 200 pixels removes almost every icon.
4. Download one image, or everything as a ZIP. Files are named after the PDF and page, like report-p3-2.jpg.

Password-protected PDFs work, with one trade-off: the encrypted file is opened by a different engine that hands over decoded pixels only, so every image comes out as a PNG instead of the original JPEG. Same pixel dimensions, lossless, but larger files. The tool tells you when this happens.

Nothing is uploaded. The PDFs people pull images from are often not theirs to share: a supplier's catalog, a client's report, an internal deck. Running extraction in your browser means the file stays on your computer. Here's why OxygenPDF works that way.

Other Ways to Do It

• Adobe Acrobat Pro has Export PDF > Image > Export all images, with an option to skip images smaller than a set size. It re-encodes into the format you choose, so you pick JPEG and get a new JPEG, not necessarily the original bytes. It's also a paid product.
• pdfimages (part of poppler, on the command line) is the reference tool. pdfimages -list file.pdf lists every image with its size and resolution, and pdfimages -all file.pdf out writes JPEG, JPEG 2000, JBIG2 and CCITT images in their native format. It's what I used to check our tool's output. If you're comfortable in a terminal, it's excellent.
• Copy and paste from a PDF viewer sometimes copies the original image, depending on the viewer and the image. It's unpredictable, and it's one image at a time.
• Screenshots give you the image at screen resolution. Fine for a slide; poor for anything you plan to print or crop.

Frequently Asked Questions

How do I extract images from a PDF without losing quality?

Extract the embedded image objects instead of taking screenshots or converting pages to images. Extract Images saves JPEG and JPEG 2000 images as their original bytes and everything else as lossless PNG, so nothing is resampled or re-compressed.

Why is the extracted image bigger than it looks in the PDF?

Because the PDF scales images to fit the page. A photo shown at 3 inches wide can contain 3,000 pixels, ten times what a 300 DPI print of that space needs. Extraction gives you all of them.

Can I extract images from a scanned PDF?

Yes. Each page of a scan is usually one image, and extraction returns it at the scanner's original resolution. For the text on those pages, use OCR instead.

Why are some images missing?

They're probably not images. Charts, diagrams and many logos are vector drawings, which no image extractor can return; render the page to capture them. A few images use JBIG2 compression or are embedded inline as small icons, which the tool skips, and it reports how many it skipped.

Why do I get the full photo when only part of it shows in the PDF?

The PDF is clipping the image to a shape or frame. The embedded image is the whole original, and extraction saves all of it.

Can I extract images from a password-protected PDF?

Yes. Enter the password when asked. The images come out at full resolution, but as PNG rather than the original JPEG files, because the encrypted file is opened by an engine that only provides decoded pixels.

Get the Originals

A screenshot is a copy of what your screen happened to show. Extraction is the file the author actually used. When you need a photo, a figure or a scan from a PDF for anything beyond a quick look, take the original.

Extract the images from your PDF, at full resolution, without uploading it.