How to Extract Images from a PDF (Not Screenshots of the Pages)

A supplier sends a 40-page catalogue and you need the product shots out of it. Screenshotting each one throws away resolution, and converting the PDF to JPG hands you whole pages with the text baked in. What you actually want is the pictures themselves: the image files someone dropped into the document before they exported it.

Extracting images isn't converting pages

These are two different jobs and the tool names sound almost identical, so it's worth being clear about which one you need.

• PDF to JPG renders each page as one flat picture. Text, photos, page numbers and all, at whatever resolution you ask for.
• Extract images reaches inside the file and pulls out the individual embedded pictures, each as its own file, at the exact resolution it was stored at.

Nothing is re-encoded on the way out. A 3000×2000 photo comes back as a 3000×2000 photo even if it was shrunk to a thumbnail in the layout, because the PDF kept the full image and only displayed it small. That's often a pleasant surprise: the pictures are bigger than they looked on the page.

What lands in the ZIP

Everything comes back in a single ZIP so you're not clicking through 60 downloads. Each file is named after the page it came from, in the order it appears, so page7_23_Im4.png is the fourth image object on page 7. Sorting the folder by name gives you the pictures in reading order, which makes it easy to match them back to the document.

Formats are whatever the PDF stored, usually PNG or JPG. You may get more files than you expected: a page background, a logo repeated in the header of every page, a hairline rule that was saved as a 2-pixel image. Skim and delete rather than assuming each file is a photo.

When the picture you want doesn't come out

The tool only finds raster images, the kind made of pixels. Some things that look like pictures on the page aren't stored that way:

• Logos, icons and charts drawn as vectors.A chart exported from Illustrator or generated by a reporting tool is often a set of drawing instructions, not an image. There's no image object to pull out, and the tool will tell you it found none.
• Photos sliced into strips. Some layout software cuts a large image into horizontal bands when it writes the PDF. You get the bands, in order, and have to stitch them back together.
• Images cropped in the layout. The crop is a display instruction, so you get the full original picture back, including the parts that were hidden on the page.

For anything in that first category, render the page instead and crop the result. You lose the original resolution, but a 300 DPI page render of a vector chart is sharp enough for a slide deck or a document.

Scanned documents behave differently

If the PDF came out of a scanner or a phone camera app, every page is one big image. Extracting gives you exactly that: one photo per page, the scan itself, at the resolution it was scanned. That's genuinely useful when you want the raw scans back out of a bundle someone merged, and useless if you were hoping to isolate a figure printed on the page. For a figure, extract the page image and crop it.

Common questions

Will the extracted images be lower quality than in the PDF?

No. The embedded image data is copied out as-is rather than re-saved, so there's no second round of compression. If a photo was already squashed to 72 DPI when the document was made, that's what you get back. The extraction doesn't make it worse, but it can't recover detail the PDF never contained.

Can I pull the images out of just one page?

Not directly, but it's a two-step job: extract that page into its own PDF first, then run the image extraction on it. On a long document that's usually faster than digging through a ZIP full of headers and rules, and the filenames stay tidy.