Someone sends you a brochure or a conference programme as a PDF, and the photograph you actually want is in the middle of page four. Screenshotting it gives you whatever your screen was showing. Converting the page to an image gives you the photo plus the headline, the caption and half a column of body text. What you want is the picture itself, at the size it was put in - and it is usually in there at a far higher resolution than the page lets on. Extract Images from PDF reaches inside the file and takes it out.
Why the pictures are bigger than they look
A PDF stores a photograph once, at whatever resolution it was given, and separately records how big to draw it on the page. Those two numbers are not related. A 4000-pixel-wide phone photo dropped into a report and dragged down to thumbnail size is still a 4000-pixel photo in the file; the page just paints it small.
That is why extraction so often beats every other route, and why picture-heavy PDFs are so much larger than people expect - most of the weight is full-resolution originals nobody ever downsampled.
It is also the difference between this tool and PDF to JPG. PDF to JPG photographs whole pages at a resolution you choose: text, background, borders and all. Extraction ignores the page and takes out the stored images. If you want a picture of a page, use the first; if you want the pictures on it, use the second.
Extract images from a PDF in four steps
- Open Extract Images from PDF and drop your file on it. It is read in your browser tab and never uploaded.
- Leave the page box blank to sweep the whole document, or type a range like 4-9 to look at part of it.
- Choose Original, PNG or JPG, and set how small an image has to be before it is ignored.
- Click Find images. Every picture appears as a thumbnail with its true pixel size; download one with the arrow on its card, or take the lot as a zip.
The whole sweep happens on your machine - which matters when the document is a client's brochure, a passport scan or a set of medical images.
Original, PNG or JPG
Original is the setting to leave alone. When a PDF embeds a photograph it very often stores the photographer's JPEG untouched, and where that is true the file is handed back byte for byte - not decoded, not re-compressed, not re-saved. What lands in your downloads folder is the same file that went in, down to the last bit.
That only holds when the stored bytes would open correctly on their own, and a PDF has several ways of making that untrue: transparency held in a separate mask a plain JPEG cannot carry, an inverted image, CMYK - which browsers cannot display at all - or a colour palette, so the numbers in the file are palette entries rather than colours. In any of those the picture is rebuilt from the decoded pixels instead, because the raw bytes would give you the wrong colours. You never have to work out which case you are in; the tool checks each image and picks.
Pick PNG when you want one format throughout and transparency kept - a logo lifted from a brand PDF is the usual reason. It throws no detail away on an opaque image, where every pixel comes back bit for bit, but it is not built for photographs: a rebuilt PNG of a photo is normally several times the size of the JPEG it came from. Pick JPG for small files where transparency does not matter; anything transparent is flattened onto white. If they still need to be smaller, Compress Image to Size takes them to an exact figure in KB.
A worked example: the photo that was 600 PPI all along
Take a product catalogue. On page three is a photograph of a chair, printed about two inches wide - a modest thumbnail in a column of text. Screenshotting it at that size gets you roughly 200 pixels across, unusable for anything but the web.
Run the page through extraction and the chair comes out at 1200 x 800. Divide the pixels by the printed width and you have the resolution the file was holding all along: 1200 / 2 = 600 pixels per inch, twice what a commercial printer asks for and about six times what a screen can use. The image was always that good; the page was drawing it small.
That photo is almost certainly stored as a plain JPEG, so on Original you get the photographer's file back untouched. PNG gives you the same 1200 x 800 pixels in a considerably larger file. JPG re-encodes them - visually fine, but a second round of JPEG on top of the first, which is exactly what Original exists to avoid.
When a PDF comes back with nothing in it
Plenty of documents genuinely contain no images, which surprises people who can see pictures on the page. Charts, logos, icons, maps and diagrams from a word processor or a design tool are usually vector artwork - lines, curves and fills, drawn fresh each time the page is displayed. There is no picture inside to take out, only instructions for drawing one. The same goes for the one-colour stencils PDFs use for shapes and shading.
The other reason is the size filter, and it is doing you a favour: documents are full of images that are not pictures - one-pixel gradient strips stretched across a header, rule artwork, dither tiles. The default skips anything under 24 pixels on either side. Slide it to 0 to see everything a file contains.
If the page really is a photograph - a scan, where the whole sheet is one big image - extraction will find it, and it comes out as the full scanned page. To pull the text out of that instead, use OCR PDF.
One picture, one file
A logo that appears on all forty pages of a report is stored once in the file and drawn forty times. It is saved once, too, rather than handed back as forty identical files with forty different names. The same goes for a photograph placed twice on one page.
That works because images are matched by their identity inside the document rather than by the name they turn up under - the same picture reached from a second page is announced under a completely different one, so matching on names would hand you duplicates.
What you get in the zip
One file per image, named after the PDF plus where it was found: catalogue-p3-2.jpg is the second image on page three of catalogue.pdf. That keeps the folder sortable by page when you are matching pictures back to the document. The archive stores the images rather than compressing them again - JPEG and PNG are already compressed, so a second pass would cost time and save close to nothing.
One thing extraction quietly does for you: images rebuilt as PNG or JPG carry no EXIF, so the camera model, timestamps and any GPS coordinates that travelled into the PDF do not travel out of it. Files taken out with Original keep whatever metadata they had, which is the honest trade for byte-identical bytes - run those through EXIF Viewer & Remover if you are about to publish them.
Frequently asked questions
- Is this the same as converting a PDF to images?
- No, and the difference matters. Converting a PDF to images photographs each whole page - text, background and layout included - at whatever resolution you choose, which is what you want for previews or for sending a page to someone who cannot open PDFs. Extracting images reaches inside the page and takes out the pictures themselves, at the resolution they are stored, with no text or background around them. If you want a picture of the page, convert it; if you want the pictures on it, extract them.
- Why is the extracted image a different size from what I see on the page?
- Because those are two separate numbers in the file. A PDF records the picture at one resolution and separately records how large to draw it, so a photograph placed as a small thumbnail can easily be several thousand pixels wide underneath. Extraction always gives you the stored version, which is the larger and more useful one. It can also be smaller than expected - a low-resolution image stretched across a page comes out at its real size, which is a useful thing to discover before you print it.
- Are the images uploaded anywhere?
- No. The PDF is opened and its images taken out entirely inside your browser tab, using your own machine's processor. Nothing is sent to a server, nothing is stored, and the tool keeps working with the network disconnected. For a document holding photographs of people, a portfolio you have not published, or scanned identity papers, that is the whole point.