Why a PDF's images are harder to reach than a Word document's
An Office file stores pictures as ordinary files inside a ZIP, so extracting them is a copy. A PDF has no such folder. Each page holds a content stream — a list of drawing instructions — and a picture appears as an operator saying "paint image object X here", with the object defined elsewhere in the file and possibly compressed with a scheme the page never names directly.
Extraction therefore means interpreting the page rather than reading a directory: walk the operator list, collect every image reference, resolve each object, and decode it. That is also why a picture may appear on twenty pages while existing only once in the file — the operator is repeated, the object is not.
How to extract images from a PDF
- Open the PDF. Drop the file onto the box above, or click to browse. Each page's drawing operations are read locally, with progress shown as pages are scanned.
- Wait for the scan. Every page is checked for image-painting operators, and each referenced picture is decoded once even if it appears on many pages.
- Review what was found. Pictures are listed largest first, with their pixel dimensions and the page they first appear on, so the full-resolution originals are at the top.
- Save the ones you want. Click Save PNG on any image. Each is written at the resolution stored in the document, not the size it happens to be displayed at.
What comes out
- Every embedded raster image — photographs, scans, screenshots and any logo stored as a picture.
- At stored resolution, which is frequently far larger than the size it appears on the page. A photograph placed in a small frame is still extracted at full size.
- Deduplicated, so a header logo repeated on every page is listed once rather than forty times.
- Ordered largest first, because the images worth having are almost always the biggest.
- As PNG, a lossless format, with the page number each image first appears on.
Which PDFs and images work
Any unencrypted PDF works, at any version. Images compressed with the usual schemes — JPEG, Flate, JBIG2, CCITT fax and JPEG 2000 — are decoded by PDF.js and come out as pixels regardless of how they were stored, so a scanned document yields its page images and a report yields its photographs.
Output is always PNG. That is a deliberate choice: re-encoding a JPEG to JPEG would lose quality a second time, while PNG is lossless. The file may therefore be larger than the original was inside the PDF. PDFs protected by an open password must be unlocked first.
Why the document is never uploaded
Pages are interpreted by PDF.js inside your browser and images are drawn to a local canvas. No server takes part, so the document is never transmitted and nothing survives closing the tab.
PDFs are where scanned identity documents, signed contracts, medical letters and bank statements live — and the images inside them are usually the most sensitive part of the file, because a scan is a picture of the original. Extracting locally is the difference between reading your own document and handing it to a stranger's server.
What cannot be extracted
The most common disappointment has a simple cause: vector artwork is not an image. Charts, diagrams, logos and illustrations drawn with lines, curves and fills are instructions, not pictures, so there is no file to save — a PDF that looks image-heavy but returns nothing is almost always vector throughout. To capture those, export the page as an image from a PDF reader instead.
Three further limits:
- Masks and transparency are flattened. An image with a soft-edged mask is extracted as its underlying rectangle, so you may get more than the page shows.
- Tiled scans come out in pieces. Some scanners store a page as horizontal strips, which extract as separate images rather than one page.
- Colour profiles are not applied. Images in exotic colour spaces are converted for display and may shift slightly.
Why people extract images from PDFs
- Reusing a photograph supplied only inside a report or brochure.
- Recovering scans — pulling the page images out of a scanned document to run through OCR at full resolution.
- Design and print work — retrieving original artwork when the source files are lost.
- Presentations — lifting a chart or diagram that was pasted in as a picture.
- Archiving — separating the images in a document for a media library.
Extracting images compared with screenshotting the page
Screenshotting captures the picture at screen resolution, after the PDF has scaled it down to fit the page. Extraction returns the file as stored — routinely several times larger. For anything you intend to print, crop or reuse, that difference is the whole point.
For the words rather than the pictures, use the PDF text extractor, or the PDF table extractor for tabular data. To find out whether a PDF is a scan before you start, the PDF metadata extractor reports whether a text layer exists. For pictures inside a Word, Excel or PowerPoint file, the Office image extractor copies them out losslessly instead.
For a full breakdown of what happens when you upload a PDF to an online service, read is it safe to upload a PDF online.
PDF image formats and extraction edge cases
PDF supports several raster image filters: DCTDecode (JPEG), FlateDecode
(zlib/deflate), CCITTFaxDecode (Group 3/4 fax — common in black-and-white scans),
JBIG2Decode (bilevel compression used by Acrobat's scanning workflow) and JPXDecode (JPEG
2000, rare). Both inline images — defined with BI/ID/EI
operators in the content stream — and XObject images referenced by name appear in the
output.
Three edge cases: (a) an image with a separate soft-mask (SMask) stream carries its transparency channel there — the extractor composites the alpha before saving, so the downloaded file has a proper alpha channel rather than a white background; (b) the same XObject may be referenced many times on a page (a logo in every header and footer) — the extractor deduplicates by XObject name and saves it once; (c) vector drawings built from PDF path operators are not raster images and are not extracted — only XObject images and inline rasters are included.
Frequently asked questions
How do I extract all the images from a PDF?
Drop the PDF onto this page. Every embedded picture is listed with a Save PNG button. Browsers restrict saving many files at once, so images are saved individually.
Is my PDF uploaded to a server?
No. Pages are interpreted in your browser and images are drawn to a local canvas. Nothing is transmitted.
Why did my PDF return no images?
Its artwork is vector — drawn with lines and shapes rather than stored as pictures. Charts, logos and diagrams are usually vector, so there is no image file inside to extract.
What resolution do I get?
The resolution stored in the document, which is often much higher than the size the picture appears at on the page.
Why are images saved as PNG rather than JPEG?
PNG is lossless. Re-encoding an extracted JPEG back to JPEG would degrade it a second time, so PNG is used even though the file can be larger.
Why does one picture appear as several images?
Some scanners store a page as horizontal strips. Each strip is a separate image object in the PDF, so each extracts separately.
Can it extract images from a scanned PDF?
Yes — a scanned page is itself an embedded image, so it extracts at full resolution. That is the best input for OCR.
Does it work on Mac, Linux or a phone?
Yes. Everything runs in the browser, so no PDF software is needed.