PDFGym
EN

Extract images from a PDF

or drop one here
Choose a PDF first.

Many PDFs carry their photos as separate objects stored between the words, and grabbing them with a screenshot throws quality away. This page opens that hidden storage, lists every picture with its pixel size and format, and saves exactly the ones you tick. Stored JPEGs come out identical to what went in, and the file never leaves your own computer.

The logo trapped in a 40-page brand book

When someone answers a request for image files by sending a PDF, pulling the stored originals out beats redrawing or screenshotting them.

Take a freelance designer facing a Friday morning deadline. The client needs the company logo for a two-metre trade-show banner, and the only file in the inbox is brand-book-acme.pdf: 40 pages, 31 MB, the logo on page 3 and product photos scattered from page 12 onward. Asking the client for source artwork would cost a day nobody has.

The catalog shows the page-3 logo stored at 2,400 px wide even though it prints barely 5 cm across, which is plenty for the banner printer. The designer ticks that one object plus six product shots, leaves two dozen tiny icons and background tiles unticked, and downloads a single ZIP. If the real trouble had been the 31 MB attachment limit instead, Size breakdown would show which pages carry the weight, and Compress PDF would shrink them in place.

From file to saved pictures in four passes

  1. Open the PDF

    Press Choose a PDF, or drop the file anywhere on the page. The scan starts at once and the catalog of found pictures appears underneath when it finishes.

  2. Tick the pictures you need

    Every card names its page and position, such as Page 12 · image 3, with its dimensions in pixels, placed resolution, codec, colour space, and stored file size. The list opens on Larger and Extractable, so objects under 256 px on a side and anything the browser cannot verify stay hidden until you switch the filters to Any size or All found. Use Select and Deselect to change many cards at once.

  3. Pick how they should come out

    Keep original files copies stored JPEGs without touching a byte and writes supported raw samples as checked PNG files. Convert to PNG turns every compatible picture into PNG instead, and hides the ones whose masks or sample mapping make conversion unsafe.

  4. Extract, check, download

    Press Extract selected images. Each result is opened again, its dimensions are compared with the recorded size, and copied JPEG streams are matched against the stored bytes before the download button appears. Several pictures arrive in one ZIP named after the source file, such as brand-book-acme-images.zip; a single picture downloads on its own with the source name in front, and every file inside follows the page-012-image-03 pattern so the set sorts in reading order.

How to read the cards and the receipt

Two numbers on each card answer different questions. The pixel size is what you take home; the DPI figure describes how densely those pixels sit on the printed page. A 2,400 px logo placed at 300 DPI prints crisply at 20 cm wide, while the same pixels stretched across a poster would look soft. Judge print jobs by stored pixels, not by how large the picture looked in the reader.

After the run, the receipt keeps every object accounted for: how many pictures were found, extracted, skipped, or left unticked, which codecs appeared, how many files changed format, how the byte counts compare, and how many repeated objects were drawn more than once yet written a single time. A proof line above it counts the files that reopened cleanly, the dimension checks that passed, and the stored streams copied exactly.

When one logo prints on ten pages, the PDF usually stores it once and draws it ten times. The tool writes that object a single time, names it after the page where it first appears, and the Duplicates row tells you how many objects behaved this way.

When a picture refuses to come out

A card marked Skipped is the tool declining to guess, and the reason sits on the card. Stencil masks carry shape without colour, so there is no picture to save. An image with a separate transparency mask is incomplete on its own. Fax-era CCITT and JBIG2 scans need a full page render to become viewable, and some JPEG 2000 streams cannot be reopened by the browser for the final check.

Two more refusals protect quality: images whose samples remap on conversion, and raw samples in depths or colour spaces the converter cannot prove. Switching from Convert to PNG back to Keep original files brings back every object that is safe to copy but unsafe to convert.

Missing pictures are usually a filter, not a failure. The catalog opens on Larger, hiding everything under 256 px on a side, and on Extractable, hiding refused objects. Move both filters to Any size and All found to audit the whole document, then tick only what you need. Tiny inline graphics stay counted but unextracted by design, and anything you leave unticked appears as not selected in the receipt rather than silently vanishing.

When the page matters more than the pictures

Extraction returns loose objects without captions, surrounding text, or layout. When the assignment needs the photograph together with its headline and caption, render the whole page instead with PDF to images or PDF to JPG.

If you pulled pictures out only to rebuild a lighter handout, Images to PDF packs a chosen set back into one document. And when print quality is the goal, remember extraction cannot invent pixels: a 400 px web graphic stays a 400 px file, so for a sharp banner ask the sender for the source artwork instead of stretching what the PDF stored.

Short answers about saving PDF pictures

Can I pull one photo out of a 200-page catalog?
Yes. Open the file, leave the filters on Larger and Extractable, tick only that card, and press Extract selected images. A single picture downloads as its own file, named after the source and the page where the object first appears.
Will the saved file print as sharply as the PDF page?
It prints exactly as the stored pixels allow. Compare the card pixel size with your target print width; the DPI figure tells you how dense those pixels were on the original page. Nothing in the copy step sharpens or softens the picture.
Why does my ten-page letterhead yield the logo only once?
The PDF stores the logo object once and draws it on every page. The tool writes each unique object a single time, and the Duplicates row of the receipt counts objects drawn more than once.
Do I gain quality by converting everything to PNG?
No. Conversion changes the container, not the detail. Keep original files for identical JPEG bytes, and choose Convert to PNG only when the receiver or the next program insists on that format.
Does the PDF or any picture leave my computer?
No. The scan, the checks, and the ZIP building all happen on your device, so the PDF never uploads anywhere.

Other tools that keep files on your device

Each one below opens in its own page and ends the same way: a named download, a short receipt, and nothing sent anywhere.

Keep working with the pictures