Extract the images from a PDF at full quality
Eight brochure photos pulled out at their original size, not screenshots.
A real run on 13 November 2026, on a Linux computer, using Claude Code with Opus, on a Claude plan. The waiting is sped up; it took 1 min 49 s in real time.
The short answer.
Screenshots and Preview's copy give you the picture at screen size. To get the stored original, use a tool that extracts embedded images: the free pdfimages -all command, PDF24's Extract images tool, or LibreOffice Draw (right-click a picture, Save). Stella (stella.sh) pulls every photo out of the PDF on your computer at its full stored size, skips repeats and logos, and names each one after its caption.
Stella does this on Windows, Mac and Linux.
What happened.
8 photos in under 2 minutes. Checked afterwards against an answer key written before the run: each of the 8 photos found at its full original size and matching it, the cut-out picture keeping its transparency, no flood of duplicates or page renders (exactly 8 files), and the brochure unchanged.
The walking club, its brochure and photos are fictitious, drawn for this run. The work is sped up 6×.
How Stella does it.
- Finds the brochure PDF in Downloads and lists every picture stored inside it
- Copies the photos out as their original files, without re-compressing them, at full size (1200×1600 up to 2000×1125 here)
- Keeps the transparent oval cut-out of one photo by saving it as a PNG
- Leaves out the small logo repeated on every page and a second copy of the cover photo
- Names each file by page, position and its caption from the brochure, like “p1-1 Harbour at dawn.jpg”, in a new folder next to the PDF
The prompt.
There's a brochure PDF in my Downloads with photos in it. Get all the photos out as separate picture files at full quality, not screenshots of the pages.
This is the exact prompt from the recording. If your files are somewhere else, say where.
Do it yourself.
Command line: pdfimages
With poppler installed (most Linux systems;
brew install poppleron a Mac; on Windows via the poppler builds or WSL), runpdfimages -list brochure.pdfto see every picture and its real size, thenpdfimages -all brochure.pdf phototo save them in their original format.On Windows: PDF24
The free PDF24 Creator has an Extract images tool that runs offline. Add the PDF and it saves every embedded picture to a folder.
Any computer: LibreOffice Draw
Open the PDF in the free LibreOffice Draw, right-click a picture, choose Save (or Export), and save it as a file. One picture at a time, but at its stored size.
Tidy the results
Extractors save everything: logos, page decorations, masks as separate black-and-white images, and the same photo twice if it's used twice. Sort by size and delete the small ones and repeats.
Avoid screenshots and copy
Copying a picture in Preview or Acrobat, or screenshotting it, gives you what's on screen, often a fraction of the stored size. Use those only if nothing above is available.
Where it stops.
- This gets the pictures stored in the PDF. In a scanned PDF each whole page is one picture, so the photos would have to be cropped out of the page instead.
- Some PDFs store pictures already shrunk; full quality means the size stored in the file, which may be smaller than the original photo.
- Extraction runs on your computer, nothing uploaded. Stella reads the brochure's captions to name the files, so that text goes to the AI model you use.
- Recorded on Linux with free tools. On a Mac or Windows Stella uses the same kind of tool, installing it if needed.
Questions.
How do I extract images from a PDF on a Mac?
Preview can't save embedded pictures at full size. Use pdfimages after brew install poppler, LibreOffice Draw, or Stella.
Why are my extracted images black or inside out?
Some pictures are stored with a separate mask for transparency, or in CMYK colour. Extractors may save the mask as its own black image; pick the other one, or use a tool that combines them (as Stella did for the oval photo).
Can I get every page as a picture instead?
Yes, that's a different job: convert PDF to JPG makes one image per page.
Is it legal to extract images from a PDF?
Extracting is fine for your own files or with permission. Reusing someone else's photos needs their permission, whichever way you get them out.
Is Stella free for this?
Yes. Free with the ChatGPT or Claude plan you already have. Sign in with your ChatGPT plan, or use your Claude plan through Claude Code, and this task costs nothing extra. New accounts also get a small free trial of Stella’s own models.
Can I control what Stella changes on my computer?
Yes. Stella works on Windows, Mac and Linux with your own files and shows each step in the chat, so you can watch it and stop it at any point. Tell it what it may not do, as the prompts on these pages do (“don't delete anything without asking”). In the Downloads recording it deleted nothing until it was told to go ahead. It also refuses commands that would wipe a whole disk, your home folder or system folders.
Checked against these sources on 13 November 2026.