The thing this tool actually does (and the thing it doesn't)
The PDF Image Extractor pulls every embedded raster image out of a PDF and gives you back separate PNG files — one per image, at the original resolution it was stored at inside the PDF. Photos, screenshots, logos, scanned figures, embedded diagrams: each comes out as its own file you can save, edit, drop into a doc, or use anywhere else.
This is not the same thing as PDF to PNG, and the distinction matters. PDF to PNG renders every page as one image — a 10-page PDF gives you 10 PNGs, each one looking like a snapshot of a page. This tool instead reaches into the PDF's underlying structure and pulls out the individual images that were embedded when the PDF was created. A 10-page PDF with three photos per page gives you ~30 PNGs, each containing one original image.
Concretely: if you have a vacation album someone exported as PDF, and you want the source photos back — you want this tool. If you want screenshots of each page as it looks — you want PDF to PNG.
The whole extraction happens in your browser. PDF parsing uses pdfjs-dist (Mozilla's PDF.js, the same library Firefox uses to display PDFs internally). Image decoding and PNG encoding use the browser's native canvas. There's no upload, no server, no queue, no daily limit. Open your network tab while you extract — zero outbound requests. That's verifiable in 5 seconds, not a promise you have to trust.
How extraction works under the hood
A PDF page isn't a flat image. It's a list of drawing instructions — "fill this rectangle, draw this line, render this glyph from this font at this position, paint this raster image at this transform." We use pdfjs-dist to parse each page and walk its operator list. Every time we hit a "paint image" operator, we read the image's data, decode it, and re-encode it as a PNG.
The output filenames identify each one: contract.pdf with images on pages 2 and 7 might produce contract-p2-img001.png, contract-p2-img002.png, contract-p7-img001.png. The page number and image-on-page number are right there, so you can trace any extracted image back to where it came from.
Each image comes out at the resolution it was stored at inside the PDF. If the original was a 4032×3024 photo and the PDF stored it at that resolution, you get a 4032×3024 PNG. If the author downsampled the image to 1920×1440 during PDF creation (a common compression trick), you get 1920×1440 — that's the actual data in the PDF, even if the original source had been higher-res.
How to use it
- Drop or pick your PDF. Up to 100 MB, 200 pages, 500 images per document.
- Click "Extract images." The tool walks each page in order, finds every embedded raster image, decodes it, and stages it as a PNG. Progress shows per page.
- Extracted images appear in a grid. Each one shows its source page number, file size, and resolution, with a download button on the tile.
- Download a single image you want, or hit "Download all" to grab everything in order. There's no zip step — files come down individually because that plays best with most browsers' download managers.
If the PDF has no embedded raster images — say it's a text-only document, or all the visuals are vector graphics (more on that below) — the result panel says so honestly: "No raster images found in this PDF." We don't fake it.
What extracts and what doesn't
| Content type | Extracts as PNG? | Why |
|---|---|---|
| Embedded photos (JPEG) | Yes | Raster image stored in the PDF; decoded and re-encoded as PNG |
| Embedded screenshots (PNG) | Yes | Same — pdfjs-dist returns the decoded pixels, we save as PNG |
| Scanned page images | Yes | A scan is a raster image, even if the page looks like text |
| Logos saved as raster (PNG/JPG before placement) | Yes | Comes out at whatever resolution the PDF stored |
| Logos placed as vector (SVG/EPS) | No | Drawing instructions, not pixels — render the page instead |
| Charts and diagrams drawn as PDF vector ops | No | Same — these are lines and curves the PDF draws on demand |
| Body text | No | Text is rendered from fonts at display time, not stored as pixels |
| JBIG2-compressed scans | Mostly yes | pdfjs-dist decodes JBIG2; rare malformed cases skip |
| Inline images (small icons embedded inline in the operator stream) | Yes | Handled the same as XObject images |
| Image masks (transparency layers) | The image; mask is applied at extract | You get a PNG with alpha when the source had a soft mask |
The single sharpest "no" on that list is vector graphics. If your PDF has a beautiful pie chart, a line graph, a CAD-style diagram, or a logo drawn with vector instructions, this tool cannot extract those as images — because they aren't images. They're drawing operations the PDF executes when displayed. There's nothing to extract; there's only something to render. For those cases, the right tool is PDF to PNG: render the whole page and crop the part you want.
A worked example: a 12-page travel magazine
Imagine a 12-page travel magazine PDF, designed in InDesign. Each page has 2–4 photos and some body text. Drop it in.
The extract walks all 12 pages. It finds 34 embedded raster images: a mix of full-page hero photos (3000×2000-ish), in-line shots (1600×1200), and small ornamental dingbats (200×150). Total extract time: about 6 seconds on a modern laptop. Total output: 34 PNGs totaling around 180 MB on disk, because PNG is lossless — the source was compressed JPEG inside the PDF, but the output is raw pixel data.
That last point is worth knowing. If you want the extracted photos as JPEGs (smaller files), you can either run the PNGs through an image format converter after, or — better, if you already know the originals were JPEGs — accept the larger PNGs and convert the few you actually need. There's no point JPEG-encoding 34 images when you'll only use 5.
What you don't get from this magazine: the page titles, body text, page numbers, or any of the layout. Those weren't raster images in the source — they were text and vector elements rendered at display time. To get the magazine pages as they look (text and all), use PDF to PNG instead.
Why use this over PDF to PNG, or over iLovePDF / SmallPDF / Adobe
vs. PDF to PNG: These are different tools for different jobs. PDF to PNG gives you screenshots of pages — useful for sharing what a page looks like, for thumbnails, for embedding a page preview in a doc. This tool gives you the source photos back, at the resolution they were stored at, without the surrounding page layout. If you want to recover the original images from a PDF someone made, you want this tool. If you want to see the page as a picture, you want PDF to PNG.
vs. iLovePDF and SmallPDF: Both have image-extraction tools, both upload your PDF, both impose page caps, both push you toward an account or upgrade. For any PDF you'd rather not share with someone else's servers — internal documents, photo albums, sensitive scans — that's the wrong shape. We have no upload, no cap below 100 MB / 200 pages / 500 images, no account.
vs. Adobe Acrobat Pro: Acrobat does this well, locally. If you already pay $20/month for Acrobat Pro, use it for serious PDF work. If not, paying for the whole Acrobat suite to get image extraction is the kind of overbought subscription this tool is trying to make unnecessary.
vs. ABBYY FineReader: FineReader is a serious tool, mostly for OCR. Its image extraction works but the workflow is heavyweight and the license is hundreds of dollars. Same overbought story.
vs. "render the page and crop the image manually": That works for one or two images. It doesn't scale, and it gives you whatever resolution your screen rendered at — not the original resolution that was actually stored. The whole point of this tool is recovering originals at full fidelity.
Edge cases worth knowing
Downsampled photos. Many PDF creation tools (Word, InDesign, Acrobat Distiller) downsample large photos during PDF export to keep file size reasonable. If the PDF only needed to display a photo at letter-paper size, it might store the photo at 1200×900 even if the original was 4032×3024. We extract what's in the PDF, which is the downsampled version. The original-original isn't recoverable from the PDF — it was discarded at PDF creation time.
Image masks and transparency. Some PDFs use soft masks to give photos alpha channels (rounded corners, fades). We apply the mask and output a PNG with alpha. The result is what you'd see on the page, not the raw rectangular source.
Color profiles. Photos in PDFs sometimes carry color profiles (sRGB, Adobe RGB, CMYK). We extract pixel data and write PNG in sRGB. For color-critical work (print prepress), the color shift can matter; for general use, it doesn't.
The same image embedded multiple times. If a logo appears on every page, PDF authors often embed it once and reference it from each page. We dedupe — you get one copy of the logo, not 12 — and label it with the first page it appeared on.
Related PDF and image tools
- PDF to PNG — the page-level version of this idea. Use it when you want each page as a picture, not the source images.
- PDF to JPG — same as PDF to PNG but with JPEG output (smaller files, lossy).
- Extract Text from PDF — for the words. Not OCR; works on text-bearing PDFs.
- PDF Page Counter — drop a PDF, see how many pages plus the full metadata block (title, author, creator app). No transform.
- Split PDF — for very large PDFs, split into chunks before extracting to keep memory usage down.
- Image Compressor — if the extracted PNGs are larger than you need, recompress them with quality controls.
- Image Format Converter — convert the extracted PNGs to JPG, WebP, or AVIF.
Frequently asked questions
How is this different from PDF to PNG?
PDF to PNG renders each PDF page as one image — an N-page PDF gives you N PNGs, each looking like a page. This tool extracts every embedded raster image (photos, logos, screenshots, scanned figures) out of the PDF's underlying structure. A 10-page PDF with three photos per page gives you ~30 PNGs, each containing one original image at the resolution it was stored at. Use PDF to PNG for page-level snapshots; use this tool to recover source images.
What kinds of content will extract?
Raster images (bitmaps): photos, screenshots, scanned content, logos that were saved as PNG or JPG before being placed in the PDF. The resolution the PDF stored is preserved — a 4032×3024 photo embedded in the PDF comes out as a 4032×3024 PNG, even if the page only displayed it at 800×600.
What won't extract?
Vector graphics. If the PDF contains SVG-like content drawn by the PDF itself — body text rendered from fonts, charts drawn as lines and curves, geometric diagrams, vector logos — there's no raster source to extract. The PDF stores those as drawing instructions, not pixels. To get them as images, render the whole page with PDF to PNG and crop the part you need.
Why are some extracted images smaller than I expected?
PDF authors often downsample images during PDF creation to keep the file size reasonable. A photo that was 4032×3024 in the source might be stored at 1920×1440 in the PDF if the document only needed to display it at a smaller size. We extract whatever resolution is in the PDF — that's the actual source data, not the pre-PDF original (which is no longer recoverable from the file).
Is my PDF really not uploaded?
Correct. PDF parsing uses pdfjs-dist running in your browser; image decoding and PNG encoding use the browser's native canvas. Zero outbound requests during extract. Open your browser's network tab and check while you run the tool.
Does this work for password-protected PDFs?
pdfjs-dist can attempt to open encrypted PDFs but won't decode page content without the password — and we need page content to walk the operator list and find images. Unlock the PDF first using a desktop reader (Adobe Acrobat: File → Properties → Security; macOS Preview: File → Export, uncheck Encrypt) and run the unlocked copy through this tool.
Is there a limit?
200 pages, 500 images, 100 MB per PDF. Past that the browser starts to struggle with memory because every extracted image lives in memory until you download it. For larger PDFs, process in chunks with Split PDF first.
Why PNG and not the original format the image was stored in?
Because PDFs internally use several image formats (JPEG, Flate-compressed bitmaps, JBIG2, JPEG2000), and decoding everything to PNG gives a single consistent output format every tool can read. PNG is also lossless — so the output is as good as the source data we decoded, with no second-generation compression artifacts. If you want JPEGs for smaller files, run the PNGs through an image format converter.