Extract PDF images

A PDF that looks illustrated does not necessarily contain raster images. Charts, logos, maps and diagrams exported from design and office software are frequently vector artwork — paths, fills and strokes drawn by instructions — and there is no bitmap in the file to extract. The empty-state message tells you which case you hit: it distinguishes "no images found" from "images were found but none could be decoded". Formats pdf.js cannot expand, notably JPEG 2000, tiled images and stencil masks, land in that second bucket and are counted as skipped rather than silently dropped. When there is genuinely nothing extractable, the useful fallback is PDF to PNG, which rasterises whole pages instead.

Extracted images come out at the resolution stored in the PDF, which is often much larger than the size they occupy on the page. A photograph placed into a five-centimetre-wide frame may well be a 3000-pixel-wide original; the layout scaled it down for display, but the bytes in the file are untouched, and that is what you get back. It works the other way too — a low-resolution image stretched across a page extracts small and blurry, because that is genuinely all the data there is. The dimensions printed under each thumbnail are the real ones, so you can judge before downloading. Cropping applied in the layout is not baked in: you get the whole source image.

A repeated export means the image is painted on more than one page. PDFs commonly store one copy of a logo or header graphic and reference it from every page, and this tool tracks what it has already saved per page rather than per document, so an image appearing on twelve pages yields twelve identical PNGs. That is deliberate: it keeps the page number attached to each file, which is what you want when pulling figures out of a report. Sorting the downloads by file size in your file manager makes the duplicates obvious and easy to delete in one pass. Scanning also stops after the first 300 images so a heavily illustrated document cannot lock up the tab, and the status line says when that cap was reached.

Alpha channels are preserved: images stored with transparency come back as PNGs with the transparency intact, which is why PNG is the output format here rather than something smaller. One-bit black-and-white images — the bilevel bitmaps typical of fax-style scans and stencil artwork — are expanded into full greyscale pixels so they open normally everywhere. What the tool does not do is colour management: it takes the pixel data pdf.js decodes and writes it out as-is, so an image stored in a CMYK or non-sRGB space can look different from how a PDF viewer renders it in place. For print work, check an extracted file against the page before you use it.

Extract PDF images and PDF to PNG answer different questions, and picking the wrong one is the usual reason people conclude a PDF holds no images. Extract PDF images gives you the source assets that were placed into the document — the individual photos, scans and bitmap graphics, each at its own size, with the surrounding text and layout absent. PDF to PNG gives you a picture of each page exactly as it looks when printed, at a DPI you choose, with the text rendered into the image. If you want the photograph from page 7, use this tool; if you want page 7 itself, use PDF to PNG. A scanned document is the case where the two converge, since every page is one big image, so extracting yields much the same thing as rasterising, minus the choice of resolution.

Your PDF stays on your device and so do the extracted images. Everything runs inside the tab: pdf.js decodes the embedded image data, a canvas re-encodes it to PNG, and the thumbnails you see are object URLs pointing at blobs held in the page's own memory. Those URLs are released when you press Reset or leave the page, so nothing lingers once you are finished. No copy is uploaded, cached on a server, or written to disk except by a download you explicitly click. The tool is free, needs no account and adds no watermark or metadata of its own — the PNGs contain only the pixels that were already sitting inside your document.