PDF to JPG / PNG Converter

How a PDF page becomes an image

A PDF page has no pixels — it is measured in points, 72 to the inch, and its text and graphics are vector instructions. To make an image you must pick a density: at D dpi, a page W points wide becomes W × D / 72 pixels wide, rounded to the nearest whole pixel. This tool does exactly that with Mozilla's PDF.js (the renderer built into Firefox, pinned and served from this page): each page is drawn onto a canvas at your chosen dpi and the browser's own encoder produces the JPG or PNG. Nothing is uploaded at any point — the file is read with the local FileReader API and the results are handed back as downloads.

Worked example: A4 at the three resolutions

An A4 page is 595.28 × 841.89 points (210 × 297 mm). Run it through the formula and you get exactly what this tool outputs:

Note the growth: doubling the dpi quadruples the pixel count (300 dpi A4 is 8.7 megapixels — sixteen times the pixels of 72 dpi). That is why the JPGs get heavier fast and why a phone can struggle with a long document at 300 dpi. A rotated or landscape page simply swaps the sides: landscape A4 at 150 dpi is 1754 × 1240.

JPG or PNG — an honest comparison

JPG is lossy. It discards detail your eye barely notices, which makes it the right choice for scanned pages and photo-heavy pages: at quality 0.70–0.85 a scanned A4 at 150 dpi typically lands around 150–400 KB and reads cleanly. Its weakness is sharp edges — at low quality, text grows faint ringing artefacts. PNG is lossless. Text edges stay pixel-perfect and diagrams stay clean, but there is no quality dial to pay with: a dense text page at 300 dpi can easily be 2–5 MB. The quality slider therefore only applies to JPG; when PNG is selected it is dimmed, because PNG never discards anything. Rule of thumb: scans and photos → JPG; screenshots, line art, or anything you will zoom into → PNG.

“Download all” hands you one archive

Once a document has two or more pages the button reads “Download all (.zip)” and packs every image into a single archive: a 30-page PDF becomes one save instead of thirty. Because it is now a single file, your browser no longer asks for permission to download multiple files — that prompt was this button's doing, and it is gone. The names inside the archive are byte-for-byte the ones the per-page buttons produce, zero-padded to the document's page-count width (report-page-03.jpg), so the entries sort in page order the moment you unzip. A one-page PDF skips the archive and saves the image directly; wrapping a single file in a zip would only cost you an extra unzip.

The archive is written here in the tab, by the same small first-party writer three other Polyatic tools use, and it stores the images uncompressed — zip's “store” method rather than deflate. That is a decision, not a shortcut: JPEG and PNG bytes are already compressed, so deflating them a second time typically saves a percent or two, sometimes adds bytes, and costs real time on a long document. Expect the .zip to weigh roughly the sum of the page images plus about a hundred bytes of bookkeeping per entry. Two honest consequences: the archive is not smaller than the images, and nothing streams — every page's bytes and the finished archive sit in memory together, so the same memory ceiling that bounds a long render at 300 dpi bounds the archive too. If a very long document ever fails to pack, the per-page Download buttons are still there and still work.

Honest limits

• A 16.8-megapixel cap per page. Pages are capped at 4096 × 4096 total pixels — the largest canvas iOS Safari reliably encodes. Ordinary A4/Letter pages never reach it (300 dpi A4 is half the cap), but an A0 poster at 300 dpi would be 139 megapixels, so it is scaled down to about 3444 × 4870 with its aspect preserved and marked capped in the result list. Past the cap, picking a higher dpi genuinely gains nothing.
• Text becomes pixels. The images are pictures of the pages — text is no longer selectable or searchable, and there is no OCR. To shrink a PDF that should stay a PDF, use the PDF Compressor.
• Memory bounds the document size. Every page is rendered in your browser's memory; a 20-page report at 300 dpi is fine on a laptop, but a 200-page scan at 300 dpi may be slow or exhaust a phone's RAM. Drop to 150 dpi for long documents.
• Password-protected PDFs fail with a message. A file locked with an open password cannot be read — unlock it in the application that created it first.

Privacy

The file you choose is read locally, rendered locally, encoded locally, and offered back as downloads — it is never transmitted anywhere. No account, no cookies from this tool, no telemetry attached to your file. The two library files that do the rendering — PDF.js and its worker — are pinned open-source releases served from this same site (see vendor/VENDOR.md in the page source), joined by one small first-party script that writes the zip, and the page keeps converting with your network disconnected.