Convert PDF to JPG

What this does

It turns the pages of a PDF into image files: JPG by default, PNG if you prefer, at the resolution you choose. Convert the whole document or only the pages you name. One page downloads as a single image; several arrive together in a zip, named report-page-01.jpg, report-page-02.jpg and so on, padded so they sort in page order in every file manager.

The pages are drawn by Mozilla’s PDF renderer, the same engine Firefox uses to open PDFs, running inside your browser, and encoded on your own processor. Nothing is uploaded, which matters for a tool like this: the PDF you need as a JPG is usually on its way into an upload form somewhere, and the fewer places it has been before that, the better.

How to use it

1. Drop your PDF onto the tool, or press “Choose a PDF”. If it is protected with an open password, you will be asked for it.
2. Pick the format and the resolution. The line above the button answers back before you commit: how many images, and the exact pixel size of the first, such as “Page 1 comes out at 1240 × 1754 pixels”.
3. To convert only some pages, choose “Only these” and list them, such as 1-3, 7. Then press “Convert to JPG” and download.

JPG or PNG: what the numbers say

JPG is the format people ask for, and for photographs it is the right one. For a page of text it often is not, and this is measurable rather than a matter of taste. These are two A4 pages converted by this tool, one a page of report text with a table, one mostly a photograph:

The reason is how each format compresses. PNG is lossless and excels at large flat areas and sharp edges, which is exactly what text on white paper is. JPG is built for the smooth gradients of photographs, and spends bytes blurring the edges of letters that PNG stores exactly. So: text, forms and diagrams make better PNGs; photos and scans make better JPGs; and when a form insists on JPG, JPG is what it gets.

How the JPG setting was chosen

The obvious way to make a JPG in a browser is to ask the page’s canvas for one, which uses the browser’s built-in encoder. This tool uses mozjpeg instead, the same encoder this site’s JPEG quality benchmark measures, and the setting was tested before it was fixed. On eight test pages (report text, photographs, a chart, a form and a table), mozjpeg at quality 88 on the standard quantisation tables matched or beat the browser encoder’s image quality, as measured by SSIM, on every page, while producing files 6 to 18% smaller.

The test also caught a trap worth knowing about. mozjpeg’s own default tables, and its habit of switching off chroma subsampling at quality 90 and above, made photo pages a third larger than the browser’s files for no visible gain. The defaults are tuned for photographs viewed as photographs, not for photographs sitting on a page, and the measurement is the only reason this page does not use them.

Which resolution, in pixels

PDF pages are measured in points, 72 to the inch, so the resolution decides the pixel size directly:

One limit applies above that. Browsers cannot draw arbitrarily large images, and when asked to they tend to fail silently, handing back a blank page. The tightest limit in common use, on iPhones and iPads, is about 16.8 million pixels, so a page that would be larger than that, an A3 poster at 300 DPI say, is drawn at the highest resolution that fits, and the tool tells you when that has happened rather than giving you a white rectangle.

Limitations

Images are pictures of text, not text. A JPG of a page cannot be searched or copied from. If you need the words, keep the PDF.

Links and form fields become part of the picture. Filled-in forms come out looking as they do in a viewer; nothing in the image is clickable or editable.

One PDF at a time. Convert one document, download, and drop the next; to turn several PDFs into one set of images, merge them first.