Pages—
Resolution150 dpi
Pixel size—
Total—
Converting a PDF to images is a choice about resolution, and almost every tool hides it behind a word — normal, high, best — that tells you nothing about what you will get. A resolution is two numbers: the pixels each page comes out at, and what those pixels weigh. This tool renders every page in the browser and prints both before you download anything, so the difference between 96 and 300 dpi is a measurement rather than an adjective. On the seven-page sample here that is 794 × 1123 pixels and 214 KB against 2480 × 3508 and 1.2 MB.
Step by step
01Open the document
Drop it anywhere on the page. Every page starts rendering immediately and the grid fills in as each one lands, so a long document shows progress rather than a spinner.
02Pick the resolution for where it is going
Screen is 96 dpi and matches a monitor 1:1. Office print is 150. Print and OCR is 300, which is what commercial printing needs and what text-recognition engines are tuned for.
03Pick the format against its real weight
JPG, PNG, and WebP all render the same pixels and store them very differently. The total is measured on the encoded images, so the comparison is a number rather than a rule of thumb.
04Take one page or all of them
Click any page in the grid to download just that image; the header button zips the set. Both are instant, because the images already exist.
A worked export
The seven-page A4 sample this tool opens with, rendered at each preset and encoded in each format. Every figure below is what the tool prints on it.
Given
Resolution, then format at 300 dpi
96 dpi 794 × 1123 px 214 KB total largest 33 KB
150 dpi 1240 × 1754 px 446 KB total largest 69 KB
300 dpi 2480 × 3508 px 1.2 MB total largest 181 KB
at 300 dpi:
JPG q90 1.2 MB
PNG (lossless) 1.9 MB +58%
WebP q90 514 KB −57%Resolution is quadratic and file size follows it: tripling the dpi multiplies the pixel count by nine and, on this document, the bytes by about 5.6. That is the argument against reaching for the highest setting by default — 300 dpi is right for print and for OCR, and wasteful for anything that will be looked at on a screen at 96. The format choice is the larger surprise: WebP stores the same pages at well under half what JPEG needs, and PNG at nearly twice, because PNG is lossless and a rendered page is full of the smooth gradients lossless compression is worst at. WebP is read by every current browser; JPG is the safe answer when the images are going into a system you do not control.
Before you convert
Resolution is quadratic
Doubling the dpi doubles both dimensions, so it quadruples the pixel count — a 300 dpi A4 page is 8.7 megapixels against 2.2 at 150. File size does not rise quite as fast, because compression benefits from the extra redundancy, but it rises fast enough that 'just use the highest' turns a 200 KB export into a multi-megabyte one.
96 for screens, 150 for an office printer, 300 for print and OCR
Those three numbers cover almost every real use. 96 is roughly a monitor's own density, so the image is 1:1 with the page on screen. 150 is where an office laser stops showing softness. 300 is the commercial print standard, and also what OCR engines expect — feeding text recognition a 96 dpi render is the usual reason it performs badly.
Rasterising destroys everything except the picture
The output has no selectable text, no working links, no bookmarks, and no accessibility structure. When the point is to stop someone copying the text that is the feature; when the point is to share the document it is a real loss, and worth being deliberate about.
PNG is the wrong default for a rendered page
PNG is lossless, which sounds like the safe choice and costs heavily on a page of anti-aliased type and smooth fills — on the sample here it is 58% larger than JPEG at the same resolution. Reserve it for pages that are line art or screenshots of flat colour, where lossless is both cheaper and visibly better.
There is a hard ceiling, and it fails silently
Browsers cap a canvas at roughly 268 million pixels and hand back a blank one past it rather than an error, which looks exactly like an empty document. The resolution slider therefore stops at whatever keeps the longest page edge under 8,000 pixels for the document you loaded, and says so.
OCR wants resolution more than it wants quality
Text recognition works on the shape of a glyph, so a 300 dpi render at JPEG quality 80 is far more useful than a 96 dpi render at quality 100. If the images are going into an OCR pipeline, raise the dpi first and leave the quality slider alone.
The judgement call
Where the images are going decides both numbers.
Posting a page in a chat or a ticket
96 dpi, JPG
It matches a monitor 1:1, and on the sample here the whole document comes to 214 KB rather than 1.2 MB.
Feeding pages to OCR
300 dpi, JPG
Recognition accuracy tracks resolution far more than it tracks compression quality, and 300 is what the engines are tuned for.
Sending images to a printer
300 dpi, PNG or JPG q95
Print needs the resolution, and at that point the compression choice is about whether the printer's workflow prefers lossless.
Images for a web page you control
150 dpi, WebP
Every current browser reads WebP, and on the sample it stores the same pages at well under half what JPEG needs.
A page that is a diagram or flat-colour screenshot
PNG
Lossless is both smaller and sharper on hard edges and flat fills — the case PNG was designed for and the one where JPEG's ringing shows.
You want the recipient to be able to copy the text
Do not rasterise
Images have no text layer. Send the PDF, or extract the pages you need as a PDF instead.
Reference
FAQ
96 dpi for something that will only be viewed on a screen, 150 for an office printer, and 300 for commercial print or OCR. On an A4 page those come out at 794 × 1123, 1240 × 1754, and 2480 × 3508 pixels — the Tinapps PDF to JPG tool shows the exact pixel size and the real file size at whatever resolution you pick, so the choice is a measurement rather than a guess.
Because resolution is quadratic: doubling the dpi quadruples the pixel count. Rendering a seven-page document at 300 dpi instead of 96 takes it from around 214 KB to 1.2 MB. Dropping to the resolution the images will actually be viewed at is almost always the fix, and switching from PNG to JPEG or WebP is the second one.
JPG when the images go into a system you do not control, because everything opens it. WebP when you do — it stores the same rendered pages at well under half the size of JPEG at matching quality. PNG only for pages that are line art or flat-colour screenshots; on a page of type it is lossless in the expensive sense, measuring 58% larger than JPEG on the sample in the Tinapps PDF to JPG tool.
Yes. Rasterising a page turns it into pixels, so the output has no selectable text, no working links, no bookmarks, and nothing for a screen reader. That is exactly the point when you want to stop someone copying the content, and a real loss when you are simply trying to share the document.
300 dpi. Text recognition works on the shape of each glyph, so resolution matters far more to it than compression quality does — a 300 dpi render at JPEG quality 80 will be recognised more accurately than a 96 dpi render at quality 100. Rendering too low is the usual reason OCR performs badly on an otherwise clean document.
Yes. The Tinapps PDF to JPG tool renders each page with pdf.js and encodes it with the browser's own canvas, so the document never leaves the device and there is no queue between choosing a resolution and seeing the result. It also keeps working with the network disconnected.
Every page is rendered anyway, so click the page you want in the grid and it downloads on its own — the header button is only there for taking the whole set as a zip. Each tile shows its file size, so you can see what you are about to save before you save it.
Yes, and it comes from the browser rather than the tool: a canvas larger than roughly 268 million pixels silently returns blank, which looks like an empty document. The Tinapps PDF to JPG tool caps the resolution slider at whatever keeps the longest page edge under 8,000 pixels for the document you loaded, and says when you have reached it.
Pages are rendered and encoded entirely in your browser. The document is never uploaded, and the images exist only as the files you download.
Keep going
JPG to PDF
The other direction, with the print resolution reported per image.
Image compressor
Take the exported pages further down with a quality ladder.
Format converter
Compare PNG, JPEG, and WebP on a single image at once.
Extract pages
When the pages should stay a PDF rather than becoming pictures.