Turning a PDF into Images When a Site Refuses PDFs

PNG or JPEG is the decision that shows

JPEG achieves its size by throwing away high-frequency detail, on the reasonable assumption that a photograph has more of it than a viewer can perceive. A page of text is the pathological case for that assumption: a letterform is nothing but a high-frequency edge, black meeting white in the space of one pixel. The encoder treats each of those edges as noise to be smoothed and leaves a faint grey ringing around every character — the effect known as mosquito noise. It is subtle at high quality and unmistakable once the file has been squeezed to something a form will accept.

PNG makes the opposite trade and it happens to suit documents. It is lossless, so an edge stays an edge, and it compresses by finding runs of identical pixels — which a page that is nine parts white to one part black provides in abundance. The result is often not just cleaner but smaller. The rule of thumb is about content rather than format preference: if the page is text, tables, forms or line drawings, PNG. If it is a photograph that happens to be inside a PDF, JPEG.

What you give up the moment you render

A PDF built from a word processor holds text as text: the recipient can select it, search it, copy an account number out of it, and any system that indexes documents can read it. Rendering each page to an image ends all of that. What arrives is a picture of a page, and the words in it are shapes. Nobody can search for a reference number, a screen reader has nothing to announce, and a system that expected to extract a total will find none.

That is a real cost, and it is worth being deliberate about paying it. Convert because the destination will not take a PDF, not because images feel simpler. If the reason you are converting is that the PDF is too large, the conversion will not help — an image of a page is almost always heavier than the page it came from, because the compact instruction "draw this glyph here" becomes several thousand pixels.

Resolution, in pixels rather than adjectives

Tools tend to offer low, medium and high, which tells you nothing about whether the result will be readable. The number that matters is dots per inch applied to a fixed page. An A4 page at 150 dpi becomes 1240 by 1754 pixels — taller than most screens, so it displays at full detail without zooming. At 300 dpi it becomes 2480 by 3508, which is four times the pixel count and four times the weight, and buys nothing on screen. It buys a great deal if the page will be printed, because a print at 150 dpi looks soft, and it buys accuracy if a text-recognition system will read the image, because recognition accuracy falls off sharply below 300.

One image per page, and what to do with thirty of them

Rendering produces one file per page, which is obvious in principle and awkward in practice when the form has a single upload field. Three ways out, in the order worth trying: send only the pages that are actually asked for, which is usually one or two and is what extraction is for; check whether the field accepts multiple files, which many do without saying so; or ask whether a PDF is genuinely refused, because a surprising number of forms reject it by mistaken configuration rather than by policy.

This site's converter stops at the first 30 pages, which is a deliberate limit rather than a failure: rendering happens on your own processor, and a hundred-page document at 300 dpi is a hundred images of nearly nine megapixels each, which is enough to exhaust a phone. If you need more than thirty pages as images, split the document first and convert each part — the constraint is memory, and halving the input halves the peak.

The round trip that quietly costs you

A tempting move is to convert a PDF to images, then rebuild a PDF from those images — to flatten a form, say, or to strip something. It works, and it is worth knowing what it costs: the result is a PDF whose text is no longer text, several times heavier than the original, and it can never be undone. Whatever you were trying to achieve, check first whether there is a way to achieve it without passing through pixels, because the round trip is a one-way door.

A decision table you can keep

Two questions settle almost every case. What is on the page — text or photograph — chooses the format. Who reads the result — a screen, a printer or a machine — chooses the resolution. Everything else is a consequence of those two, and getting them right on the first attempt is the difference between one upload and a conversation with a support desk.

Frequently asked questions

Why is my PNG bigger than the whole PDF was?

Because you replaced a description with a picture. In the PDF, a page of text is a short list of instructions naming glyphs and positions, and the fonts are shared across every page. Rendering turns that into two million pixels that have to be stored one way or another. The heavier the resolution, the wider the gap. If size is the problem you were trying to solve, converting to images is the wrong direction entirely.

Can I get the text back out of the images afterwards?

Only by running text recognition over them, which reads the shapes and guesses the characters. It is good and it is not the original: it makes mistakes on unusual fonts, small print, tables and anything handwritten, and it gives you no way to know which words it got wrong. Keep the original PDF. Recognition is what you use when the original is genuinely gone, not a substitute for not having thrown it away.

The form wants one image and my document has six pages. Now what?

Send the page that answers the question, not the document. A form asking for proof of address wants the page with the address on it, and a six-page bill attached in full is a reviewer's chore rather than a stronger case. Extract the page first and convert only that. If several pages are genuinely required and only one file is allowed, ask — that combination is usually a configuration mistake, and support can either raise the limit or accept a different format.

Does converting to images remove hidden data from the PDF?

It removes what was in the document — layers, annotations, form values, the object graph, text under a black rectangle — because rendering keeps only what is visible on the page. That makes it a genuine way to flatten something, and it is the reason people reach for it. It does not remove what the image itself carries: the rendered file gets its own metadata, and if you later photograph or re-edit it, more can accumulate. Flattening and stripping metadata are two separate jobs.

Why does the converter stop at thirty pages?

Because the rendering happens on your device rather than a server, and each page becomes a full-size image held in memory before it can be saved. Thirty pages at a readable resolution is already a substantial amount of memory on a laptop and close to the ceiling on a phone. The limit is there so the tab finishes rather than being killed halfway through. If you need more, split the document and convert the parts — the total work is the same, but the peak is not.