Convert PNG to PDF

See the page before you make the PDF

Every image-to-PDF site does roughly the same thing: you hand over your files, you press a button, you download something, and only then do you find out that your screenshot came back shrunk into the middle of a Letter page with two inches of white space around it. If that is wrong, there is nothing to correct it against — you change a dropdown, download again, and compare two files in your Downloads folder.

This tool draws the page first. The preview beside the settings is not an approximation: it is rendered from the same layout function that positions the images inside the PDF, using the same numbers. Change the page size, the margin, the orientation or the order and the sheets redraw immediately. What you see is the arithmetic that will run, not a picture of what it might look like.

How PNG to PDF conversion actually works

A PDF does not contain a picture of your image. It contains the image, embedded as a stream of bytes, plus instructions saying where on the page to draw it and how big. For PNG and JPEG that stream can be your original file, unchanged — the format is one a PDF reader already knows how to decode. That is why a PNG that is already a PNG is embedded byte for byte here, with no canvas round-trip and no quality loss.

Formats a PDF cannot carry directly — WebP, GIF, BMP — have to be decoded and re-encoded before they can go in. That happens in your browser, and the receipt says which files it happened to, because a silent re-encode is the kind of thing you only notice much later when you are looking for the original.

Pixels are not points

This is the part that surprises people, and it is worth two sentences of arithmetic. A PDF measures everything in points, and there are exactly 72 points to an inch. An image measures in pixels, and a pixel has no physical size at all until you decide how densely the pixels sit — the resolution.

So a 3000-pixel-wide image is not 3000 points wide. At 300 dpi it is ten inches; at 96 dpi it is over thirty-one. On Letter or A4 the page is fixed and the image is scaled to fit inside your margins, so the question never comes up. InMatch image size mode the page is derived from the pixels, and then the resolution is the only thing that decides how big the page turns out. That is why the setting appears in that mode and nowhere else.

Transparency, and why it comes back black

A PNG can carry an alpha channel, and a PDF can carry that alpha through. What a PDF cannot do is guarantee what sits behind it. Some readers composite onto white, some onto the application background, and a few — including several print paths — end up compositing onto black. A logo that looked fine on screen arrives as a dark rectangle with a shape cut out of it.

So the tool checks. Every image that can carry transparency is sampled after decoding, and if any pixel is actually less than fully opaque the warning appears before you run anything, with a background color you can change. White is the default because it is what people mean by "a page". If you genuinely want the alpha channel preserved, one button keeps it, and the receipt records the choice.

Order, on a phone as well as a desk

With more than one image the order is the whole job, and drag-and-drop is a desktop gesture wearing a mobile disguise. Here the handle at the left of each row responds to a mouse drag and to a press-and-hold on a touchscreen — and next to every row there are plain up, down and remove buttons that need nothing but a tap. Reordering never re-reads the files, so nothing has to be handed in twice.

What this tool will not do

It does not do OCR, so the words inside your screenshot stay pixels — the PDF will not be searchable, because there is no text layer to search. It does not compress the images on the way in; if you need a smaller result, shrink the images first and then convert. And it does not go the other direction: pulling images back out of an existing PDF is a different job, needing a different renderer.

It also refuses things rather than failing at them. A file that is not an image is named and set aside while the rest run. An HEIC straight off an iPhone is identified as an HEIC and turned away with the reason, because PDF cannot carry that format and most browsers cannot decode it either. Anything over eighty megapixels is stopped before it starts, which is a better outcome than a phone tab that dies halfway through with no message at all.

Nothing is uploaded

The PDF is assembled by JavaScript running in this tab. Your browser reads the image bytes from disk, this page writes a PDF around them, and the download link points at a blob that never existed anywhere but in your own memory. There is no upload endpoint, which is a stronger promise than "we delete your files after an hour" — the files were never anywhere to delete.

You do not have to take that on faith. Open your browser's developer tools, switch to the Network tab, and convert something. Nothing leaves.