Image to PDF

The document is built here, not somewhere else

This one is worth being precise about, because “convert images to PDF” is normally a service you post your photographs to. Here the PDF is written byte by byte by JavaScript in your own tab. Your images are decoded, drawn, encoded and assembled without anything being sent anywhere, and the finished document exists only in memory until you save it.

Not even a library is involved. A PDF containing images is a small, well-specified structure, a catalogue, a page list, and one image and one instruction per page, so it is written directly. That keeps the page small, and it means there is nothing in it that could send anything anywhere.

Nothing identifying is written into the file

Most PDF writers add an author, a producer string naming the software, and a creation timestamp. None of those are added here. The timestamp in particular travels with the document to everyone it is ever forwarded to, and a tool whose point is that your files stay on your machine should not be stamping them on the way out.

The images lose their own metadata too, EXIF, camera and lens, and any GPS coordinates, because they are re-encoded on the way in. That is a side effect rather than a feature, and a useful one.

Why the photos are re-encoded

A PDF can carry a JPEG untouched, so passing one straight through would be lossless and instant. It is not done, for a reason you would notice immediately: rotation.

A phone stores a photograph in whatever orientation the sensor read it, along with a tag saying which way up it goes. PDF has no equivalent tag, so an embedded photo appears on its side. Drawing it onto a canvas first applies the rotation permanently, which is also the only path that handles PNG, WebP and everything else. One path, got right once.

Page size, and when it matters

A4 is the standard almost everywhere outside North America; US Letter inside it. Pick one of those if the document might be printed, attached to a form, or sent to somebody who will do either.

Match each image makes every page exactly the shape of its picture with no border at all. It is the right answer for screenshots, receipts and anything only ever read on a screen, and an awkward one to print. A printer will scale each page to fit its own paper, and the pages will not agree with one another.

On a fixed sheet the page turns for a landscape image rather than shrinking it into a portrait one between two empty bands. Same paper, rotated.

Quality, and why it is not a slider

Three named settings rather than a number, because the decision being made is about purpose rather than about a value, and a slider invites fiddling with something you can only judge by rebuilding the whole document.

• Smaller file caps each image at 1,600 pixels. For email, forms and anything with an attachment limit.
• Balanced caps at 2,400. Sharp on any screen and fine for ordinary printing.
• Best quality keeps essentially the full resolution and produces much larger files.

For scale: ten twelve-megapixel photographs embedded untouched come to well over a hundred megabytes, which most mail servers will refuse.

Reordering is buttons, deliberately

Dragging is the obvious gesture and a poor one to offer as the only one: there is no keyboard equivalent, a screen reader has nothing to announce, and on a phone it fights with the scroll. Two arrows work everywhere and cost one extra tap on the rare occasions anyone reorders at all.

Dropping files onto the page to add them is supported, because there the drag has a keyboard equivalent sitting right next to it.