JPG to PDF

1. Choose the images on your device. Drop the JPG, PNG, WebP or HEIC files onto the panel above, or press Choose images. Up to 20 at a time, 20 MB each, and the files stay where they are — there is no transfer step.
2. Put them in page order. The number beside each row is the page it will become. Use the Move up and Move down buttons on a row to shift it, and the remove button to drop one out of the set.
3. Set the page size. Leave it on Fit each page to its photo for edge-to-edge pages, or pick A4 or US Letter and a margin when the document has to be printed or a form demands a paper size.
4. Press Make PDF. Your device reads the photos one at a time and writes each of them into the document. If you set a maximum size, the pictures are re-drawn smaller until the finished file fits it.
5. Download the PDF. The panel prints the page count and what the document really weighs. Download it, open it to check the order, and run it again with a different order if you need to.

Most JPG-to-PDF converters take a copy of your pictures first and build the document on a machine somewhere else. That is a reasonable way to write software and a poor fit for what this tool is actually used for: passport scans, birth certificates, payslips, bank statements and medical letters.

Here the PDF writer is a script that arrives with the page and runs in this browser tab. Your device opens the photos, assembles the document and hands you the finished file, so there is no copy of your paperwork anywhere but on your own computer. The only thing that ever travels is the code, in one direction, on its way to you.

A PDF does not store pixels. It stores streams, and a JPG is already a stream of exactly the kind a page wants, so the picture is copied straight in — the reader opens the same bytes your camera wrote. Nothing is decoded, so twenty photos cost this tool about as much memory as one.

A PNG, a WebP and an iPhone HEIC get no such shortcut. Each has to be opened into raw pixels and written back out as a JPG before it can be a page. Packing a PNG in unchanged is possible in theory and worse in practice: it costs the memory of a full decode and usually produces a document larger than the sum of the pictures that went into it. The panel counts these files and tells you before you start, and the result reports how many were actually re-drawn.

There is one exception on the cheap side. A photo taken sideways carries a tag telling a viewer to rotate it, and no PDF reader reads that tag, so a tagged photo is turned in the pixels first. It is the difference between a portrait page and a page lying on its side.

Forms and email systems put a ceiling on the file, not on the photos. So the maximum applies to the finished PDF: the budget is divided across the images in proportion to what each one already weighs, and whatever a page does not spend is handed to the next one rather than thrown away.

The verdict is then measured on the document that actually came out, never guessed from the plan. If the number cannot be reached you still get the PDF, with a line saying how close it got and what the smallest achievable size is, and the pictures keep their full pixel dimensions — a document that quietly shrank your scans to hit a number would be unreadable at exactly the moment it mattered.

• Visa and passport applications, which want every scanned document as a single PDF.
• Job applications and university forms that accept one attachment and no more.
• Expense claims, where a set of receipt photos has to go in as one file.
• Bank and insurance uploads that reject an image but accept a document.
• Sending a phone-camera scan of a signed contract back to whoever sent it.
• Keeping a set of photographs in a fixed order that a folder of files cannot hold.

Up to 20 images in one PDF, 20 MB per image and 80 MB across the set. JPG, PNG, WebP and HEIC all go in; the output is always a PDF, one page per image, in your order.

Past those numbers the binding limit is the device in your hand, not a policy. A photo has to be unpacked into raw pixels before it can be re-drawn, which takes several times the file size in memory, so the panel costs the whole document before it starts and refuses in plain words if this phone or laptop cannot hold it — with the suggestion to build it from fewer images at a time. It is the honest answer: there is nowhere else for the work to go.