Convert JPG to PDF

The photo that arrives lying on its side

Almost every complaint about this category of tool is the same complaint. You feed in twenty holiday shots, you get back a document where half of them are on their side, and there is no setting anywhere that seems to be responsible. It is not a bug in one particular website. It is a gap between two file formats, and it catches nearly everyone who writes one of these converters without knowing to look for it.

Here is what happens. A phone camera does not turn the sensor when you turn the phone. It always records the same landscape frame, and then writes a tiny note in the file's metadata — the EXIF orientation tag — saying which way up the result should be shown. Your camera roll reads that note. So does the picture viewer on your laptop, and so does every web browser. A PDF has no equivalent note, so if a converter copies your photo across and never checks, it hands the reader a sideways frame with nothing to correct it by. Worse, the page it was measured for was the wrong shape, so the picture is squashed as well as turned.

This page reads the tag, works out the shape the photo should be, and then puts the quarter turn into the drawing instruction that places it on the page. That last part is the trick worth knowing about: the turn costs nothing, because a document already draws every picture through a transformation, and giving that transformation a rotation is free. The alternative — turning the pixels themselves and saving them again — is what most tools do, and it quietly costs you image quality on exactly the photos that needed help.

JPG and PDF, in brief

JPG

Full name	Joint Photographic Experts Group
Extension	.jpg, .jpeg
Format type	Lossy raster image
MIME type	image/jpeg

PDF

Full name	Portable Document Format
Extension	.pdf
Format type	Fixed-layout document
MIME type	application/pdf

What a good JPG to PDF converter leaves alone

A JPEG is already a compressed picture, and a document can hold that compressed data directly. Nothing has to be unpacked, redrawn or squeezed again. That is the single most useful property of this particular conversion, and it is worth protecting: every time a JPEG is saved a second time it loses a little more, and the loss is cumulative and never comes back. Photographs of text and diagrams show it first, as soft fringing around edges that used to be crisp.

So the rule here is that your photo goes in exactly as it already is. There is no quality slider on this page, and its absence is the point rather than an omission — a slider would imply we were re-compressing, and we are not. Files in formats a document cannot carry directly, such as WebP or GIF, are the one exception; those get rebuilt once, and the receipt names them so a silent change never happens behind your back. TIFF is handled separately because one file can contain several compressed image directories: TIFF to PDF inventories and selects those pages before it writes the document.

A pleasant consequence is that the finished size stops being a mystery. Add up the photos, add a few kilobytes of scaffolding for each page, and that is your answer. The estimate appears before you press anything, and the receipt afterwards shows what it actually came to, so you can see how close the arithmetic landed. If the result is larger than you want, the honest fix is to make the photos smaller first, which is a different job with its own page: compress an image to an exact size.

Order is the whole job, not a detail

One picture in, one picture out, is barely a conversion. The reason people reach for this is a stack: receipts for an expense claim, a passport and a boarding pass, the pages of a form photographed one after another because there was no scanner in the house. Get the sequence wrong and the document is useless, however good the pictures are.

The list on the left is therefore the main control, not a preview. Rows carry a handle you can drag with a mouse or press and hold on a touchscreen, and beside every row sit plain arrows and a remove button that work with a single tap. Files arrive in the order your browser hands them over, which for camera files is usually chronological, so most stacks need no rearranging at all. Reordering never re-reads anything from disk, and neither does changing the page size, so experimenting is free.

Mistakes surface later than you would like, though. A stack of twenty receipts often turns out to contain one shot of a thumb, or the same page twice, and that is usually noticed after the document has been built and looked at rather than while the rows are still on screen. Rebuilding from the photographs is one route back; deleting pages from the PDF is the shorter one, since it takes the finished document, puts every sheet on screen as a picture, and lets you drop the offender without disturbing the nineteen that came out right.

Choosing a page, and what the choice means

Letter is the default because most people reading this are printing on Letter. A4 and Legal are there, and so is a fourth option that behaves differently: Fit each photo makes the page exactly the shape of the picture, with no empty margin at the sides of a wide shot. That mode is where resolution starts to matter, because a pixel has no physical size of its own — it only acquires one once you say how tightly the pixels sit. At 300 per inch a four-thousand-pixel photo is a page thirteen inches across; at 96 it is nearly three and a half feet. On a fixed page size the question never arises, which is why the setting only appears in the mode where it changes something.

What this page deliberately does not do

It will not make the text in your photographs searchable. Recognizing characters in a picture is optical character recognition, a genuinely different problem, and a converter that quietly did a mediocre job of it would be worse than one that does not pretend. It is worth knowing that the document you get is image-only for exactly that reason: nothing in it can be selected, searched or quoted, which is the first thing people are surprised by afterwards. How to copy text from a PDF explains which documents have a text layer and which never had one. It will not shrink your photos on the way through, for the reasons above. It will not merge in an existing PDF either — when the photographs need to sit alongside a contract you already hold as a document, build the photo pages here and then merge the two PDFs in a second pass, which is also where you decide whereabouts in the sequence they land. And it will not go the other way: pulling pictures back out of a document needs a renderer instead of a writer, so that lives at PDF to JPG. If your material is screenshots or logos rather than photographs — anything that might have a transparent background — PNG to PDF is the page built for that, with the background handling this one has no use for.

One more thing it will not do: refuse you. A very large stack earns a plain warning about how much memory the job is going to want, and then runs regardless, because it is your hardware and your decision. Photos it cannot identify are named and set aside while the rest continue, so a mixed selection never yields a document with pages silently missing.

Two walkthroughs cover the cases where the question is about the route rather than the tool. Saving a single JPG as a PDF compares what the print dialog on each operating system does to the picture, which is not always what you would expect, and combining several JPGs into one PDF deals with ordering, which is the part that goes wrong at scale. If the photos are still on a phone, converting photos to PDF on an iPhone covers the built-in route that needs no software at all.

Why the photographs exist in the first place

The form asked for one PDF. What you have is a phone, because the phone became the scanner some years ago: a signed lease shot on a kitchen table, a repair receipt, a utility bill photographed to prove an address, four pages of a form filled in with a pen. Benefits renewals, immigration paperwork, insurance claims, school registration, a landlord's move-out packet — the errands that still demand documents are also the ones least likely to reach somebody who owns a flatbed scanner.

A good share of this gets finished on a computer at a public library, because that is where the printer is and where the deadline can still be met. Those machines are restricted on purpose and reset between visitors, so anything arriving as an installer is out of the question, and the pictures come in over webmail or from a phone plugged into a front-panel socket. Assembling the document inside the tab fits that half hour: nothing to set up first, and the only copy made anywhere is the one in that machine's downloads folder, which the reset takes with it when you log off.

Your photos stay on your machine

There is no server involved in this conversion. The code that assembles the document is running in the page you are looking at, working on data your own browser read from your own disk, and the download link points at something that has never existed anywhere else. That is a different promise from a service that deletes your files after an hour, and a stronger one, because a file that was never sent cannot be kept.

It is also a promise you can check rather than believe. Open the developer tools, watch the network activity, and run a conversion: you will see the page fetch its own code and nothing else. If you would like the longer version of how to do that on any site, not just this one, we wrote it up in how to check that a site is not uploading your files.