JPG to PDF
Built in your browser · nothing is uploaded
A PDF is the right wrapper when several images have to travel as one document; a set of receipts, a passport photographed front and back, a signed page you photographed rather than scanned. Each image becomes a page, in the order you set.
How to turn JPGs into a PDF
JPEGs are embedded into the PDF byte for byte rather than being decoded and re-saved, which matters more than it sounds: a photo goes in at exactly the quality it arrived with, and the PDF ends up roughly the size of the photos added together. Anything that is not a JPEG is a different path. A PNG screenshot or a WebP is decoded to raw pixels and stored losslessly, so its transparency and its hard edges survive intact, and it takes up a great deal more room: three bytes a pixel before compression, which for a twelve-megapixel image is thirty-six megabytes going in. Lossless compression claws some of that back and never as much as the original PNG managed.
The page sizes are the real ones, in the points a PDF is measured in, at 72 to the inch: A4 is 595.28 by 841.89, Letter 612 by 792, Legal 612 by 1008, A3 841.89 by 1190.55 and A5 419.53 by 595.28. The margin slider runs from zero to 72 points, so its top setting is exactly one inch and its 24-point default is a little under nine millimetres. Auto orientation turns the page landscape whenever an image is wider than it is tall, so a mixed batch comes out with each photo the right way round rather than all of them letterboxed to portrait.
Nothing is ever scaled down on the way in. An image is placed at whatever size fits the page, but the full pixel data goes into the file, so the print resolution is far higher than any printer will use. Orientation sets the exact figure, and Auto, the default, turns the page to match the photo: a 4000 by 3000 photo on A4 with the default margin therefore gets a landscape page and prints about 10.1 by 7.6 inches, roughly 395 pixels to the inch. Force the same photo onto a portrait page and it shrinks to 7.6 by 5.7 inches and 526 ppi. Fit the image works the other way and is the one setting that produces an odd document. It builds each page from the image at 96 pixels to the inch, so that same 4000-pixel photo becomes a page over a metre wide. Nothing renders wrong, and the PDF format allows pages up to 200 inches square, but it is a shape for reading on screen rather than for printing.
If the result is too big to email, the margin and the page size will not help, because neither changes how many pixels are stored. Resize the photos first, or run the finished document through the PDF compressor.
What people use it for
- Sending photographed receipts as one document
- Combining a passport photographed front and back
- Turning a signed page photographed on a phone into something emailable
Questions
No. JPEGs are copied into the PDF untouched, so what you see is exactly the original file.
Because the photos are. A modern phone photo is three to six megabytes, and ten of them make a thirty-megabyte PDF. Compress it afterwards, or resize the photos first.
Yes. Each is embedded in whichever way preserves it best: a JPEG byte for byte, anything else as lossless pixels.
Because it is stored as plain pixels with general-purpose compression rather than with PNG’s own filters, and photographic content barely compresses that way. Save a photographic screenshot as a JPEG first if size matters more than sharp edges.
Yes. The alpha channel is carried across as a soft mask, so a logo with a transparent background stays transparent on the page.
It makes every page exactly the proportions of its photo, so nothing is letterboxed. Useful for screenshots, less so for anything you will print.
The image at 96 pixels to the inch, plus the margin. A 4000-pixel-wide photo becomes a page about 42 inches across, which is fine on screen and absurd on paper.
The standard ones: A4 is 210 by 297, A5 148 by 210, A3 297 by 420, Letter 216 by 279 and Legal 216 by 356. They are stored in points, at 72 to the inch.
Points. The slider runs from 0 to 72 in steps of four, so the maximum is exactly one inch and the 24-point default is just under 9 mm.
Per image, by comparing width against height. Anything wider than it is tall gets a landscape page; everything else stays portrait. Force one or the other if you want a consistent document.
Only visually. The full pixel data is stored regardless of the page size, so a small page does not make a small file.
Usually far more than you need. A 4000 by 3000 photo on A4 with the default margin and the default Auto orientation gets a landscape page, printing about 10.1 by 7.6 inches at roughly 395 pixels to the inch. Forced onto a portrait page the same photo lands at 526. A good print uses 300.
Use the arrows next to each thumbnail. The top one becomes page one.
No fixed one. Everything is decoded in your browser, so the practical limit is your device’s memory. A few dozen phone photos is unremarkable; several hundred may struggle on an older phone.
They are probably HEIC, which most browsers cannot decode. Either set the camera to Most Compatible so it saves JPEGs, or export the photos as JPEG before adding them.
Yes. A JPEG says how many components it carries in its own header, and the builder reads that rather than assuming three: a one-component greyscale scan is embedded as greyscale, a four-component one as CMYK. Getting this wrong is why a scan can arrive as a grey smear in other tools, since declaring one byte per pixel as three does not fail, it just renders nonsense.
Because the EXIF Orientation tag does not travel. A phone often stores a portrait photo as a landscape image with a tag telling viewers to turn it, and since the JPEG is embedded exactly as stored, that tag is neither read nor carried. A PDF page does have a rotation entry of its own — /Rotate, in clause 7.7.3.3 of ISO 32000-1, an integer multiple of 90 — but this builder never writes one, so nothing turns the page either. Rotate and re-save the photo before adding it.
Only the first frame. A PDF page holds a still image, so an animation is reduced to whatever it started on.
No. A photo of a page is a picture, and nothing here reads it. You need optical character recognition to make it searchable, which is a different job.
Not here. Build the second batch as its own PDF and join the two with the merge tool.
A single photo keeps its own name with a .pdf ending. Two or more come out as images.pdf, which you can rename afterwards.
Yes. The output is an ordinary PDF with no unusual features, so browsers, phones and every reader handle it.
Shrink the photos, not the page. Resizing a 12-megapixel photo to about 2000 pixels on its long edge is still sharp on paper and cuts the file by most of its size, or run the finished PDF through the compressor.
No. The PDF is written in this page by code that ships with it, and works with the network disconnected.