How to put photos into a PDF without making them worse

PDF can carry a JPEG untouched

A PDF is a container. It has carried JPEG natively for decades, which means a JPEG can be written into a page as the exact bytes that came off the camera, not decoded, not re-compressed, not touched. The same is true of PNG.

This matters because photo-to-PDF tools routinely do not do it. The easy implementation draws each picture onto a canvas and encodes the canvas as JPEG at whatever quality the author picked, usually 0.92. Every image goes through a lossy round trip it did not need, including the ones that were already JPEG and could have been copied verbatim.

The damage is small per image and cumulative across a document, and there is never any indication it happened.

What can be embedded as-is

JPEG and PNG go in directly. Nothing is decoded and nothing is lost.

Everything else (WebP, HEIC, AVIF, GIF, BMP, TIFF) has no native PDF representation and genuinely has to be redrawn. The question is what it is redrawn to, and the honest answer is PNG: lossless, larger, and no worse than the source. Redrawing to JPEG makes a smaller file by throwing away information, which is a legitimate choice but has to be a choice, not a default nobody mentioned.

Ours embeds JPEG and PNG byte-for-byte, redraws everything else to PNG, and says which path each picture took before anything is written.

Why your file is enormous

A PDF of ten phone photos is roughly the size of ten phone photos, because that is what is inside it. Around 40 MB for ten, more for a recent phone.

PDF does not compress images beyond how they already are. If the file is too big the options are to use fewer pixels or accept lossy compression: there is no third setting that makes the same pixels smaller for free.

The useful move is usually to resize before converting. A photo printed at A4 needs about 2500 pixels on its long edge for 300 DPI. A 4032-pixel phone photo is already beyond that, and anything above it is bytes that no printer and no screen will ever resolve.

Page size and orientation

Two decisions that look small and produce most of the complaints:

Page size. A4 and Letter are nearly the same and not the same: A4 is taller and narrower. A document built for one and printed on the other gets scaled or clipped. Pick the one your printer actually has.

Orientation. A landscape photo on a portrait page leaves broad white bands top and bottom, and uses about half the paper. Letting each page take the orientation of its own image is almost always what people want and almost never the default.

There is also the option of a page exactly the size of the image, which gives no margins and no whitespace. Good for a photo book, bad for anything that will be printed on ordinary paper.

What gets lost that is not the pixels

Embedding a photo in a PDF carries the compressed image data, not the file around it:

• EXIF goes. Camera, lens, exposure, timestamps, and any GPS coordinates.

Usually the point rather than a loss: a PDF of photographs headed to somebody else is exactly where you do not want the location of your house.

• Colour profiles may not survive. A photo in a wide-gamut space can shift

slightly if the profile is dropped. Visible mainly in saturated reds and deep blues.

• Rotation is the one to watch. Phones often store orientation as an EXIF

flag rather than rotating the pixels. Lose the flag and the photo lands sideways. A tool that reads the orientation and applies it before embedding avoids this, if yours does not, the fix is to rotate and re-save the originals first.

Scans and documents, not photographs

If the images are pictures of pages, receipts, a signed contract, a passport the priorities invert. Legibility beats file size, so keep the resolution high and prefer lossless. JPEG artefacts around small text are the difference between a readable scan and an argument.

And these are precisely the files worth not uploading to a stranger's server in order to staple them together.