How to Compress Photos Without Uploading Them

With a photograph, the file is the personal data

A document leaks what it says. A photograph leaks that and more, because the camera wrote things into the file that are not visible in the picture: the coordinates where it was taken, to within a few metres, the date and time to the second, the camera or phone model and often its serial number, and sometimes the name of the owner.

So uploading a photograph to shrink it is not the same kind of act as uploading a spreadsheet. A holiday snap can carry the address of the house you were standing in. That is the argument for compressing where the picture already is.

How it works when nothing is uploaded

The browser decodes the image into pixels, then encodes it again at a lower quality setting, and hands you the new file. Both steps happen in the tab, on your processor. The test is the same as for any tool that claims this: load the page, disconnect from the network, compress. If it still works, the photograph never went anywhere.

Compressing is not the same as stripping metadata

This is the assumption worth breaking. Re-encoding an image often drops its metadata as a side effect, because the encoder writes a fresh file and does not bother to copy the old tags across. Often is not always. Some tools deliberately preserve the camera data, some preserve part of it, and which one you have is rarely stated anywhere.

If the reason you are compressing is that the photograph is about to become public, do not let a side effect do that job. Strip the metadata deliberately with a tool that says it does that, and then check the result rather than assuming: the file properties in your operating system will show you whether a location survived.

Resize first — it does more than compression will

Quality settings fight for percentages; dimensions work in multiples. A twelve-megapixel photograph shown in a four-thousand-pixel-wide frame on a website is carrying three times the pixels anyone will ever see. Halving the width and height quarters the pixel count before a single quality slider is touched, and it is invisible to the viewer because those pixels were never displayed. Resize to the size it will actually be seen at, then compress what is left.

When the result comes back bigger

It happens, and it usually means the file was already efficient. A JPG that has been through a compressor once has little left to give, and a second pass mostly adds artefacts. A PNG screenshot re-encoded as a PNG can grow, because the original was already using the best filter available. And a small image converted from a format that suits it to one that does not will gain weight. A compressor worth using tells you when it cannot beat the original and hands the original back rather than quietly returning something worse.