Find out what is wrong with your image, then fix it

No. The file is read into memory in your browser and parsed there. There is no upload endpoint, and nothing on the page that can read your content, and a Content Security Policy allows no upload destination at all.

It walks the EXIF structure inside the file and looks for the GPS pointer tag. That is a direct reading of your file’s own bytes rather than a guess based on the camera or the format.

For formats whose metadata it cannot parse, it reports that it does not know rather than telling you the file is clean.

Usually because the original was already efficiently compressed, or because the image is small and flat — re-encoding those can cost more than it saves.

When that happens the tool says so and disables the download, because handing you a larger file and calling it optimised would make the whole report untrustworthy.

The file size divided by the number of pixels. It is a quick way to compare compression across images of different dimensions: a quality-85 JPG photograph usually sits around 0.3–0.6, while the same photo stored as PNG is often 2–3.

Up to 50. Each one is analysed separately and gets its own recommendation, so a batch of mixed screenshots and photographs is handled correctly rather than being given one blanket setting.

Anything that came out larger than the original is left out of the ZIP, and the summary says how many that was.

Never. The original is only read. Optimising produces a new file that you choose whether to download.