What happens to a file you drop here, and where the browser stops being enough.
Image conversion happens in your browser's own canvas: the picture is decoded, redrawn at the size you asked for and re-encoded in the target format. Nothing is uploaded, so there is no queue, no copy on a server and no retention policy to trust. Close the tab and the file is gone from memory.
This has a side effect worth knowing about: re-encoding drops the metadata the camera wrote, including GPS coordinates. Converting a photo before publishing it strips the location it was taken at.
A canvas can only decode what the browser already understands — in practice PNG, JPEG, WebP and, on most systems, AVIF. It will not open TIFF or camera RAW, and a large enough file will exhaust the memory a tab is allowed to use. Long video is beyond it entirely.
Those cases need a server, and that is the half still being built. When it opens, the tools that use it will say so before you send anything, and the privacy page will state what is stored and for how long. The rule stays the same: nothing is uploaded where the work can happen locally.
Every converter here is a definition rather than its own little application: the formats it accepts, the options it offers, the specification behind them and the results it has to reproduce. A shared engine renders it. That is why the tools behave consistently, and why a converter with no passing test cases never reaches the build.
Format behaviour is transcribed from the MDN image format guides and the relevant W3C specifications, and the citation stays next to the claim on the page. Where a format cannot store something — PNG keeps transparency, JPEG does not — the tool says so rather than silently discarding it.
5 published converters, with 15 reference cases behind them.