Image Converter

The problem

Every free converter asks you to hand your photographs to a stranger's server. Having taken them, it caps the batch, queues you behind whoever else is waiting, and sends back a file with a watermark or an upsell attached. None of that is required by the task. Decoding and re-encoding an image is arithmetic, and the browser has been able to do it locally for years — the upload exists to create the queue, and the queue exists to sell the way past it.

What got decided

Do the work where the files already are

Decoding, resizing and encoding all happen in the browser. Nothing is uploaded, which removes the privacy question, the file-size cap and the queue in one move — and means metadata doesn't survive the trip either, since there is no trip.

One file at a time, on purpose

Converting a whole batch in parallel is faster right up to the point where a large set exhausts the tab's memory and takes the page with it. Sequential conversion with per-file progress is slower on paper and finishes more often.

Detect the encoders, don't assume them

Browser support for writing AVIF and WebP is uneven. Unsupported targets are detected and disabled rather than offered and then failed — an option that isn't there is honest, and one that silently produces nothing is not.

Resize by longest edge

The control people reach for is quality, and the one that actually saves the megabytes is dimensions. Leading with longest edge puts the effective lever first.