format conversion

JPEG to WebP

Two lossy formats, so the honest description is a re-encode rather than a conversion: the WebP is built from the pixels the JPEG already settled on.

What actually changes

  • A second lossy pass never adds detail back

    The WebP is encoded from a lossless master decoded out of your JPEG, so it is a faithful copy of an approximation. Whether it is smaller is a question about the two encoders, and the results table answers it on your file rather than on an average.

    Source src/lib/convert.worker.ts — the master decode, per docs/v2/22

  • Your camera data is dropped unless you ask for it

    JPEG is the format that actually carries EXIF, and a photograph carries where it was taken. Metadata is off by default here. The orientation tag is dropped either way: the picture has already been turned the right way up, and keeping the tag would turn it again.

    Source The converter’s Output panel — src/copy/en/convert.ts

  • The same picture gives you the same file

    Measured on one card at four thread counts: WebP came out byte-identical every time. It is the property that let WebP into this tool and kept AVIF out, and in a browser it decides whether a phone and a laptop hand you the same file.

    Source docs/v2/18 K15 and docs/v2/22 — measured 2026-08-22

  • Read by every browser released since 2020

    The reason to accept a re-encode at all: the smaller file needs no fallback arranged behind it.

    Source The converter’s own format notes — src/copy/en/convert.ts

The short answer

Worth it for photographs going onto a web page, once the results table shows the WebP is actually smaller. Keep the JPEG as the original.

Convert it here

The converter takes your file, writes it as PNG, JPEG and WebP at the same pixel size, and shows you what each one weighs — measured on your file rather than on an average. It runs on the page using WebAssembly; the image never leaves your device.

Open the image converter

The other conversions

Every pair this tool can actually do. There are nine because it reads four formats and writes three, and a format cannot be converted to itself.

This page carries 4 facts and a source for each one. It is one of nine, generated from the same list the converter reads — so a page cannot exist here for a format the file picker would refuse.