Which image format should you actually use?

JPEG has been the web's default photo format since 1992, and for good reason: its transform-based compression finds photographic redundancy extremely well. A 12-megapixel camera original will drop to 300–500 KB at quality 80 with no visible loss on screen.

The trade-offs are real and permanent. Every generation of JPEG re-applies lossy compression to already-lossy data, so repeatedly re-saving degrades the image. There is no alpha channel, so transparent areas come out black unless you flatten them onto something first. And the EXIF block — timestamps, GPS coordinates, camera serial numbers — travels with the file unless you strip it.

**Use it when:** you are sending a photograph to someone who needs it to open in anything, or you are emailing it. **Avoid it when:** you need transparency, you need text to stay crisp, or the image will be edited again later.

PNG stores pixels exactly. It is the only format here that is guaranteed to be supported everywhere, and the only lossless one that carries a full 8-bit alpha channel with no premultiplication surprises.

That losslessness is why converting a JPG to PNG usually makes the file dramatically larger: you are storing noisy, already-artifacted pixels with zero compression loss tolerance. A 400 KB photo becomes a 3 MB PNG.

**Use it when:** the image has transparency, it is a logo, icon, diagram or screenshot where text and hard edges must survive perfectly, or a downstream tool needs a lossless container. **Avoid it when:** you are converting a photograph and care about bytes.

WebP has been a baseline feature in Chrome, Edge, Firefox and Safari since 2020, which makes it the safe modern default. It offers both lossy and lossless modes plus an alpha channel, and it beats JPEG by roughly 30–35% at matched visual quality for photographs.

It is also the practical sweet spot: encode it and you are done, with no build step to generate a second format and no `<picture>` markup juggling. If you convert one format, make it this one.

**Use it when:** you want one format that works everywhere and is meaningfully smaller than JPEG. **Avoid it when:** a strict legacy client constraint rules out non-JPEG formats.

AVIF wraps the AV1 intra-frame codec. At matched quality it is typically 30–50% smaller than WebP for photographic content, and it handles alpha, 10-bit colour and HDR. Google has treated it as the preferred format for years.

Two caveats. First, encoding is expensive: AV1 is an inter-frame codec, so producing a single still image is much slower than WebP or JPEG. In this tool that cost shows up as a pause on very large files. Second, support arrived later than WebP, so older Safari and Edge builds will not decode it.

**Use it when:** you control delivery and can serve a fallback. **Avoid it when:** you need one file that works everywhere, or you are converting hundreds of large images in a batch and the encode time matters.

GIF has one genuinely useful feature: multi-frame animation with transparency, supported everywhere without JavaScript. Everything else about it is a limitation. It is limited to 256 colours per frame, has no partial alpha (a pixel is either opaque or not), and its LZW compression is very weak by modern standards.

Converting a photograph to GIF will band badly and usually produce a file larger than the original JPEG.

**Use it when:** you need a short, universally-playable animation and have no better option. **Avoid it when:** the image is a still — convert to WebP or PNG instead.

BMP stores raw, usually uncompressed pixels. A 1920×1080 image takes about 6 MB with no quality setting at all. You will mostly meet it when exporting from an older Windows application or a scanner.

Converting BMP to WebP is one of the most dramatic size reductions available — frequently a 95% saving — because you are moving from uncompressed to compressed in a single step.

**Use it when:** absolutely never, if you have a choice. **Use it as input when:** a program only offers BMP and you want a web-ready version.