WebP vs JPG vs PNG: Which Should You Actually Use

The honest answer is "it depends on what's in the picture" — but that's a useless answer without numbers. So we converted the same three representative images through all three formats and measured what actually came out.

The test

Three images, chosen to represent the three cases people actually run into: a photograph (3024×4032, the kind a phone camera produces), a UI screenshot (1600×1000, flat colour and sharp edges), and a logo with a transparent background (900×900). Each was converted to JPG, PNG and WebP at quality 80 using this site's own converter, and the resulting file sizes measured directly — not sourced from a spec sheet or someone else's benchmark.

File size in KB at quality 80 (PNG has no quality setting — it's always lossless)
ImageSourceJPGPNGWebP
Photograph830 KB337 KB8,636 KB121 KB
Screenshot7 KB51 KB35 KB8 KB
Transparent logo9 KB20 KB *26 KB9 KB

* JPG can't store transparency at all — this number is what it costs after flattening the transparent areas to white, which also changes how the image looks.

The rule that predicts all three results

You don't need to memorise a table if you understand why these numbers land where they do. It comes down to one question: does the image have continuous tone, or flat colour?

JPG's compression algorithm assumes neighbouring pixels blend smoothly into each other — which is true of virtually every photograph, where light, shadow and colour shift gradually across a scene. PNG's compression assumes the opposite: it looks for exact, repeated runs of identical pixels, which is what flat-colour graphics, screenshots and text are made of. Feed either format the kind of image it wasn't designed for, and the mismatch shows up directly in the file size: our photograph became a 10.4× larger PNG, and our screenshot became a 7.3× larger JPG.

WebP wins across the board in this test because it supports two different internal modes — a lossy mode descended from the same family of ideas as JPG, and a lossless mode built more like PNG's — and its encoder is simply newer and more sophisticated than either. That's the whole explanation for why it beat both formats on every single row: it isn't magic, it's a fifteen-to-thirty-year technology gap.

So why doesn't everything just use WebP?

Mostly compatibility, and it's a smaller problem than it used to be. Every current browser — Chrome, Firefox, Edge, Safari including on iOS — has supported WebP for years, so for anything destined for the web, there's rarely a real reason not to use it. The friction shows up outside browsers: some older desktop software, print services, and upload forms still don't accept a .webp file, in the same way plenty of them historically rejected HEIC. If you're building a website, WebP is close to a free win. If you're sending a file to a print shop or an application that specifies JPG, that's what you send.

What each format is actually still good for

  • JPG — universal compatibility for photographs. Every piece of software from the last three decades opens it. Use it when you need the file to work absolutely everywhere and can't guarantee the destination supports anything newer.
  • PNG — when you need a guaranteed-lossless copy, or transparency and the destination doesn't accept WebP. Never for photographs — see the ten-times-larger result above.
  • WebP — the default choice for anything going onto the web: it matched or beat both other formats on every image type we tested, including transparency.

A trap worth knowing about: WebP at quality 100

WebP's advantage isn't unconditional. Pushed to quality 100, its lossy encoder shifts into a near-lossless mode and file sizes balloon — in separate testing on this same photograph, WebP at quality 100 came out larger than the equivalent JPG. The sweet spot for WebP is roughly 80–90, not the top of the range; see our JPG to WebP page for the full breakdown across quality settings.

Related

Common questions

Is WebP always the best choice?

For anything going onto the web, essentially yes — it matched or beat both JPG and PNG on every image type in our test, and every current browser supports it. Outside the browser, in software or services that specifically require JPG or PNG, use whichever they ask for.

Why did my screenshot get bigger when I converted it to JPG?

Because JPG's compression is built for photographs with smoothly blending colour, and a screenshot is the opposite — flat colour blocks and sharp text edges. In our test, a 7 KB screenshot became a 51 KB JPG. Use PNG or WebP for this kind of image instead.

Why did my photo become a huge file as PNG?

PNG compresses by finding exact repeated pixels, and photographs — full of sensor noise and gradual tone shifts — have almost none. Our 830 KB test photo became an 8.6 MB PNG, over ten times larger, with no quality gain since JPG at quality 80 already looked the same.

Does WebP support transparency like PNG?

Yes, and in our test it stored transparency in less than half the size of PNG — 9 KB versus 26 KB for the same transparent logo.