Image to WebP Converter — Convert to WebP Format

What WebP is, and what this converter actually does

WebP, an image format Google released in 2010, built on the VP8 video codec and a lossless companion mode, and standardised in RFC 9649 in January 2024, some 14 years later. It supports 24-bit colour, rising to 32-bit once transparency is included. Its selling point is compression: for the same visible quality, a WebP file is usually 25–34% smaller than the equivalent JPEG and 26% or more below a comparable PNG. In 1 live test on this tool an 800 by 600 image came out 23% below JPEG and 92% below PNG. This tool converts your JPG, PNG, or BMP images to WebP entirely on your own machine. Each file is decoded to pixels, drawn onto an HTML canvas, and re-encoded as WebP through the browser's built-in encoder, identical to what Chrome 32, Edge 18, Firefox 65, and Safari 16 ship. Nothing is uploaded; the images never leave your device, and there is no server in the path at all.

The 1 honest caveat about browser encoding

Be precise about what "convert" means here. The browser canvas exposes only lossy WebP encoding at a chosen quality between 30% and 100% via the slider, in 5% steps, so it produces the lossy variant, not the lossless one that command-line encoders can output. For photographs that is exactly right, since that is where the big savings live. For a flat graphic with sharp edges and few colours, a PNG can occasionally beat it, which is why the tool shows you the before and after byte counts for every file rather than asking you to trust a slogan. If a conversion would not actually shrink a file, you will see that in the numbers.

Why WebP is smaller: VP8 prediction

The lossy path borrows the VP8 video codec's key-frame compression. Instead of storing every pixel, the encoder predicts each 4 by 4 or 16 by 16 block of pixels from its already-decoded neighbours, using 1 of 4 whole-block or 10 sub-block intra-prediction modes, then stores only the residual, the small gap between prediction and true pixels. That residual is passed through a 4 by 4 discrete cosine transform and quantised, discarding detail the eye barely registers. Because a photograph is mostly smooth gradients that predict well, the residuals are tiny, so the file shrinks. This is the same family of techniques as 1992-era JPEG, but WebP's block prediction and boolean arithmetic coding are more modern, which is where its roughly one-third edge comes from. WebP also lifts JPEG's hard ceiling: it handles images as large as 16383 by 16383 pixels.

Transparency survives, unlike JPEG

One real advantage over JPEG is transparency support. A JPEG cannot store transparency at all, so converting a transparent PNG to JPEG forces a solid background. WebP keeps transparency even in lossy mode: the 3 red-green-blue channels are compressed with lossy VP8 while the 8-bit transparency mask is stored separately, so a logo with a cut-out background stays cut out. The conversion preserves that automatically: 8 bits of alpha give 256 levels of opacity, from fully clear at 0 to fully opaque at 255. A transparent PNG converted here comes back as a WebP whose transparent regions are still transparent, verified by reading the corner pixel back as alpha 0 after encoding, with the coloured areas intact and the edges clean.

Where the savings pay off: Core Web Vitals

The reason to bother is page speed. Google's Largest Contentful Paint metric, 1 of 3 Core Web Vitals, measures how long the biggest visible element takes to render, with a good score under 2.5 seconds and a poor one over 4.0, and that element is very often a hero image. Shaving 30% off every image on a page pulls that number down directly. A 500 kilobyte hero JPEG that becomes a 340 kilobyte WebP, a 32% cut of 160 kilobytes, loads noticeably faster over a 5 to 10 Mbps mobile link, saving roughly 250 milliseconds, and across a whole page of 20 images at 200 kilobytes each, a 30% cut saves over 1 megabyte per page load. Faster images mean a better LCP score, 1 ranking signal Google weighs.

Batch conversion and the size table

It handles a whole batch at once, up to dozens of files, looping over every 1 you drop in and encoding each to WebP at the 1 quality you set. For each output it lists the original size, the WebP size, then the percentage change, plus a batch total across all files, so a 10-file set shows 1 combined figure at a glance. This matters because savings are not uniform: a noisy photograph might drop 30% while an already-optimised JPEG drops 5, and a simple flat graphic might even grow, in which case the honest move is to keep the original PNG. Showing 3 numbers per file, rather than a single reassuring headline, is what lets you make that call per image instead of guessing. On a 12-file batch of mixed photos, expect totals in the 20–40% range, or 200 to 400 kilobytes saved per megabyte of input.

Getting the quality setting right

The quality slider is the 1 dial that matters. At 80 percent, the default, most photographs hit the sweet spot: a large size cut, commonly 60 to 70%, with no loss visible from a normal 40 to 50 centimetre distance. Drop to 60 or 50 for thumbnails and background images, often 150 by 150 or 300 by 300 pixels, where a little softening is never noticed. Push to 90 or 100 for a hero shot or product photo where crisp detail earns its bytes; the jump from 80 to 95 can double the file. Because output size is reported for each setting, the fastest way to choose is empirical: convert once, read the number, nudge the slider, then convert again until both the size and the look satisfy you.

Quick reference

• Format: WebP, from Google in 2010, built on VP8 plus its VP8L companion, at version 1.4 of the reference encoder.
• Reads: JPG, PNG, and BMP; writes lossy WebP through the browser canvas.
• Savings: about 25–34% below JPEG, and 26% or more below PNG.
• Encoding: lossy only, quality 30–100%; the canvas exposes no lossless path.
• Transparency: the alpha channel is preserved, even in lossy mode, unlike JPEG.
• Verify: before and after byte counts are shown per file, so every claim is checkable.
• Privacy: 100 percent in-browser; no upload, no server.
• Best setting: 80 for photos, 50 to 60 for thumbnails, 90 or above for hero images.
• Max dimensions: WebP supports up to 16383 by 16383 pixels, well beyond typical web use.
• Browser support: Chrome, Edge, Firefox, Opera, and Safari 16 and later, over 95% of users.