WebP to PNG — Save Decoded Pixels and Alpha Losslessly for Editing Masters, Icon Assets and Long-Term Archiving, Processed Locally

WebP to PNG is for people who still intend to keep working with the image, not for those who only want to send it out quickly. WebP was built to make web pages lighter, and its lossy mode hands you pixels that have already been quantised, while saving again as JPG would lose a little more; PNG writes every pixel down exactly as it is, loses no detail and supports a full alpha channel. Turning a WebP icon, a rounded button, a cut-out product shot or a transparent logo into PNG means decoding it once in the browser and then sealing it without further compression, so later colour grading, cut-outs, layer work or repeated saves will not degrade the picture. The price is a noticeably larger file — a single photo often runs to several MB — so compress it again before putting it on a web page. Import several images at once, see the dimensions and size for each in the batch list, and step through the preview with Previous and Next to check edges and gradients between the original on the left and the PNG result on the right. The dropdown switches to JPG at any time, and already-decoded images re-export automatically without being imported again. Decoding uses the browser’s own capability, nothing is uploaded and the page still works offline; the output is a clean file, and EXIF, ICC and orientation data in the container do not come along. PNG has no animation either, so a multi-frame WebP contributes only its first frame, and the batch list also lets you retry or download just a few images.

Import images

Drop WebP images here, or click to choose filesImport several at once, or press Ctrl+V to paste an image from the clipboardJPG, PNG, GIF and BMP are accepted too · up to 40.00 MB per file · 30 files maxChoose images

Target format

Output format

Both formats reuse the same decoded pixels, so switching re-exports automatically — no need to import again.

PNG keeps every pixel losslessly and preserves alpha. Best for further editing, assets and archiving, but usually the largest file.

Compression quality

PNG is lossless, so there is no quality setting for it.

Extra rotation

A few WebP files carry an orientation flag or come from a portrait shot and decode sideways. Correct them here.

Filename suffix

Leave it empty to keep the original name; the extension is replaced with the output format.

Preview and export

Import an image to compare the original and converted result side by side or with a draggable split.

Batch list

0 / 0 Done

No images imported yet.

How to use

Drag .webp icons, screenshots or assets with transparent areas into the import area, selecting several at once; Ctrl+V pastes an image too. The tool identifies the real format from the file header, so a wrong extension does not matter.
Keep the target format on PNG, where there is no quality slider to set because the output is lossless; first check that the edges and gradients match in the two preview panes.
Turn on the checkerboard to confirm transparent areas survived intact; a few WebP files carry an orientation flag and decode sideways, so correct those with Extra rotation.
Use Previous and Next in the preview to compare image by image, then download the current one, download individual rows from the batch list, or pack everything into a ZIP. Switching to JPG in the dropdown re-exports the already-decoded images automatically.

Frequently asked questions

WebP already has a lossless mode. Why convert to PNG?

It does, but a web page cannot reach it. Canvas toBlob exposes only lossless PNG export, and libwebp’s separate lossless coding needs a dedicated compiled build before it can be called from a page; the more practical difference is the ecosystem, since almost every design application, editing timeline, print workflow and archive system reads PNG directly, while a WebP carrying lossless data is still treated as an ordinary lossy image by many tools. What you get here is a master anyone can open.

Can converting to PNG bring back lost detail?

No. Lossless means nothing further is lost from this step onwards, not that information discarded in the past is restored. Banding already in a lossy WebP sky or ringing beside a strand of hair stays exactly where it is in the PNG; what changes is that later saving, grading or layering will no longer make it worse. To truly recover detail you need the camera original, the source design file or an uncompressed archive.

Is the alpha channel preserved in full?

Yes. PNG supports a complete eight-bit alpha channel, so semi-transparent shadows, feathered edges and the outside of rounded corners are written into the new file exactly as they were, rather than being composited onto a solid colour as JPG would do. You can switch on the checkerboard in the preview to see which parts are genuinely transparent; before placing the asset on a dark page, check here that no white halo or dirty fringe remains along the edge.

Why is the file so much larger, and can it go on a web page?

PNG uses Deflate, which is excellent for icons, line art and large flat areas but far less efficient than intra prediction for the continuous tones in a photograph, so a single phone photo often runs to several MB. This page is meant for intermediates and archiving, not for live assets; before publishing, export a compressed WebP or JPG from this clean master instead of uploading the archive file itself.

Why do the colours or orientation look different after export?

The conversion produces a brand-new file, and EXIF, the ICC colour profile and orientation flags inside the WebP container are not carried over automatically, so material with a wide-gamut profile or an orientation property can look slightly different in another program. Compare the two panes side by side in the preview first; if the direction is wrong, choose ninety or one hundred and eighty degrees under Extra rotation, and if the colour shift is obvious, assign the correct colour profile in the application where the image will finally be used.

What is it

These two formats disagree on one question: whether anything is lost a second time. WebP’s lossy mode trades intra prediction, transform and quantisation for file size, so the pixels you decode are no longer the original numbers from the camera or the design file; even when the source uses WebP lossless mode, what the browser hands to canvas is just that decoded eight-bit version. PNG takes the other road: it applies a prediction filter to each row of pixels and then compresses losslessly with Deflate, so what comes out is identical to what went in, and it also supports eight- or sixteen-bit depth, greyscale, palettes and a full alpha channel. The point of converting to PNG is therefore not a better-looking picture but a stable, universally readable home for the decoded pixels, so every later edit starts from the same master. One thing should be clear: lossless means no further loss from this point on, not that it can recover what a lossy WebP discarded in its first compression, so banding already present in a sky stays there. In a web page, canvas toBlob exposes only lossless PNG export and cannot reach libwebp’s separate lossless coding, so PNG is still the more direct choice for a master you intend to edit for years; APNG can carry animation, but it is not what this page outputs. The tool identifies the real format from the file header, so a wrong extension does not matter.

Origin & history

There is no new algorithm here, only two decisions — one about licensing, one about file size — colliding twenty years later. The first came in 1994, when Unisys began asserting patent fees over the LZW compression behind GIF and the web community realised that its most universal lossless image format might start costing money. Thomas Boutell opened a discussion on a mailing list, and within weeks a group of engineers assembled a replacement specification: the already-public Deflate algorithm, a full alpha channel, gamma correction and cyclic redundancy checks, which became a W3C recommendation in 1996 and ISO/IEC 15948 in 2003. The name PNG is most often explained as a recursive acronym with a degree of self-mockery; it then settled firmly into icons, screenshots, line art and editing intermediates, libpng found its way into almost every image toolchain, and APNG later added animation, yet the core specification barely changed in thirty years. The second decision had to wait until 2010: Google used the VP8 intra coding it had acquired as the basis for WebP, packing lossy images into a RIFF container and later adding alpha, lossless coding and animation, with one clear aim — sending fewer bytes. Chrome supported it early, Firefox followed in 2019, and Safari completed the set in 2020, so through those years front-end teams maintained two assets for every page. Once the mainstream browsers could all decode WebP natively, a new gap appeared: site directories filled up with compressed .webp files, while design software, print workflows and archive systems still expected a master anyone could read back losslessly. What closed it was not a new standard but two things that had existed all along: the browser unpacks WebP into pixels, and canvas toBlob writes those pixels losslessly as PNG, so local decoding plus local export does not even need an upload endpoint. A single click today looks like nothing more than a changed extension, but what it really does is resettle a file born for delivery into a master suited to keeping; AVIF and JPEG XL later pushed compression further, and newer formats need that road back to a master even more.