Why convert PNG to JPEG XL
PNG is lossless and universally supported, which is why screenshots, diagrams, logos and exported artwork are so often PNG files. Its compression, however, is based on the DEFLATE algorithm from the 1990s. JPEG XL’s lossless mode uses newer techniques — including context modelling and prediction designed for both photos and synthetic images — and the format supports alpha, high bit depth and wide-gamut colour (JPEG XL overview, jpeg.org).
The practical result is that a lossless JPEG XL file is often smaller than the PNG it came from, while decoding to identical pixels. How much smaller depends heavily on the image, which is why this tool shows the actual numbers for every file rather than promising a percentage.
What “lossless” means here
In Auto and Lossless mode the converter decodes your PNG with an exact PNG decoder (no colour conversion, no resampling) and encodes the pixels with libjxl in lossless mode. Our test suite decodes the produced .jxl and checks that every pixel, including transparency, matches the PNG.
Two cases are handled differently, and the tool tells you when they apply:
- PNGs with an embedded colour profile (iCCP). To keep colours correct, these are decoded by your browser, which converts them to sRGB. The pixels then match what you saw on screen, but not the raw values in the file, and the profile itself is not embedded. The metadata report states this.
- 16-bit PNGs. This version encodes 8 bits per channel. You will see a warning, and you should keep the original if the extra precision matters.
When to use PNG to JXL
- Screenshots and UI captures you want to archive at full fidelity.
- Graphics with transparency — logos, icons, overlays — where you cannot accept compression artefacts.
- Large lossless image collections where storage size matters.
When a lossy mode is the better choice
Photographs saved as PNG (for example exported from an editor) can be very large. If you do not need the exact pixels, the Visually lossless mode will usually produce a much smaller file that looks the same at normal viewing size. That is a deliberate trade-off: you give up bit-exactness for size.
When to stay with PNG
- When the file must open everywhere, including older browsers and tools. PNG support is universal; JPEG XL support is still growing — see the browser support guide.
- When you need 16-bit precision (this version does not keep it).
- When the PNG is already tiny: the gain may be a few bytes, or negative.
Metadata
EXIF (stored in PNG’s eXIf chunk) and XMP (in an iTXt chunk) are copied into standard Exif and xml boxes of the JPEG XL file unless you choose to remove them. Other text chunks, such as comments written by some editors, are not carried over in this version; the metadata report lists them as “Not kept”.
Related: the JXL vs WebP vs AVIF comparison explains how each format handles lossless images.
Last updated:
Frequently asked questions
Is PNG to JXL conversion lossless?
In the default Auto or Lossless mode, yes: the JPEG XL file decodes to exactly the same 8-bit pixels and transparency as your PNG. Our automated tests decode the result and compare every pixel.
Will the JXL always be smaller than the PNG?
Usually, but not always. Very small or already heavily optimised PNGs can come out slightly larger. The converter always shows the real size change and suggests keeping the original when the JXL is bigger.
What about 16-bit PNGs?
This version converts 16-bit PNGs to 8 bits per channel, so some precision is lost. The tool shows a warning for those files. Keep the PNG if you need the full bit depth.
Is transparency preserved?
Yes. JPEG XL supports an alpha channel, and the lossless mode keeps it exactly.