> Key features of the JPEG XL codec are: > lossless JPEG transcoding,
> Moreover, JPEG XL includes several features that help transition from the legacy JPEG coding format. Existing JPEG files can be losslessly transcoded to JPEG XL files, significantly reducing their size (Fig. 1). These can be reconstructed to the exact same JPEG file, ensuring backward compatibility with legacy applications. Both transcoding and reconstruction are computationally efficient. Migrating to JPEG XL reduces storage costs because servers can store a single JPEG XL file to serve both JPEG and JPEG XL clients. This provides a smooth transition path from legacy JPEG platforms to the modern JPEG XL.
https://ds.jpeg.org/whitepapers/jpeg-xl-whitepaper.pdf
If you need more profs, you could transcode a JPEG to JPEG XL and convert against to JPEG. The result image would be BINARY IDENTICAL to the original image.
However, perhaps are you talking about an image on JPEG XL, using features only in JPEG XL (24 bit, HDR, etc...) that obviously couldn't be converted in a lossless way to a JPEG.
A lot of those features (non-8×8 DCTs, Gaborish and EPF filters, XYB) are enabled by default when you compress a non-JPEG image to a lossy JXL. At the moment, you really do need to compress to JPEG first and then transcode that to JXL if you want the JXL→JPEG direction to be lossless.
Trying around with some jpg and jxl files I cannot convert jxl losslessly to jpg files even if they are only 8bit. The jxl files transcoded from jpg files show "JPEG bitstream reconstruction data available" with jxlinfo, so I think some extra metadata is stored when going from jpg to jxl to make the lossless transcoding possible. I can imagine not supporting the reverse (which is pretty useless anyway) allowed for more optimizations.
JXL to JPG is lossless as in a bit-for-bit identical file can be generated
only if you got the JXL from JPG.
So he was not wrong about this. You have perfect JPEG -> JPEG XL conversion, but not the other way around.