JPEG XL faces fierce competition from AVIF. And Google. A bit of both0:00 - What is JPEG?1:20 - JPEG XL2:07 - AVIF3:08 - JPEG XL's Strengths4:58 - AVIF Suppo...
Jpeg XL isn’t backwards compatible with existing JPEG renderers. If it was, it’d be a winner.
According to the video, and this article, JPEG XL is backwards compatible with JPEG.
But I’m not sure if that’s all that necessary. JPEG XL was designed to be a full, long term replacement to JPEG. Old JPEG’s compression is very lossy, while JPEG XL, with the same amount of computational power, speed, and size, outclasses it entirely. and PNG is lossless, and thus is not comparable since the file size is so much larger.
JPEG XL, at least from what I’m seeing, does appear to be the best full replacement for JPEG (and it’s not like they can’t co-exist).
It’s only backwards compatible in that it can re-encode existing jpeg content into the newer format without any image loss. Existing browsers and apps can’t render jpegXL without adding a new decoder.
Legacy client support. Old devices running old browser code can’t support a new format without software updates, and that’s not always possible. Decoding jxl on a 15yo device that’s not upgradable isn’t good UX. Sure, you probably can work around that with slow JavaScript decoding for many but it’ll be slow and processor intensive. Imagine decoding jxl on a low power arm device or something like a Celeron from the early 2010s and you’ll get the idea, it will not be anywhere near as fast as good old jpeg.
Google rammed webp through because it saved them money on bandwidth and because they controlled the standard. They’re doing the same thing with jpeg now that they control jpegli. Jpegli directly lifts the majority of features from jpegxl and google controls the standard.
That’s a good argument, and as a fan of permacomputing and reducing e-waste, I must admit I’m fairly swayed by it.
However, are you sure JPEG XL decode/encode is more computationally heavy than JPEG to where it would struggle on older hardware? This measurement seems to show that it’s quite comparable to standard JPEG, unless I’m misunderstanding something (and I very well might be).
That wouldn’t help the people stuck on an outdated browser (older, unsupported phones?), but for those who can change their OS, like older PC’s, a modern Linux distro with an updated browser would still allow that old hardware to decode JPEG XL’s fairly well, I would hope.
Optimized jpegxl decoding can be as fast as jpeg but only if the browser supports the format natively. If you’re trying to bolt jxl decoding onto a legacy browser your options become JavaScript and WASM decoding. WASM can be as fast but browsers released before like 2020 won’t support it and need to use JavaScript to do the job. Decoding jxl in JavaScript is, let’s just say it’s not fast and it’s not guaranteed to work on legacy browsers and older machines. Additionally any of these bolt on mechanisms require sending the decoder package on page load so unless you’re able to load that from the user’s cache you pay the bandwidth/time price of downloading and initializing the decoder code before images even start to render on the page. Ultimately bolting on support for the new format just isn’t worth the cost of the implementation in many cases so sites usually implement fallback to the older format as well.
Webp succeeded because Google rammed the format through and they did that because they controlled the standard. You’ll see the same thing happen with the jpegli format next, it lifts the majority of its featureset from jpegxl and Google controls the standard.
Adding more decoders means more overheads in code size, projects dependencies, maintanance, developer bandwidth and higher potential for security vulnerabilities.
The video actually references that comic at the end.
But I don’t see how that applies in your example, since both JPEG and JPEG XL existing in parallel doesn’t really have any downsides, it’d just be nice to have the newer option available. The thrust of the video is that Google is kneecapping JPEG XL in favor of their own format, which is not backwards compatible with JPEG in any capacity.
They’re confusing backwards and forwards compatible. The new file format is backwards compatible but the old renderers are not forward compatible with the new format.
According to the video, and this article, JPEG XL is backwards compatible with JPEG.
But I’m not sure if that’s all that necessary. JPEG XL was designed to be a full, long term replacement to JPEG. Old JPEG’s compression is very lossy, while JPEG XL, with the same amount of computational power, speed, and size, outclasses it entirely. and PNG is lossless, and thus is not comparable since the file size is so much larger.
JPEG XL, at least from what I’m seeing, does appear to be the best full replacement for JPEG (and it’s not like they can’t co-exist).
JPEG XL in lossless mode actually gives around 50% smaller file sizes than PNG
It’s only backwards compatible in that it can re-encode existing jpeg content into the newer format without any image loss. Existing browsers and apps can’t render jpegXL without adding a new decoder.
Why is that a negative?
Legacy client support. Old devices running old browser code can’t support a new format without software updates, and that’s not always possible. Decoding jxl on a 15yo device that’s not upgradable isn’t good UX. Sure, you probably can work around that with slow JavaScript decoding for many but it’ll be slow and processor intensive. Imagine decoding jxl on a low power arm device or something like a Celeron from the early 2010s and you’ll get the idea, it will not be anywhere near as fast as good old jpeg.
But how is that different to any other new format? Webp was no different?
Google rammed webp through because it saved them money on bandwidth and because they controlled the standard. They’re doing the same thing with jpeg now that they control jpegli. Jpegli directly lifts the majority of features from jpegxl and google controls the standard.
That’s a good argument, and as a fan of permacomputing and reducing e-waste, I must admit I’m fairly swayed by it.
However, are you sure JPEG XL decode/encode is more computationally heavy than JPEG to where it would struggle on older hardware? This measurement seems to show that it’s quite comparable to standard JPEG, unless I’m misunderstanding something (and I very well might be).
That wouldn’t help the people stuck on an outdated browser (older, unsupported phones?), but for those who can change their OS, like older PC’s, a modern Linux distro with an updated browser would still allow that old hardware to decode JPEG XL’s fairly well, I would hope.
Optimized jpegxl decoding can be as fast as jpeg but only if the browser supports the format natively. If you’re trying to bolt jxl decoding onto a legacy browser your options become JavaScript and WASM decoding. WASM can be as fast but browsers released before like 2020 won’t support it and need to use JavaScript to do the job. Decoding jxl in JavaScript is, let’s just say it’s not fast and it’s not guaranteed to work on legacy browsers and older machines. Additionally any of these bolt on mechanisms require sending the decoder package on page load so unless you’re able to load that from the user’s cache you pay the bandwidth/time price of downloading and initializing the decoder code before images even start to render on the page. Ultimately bolting on support for the new format just isn’t worth the cost of the implementation in many cases so sites usually implement fallback to the older format as well.
Webp succeeded because Google rammed the format through and they did that because they controlled the standard. You’ll see the same thing happen with the jpegli format next, it lifts the majority of its featureset from jpegxl and Google controls the standard.
https://xkcd.com/927/
Adding more decoders means more overheads in code size, projects dependencies, maintanance, developer bandwidth and higher potential for security vulnerabilities.
The alternative is to never have anything better, which is not realistic
Yes, it means more code, but that’s an inevitability. We already have lots of legacy stuff, like, say, floppy disk drivers
https://xkcd.com/927/
The video actually references that comic at the end.
But I don’t see how that applies in your example, since both JPEG and JPEG XL existing in parallel doesn’t really have any downsides, it’d just be nice to have the newer option available. The thrust of the video is that Google is kneecapping JPEG XL in favor of their own format, which is not backwards compatible with JPEG in any capacity.
The other thing is, disk space and internet speeds just keep getting cheaper so… Why change just for disk space?
The bottom line.
They’re confusing backwards and forwards compatible. The new file format is backwards compatible but the old renderers are not forward compatible with the new format.