Re:Overview of JPEG XT (2016)
by mkl on Tuesday March 10, 2020 @05:14AM
(#22533937)
XT, XR, XL, XS. I wish the names were more meaningful. I can never remember which one is which or what they're good for, or even which order they came in (and the jpeg.org pages don't even list years).
JPEG XL is the one currently in development, and it has what sound like some really neat features, like floating point pixel data, and lossless conversion from JPEG (and according to Wikipedia [1] back to JPEG, but I'm guessing that only applies to images that went JPEG -> JPEG XL -> JPEG).
"JPEG XT extensions are backward compatible with base JPEG/JFIF file format - existing software is forward compatible and can read the JPEG XT binary stream, though it would only decode the base 8-bit lossy image."
It is a standardization of how to put additional data in the already existing JPEG file format, the one that we already have supported everywhere.
Then, if some software "understands" what the XT parts of these JPEG files mean, it can use these new parts, still, the files are usable on every device, just these XT parts don't mean anything to the devices that don't understand XT.
Re:Overview of JPEG XT (2016)
by m1sta_ on Tuesday March 10, 2020 @06:49AM
(#22534369)
The additional data can be used, for example, but neural nets pretrained to cleanup particular images. Similarly, the data could have vectorisation data for key features.
Re:Overview of JPEG XT (2016)
by Dylan16807 on Tuesday March 10, 2020 @03:55PM
(#22538984)
Images are much bigger than javascript libraries. So by comparison it's a small library. And you're probably decoding, not transcoding.
You want to get browser support, but a gap-filler can keep you from waiting forever to reach 99.9%. And sometimes bandwidth use is more important than clunkiness.
Re:Overview of JPEG XT (2016)
by wongarsu on Tuesday March 10, 2020 @09:34AM
(#22535296)
JPEG XR is only supported by IE (since version 2011) and Edge before the switch to Blink/Chromium. Webp is supported by everyone except Safari and IE, but Safari and mobile Safari have about 15-20% combined market share.
New formats are basically only relevant if they get blessed by Google/Chrome, and even then you often need fallbacks for a long time.
Re:Overview of JPEG XT (2016)
by PaulHoule on Tuesday March 10, 2020 @09:17AM
(#22535161)
Many browsers support some formats that are favorites of some vendor, but the new formats are not supported cross-browser, so the benefits of switching formats are reduced and the costs are raised -- in particular, one benefit of smaller files is reduced storage cost. If you have to support JPEG for legacy and then something else for Microsoft's format and something else for Google you are increasing your storage cost, encoding complexity, etc.
I looked at alternative image formats for a photo site I was working on and never convinced myself that the benefits of switching formats was worth the trouble.
Re:Overview of JPEG XT (2016)
by jorvi on Tuesday March 10, 2020 @08:43AM
(#22534902)
The only ones I can ever remember are jpeg and jpeg2000.
To be honest most tech has horrible naming schemes. The iPhones were a breath of fresh air for a while (iPhone 3G/S, iPhone 4/S, iPhone 5/S, iPhone 6S/Plus) up until the 7.
Re:Overview of JPEG XT (2016)
by AlanYx on Tuesday March 10, 2020 @04:45PM
(#22539514)
The JPEG committee didn't believe it was patent encumbered, but Sisvel came along and put together a portfolio of patents it claims must be licensed to use JPEG XT. However, Sisvel at present is only claiming that devices with cameras have to license the patents.
Sisvel did the same thing with VP9 and a couple of other recent video codecs.
JPEG XL is the one currently in development, and it has what sound like some really neat features, like floating point pixel data, and lossless conversion from JPEG (and according to Wikipedia [1] back to JPEG, but I'm guessing that only applies to images that went JPEG -> JPEG XL -> JPEG).
[1] https://en.wikipedia.org/wiki/Joint_Photographic_Experts_Gro...
https://en.wikipedia.org/wiki/JPEG_XT
"JPEG XT extensions are backward compatible with base JPEG/JFIF file format - existing software is forward compatible and can read the JPEG XT binary stream, though it would only decode the base 8-bit lossy image."
It is a standardization of how to put additional data in the already existing JPEG file format, the one that we already have supported everywhere.
Then, if some software "understands" what the XT parts of these JPEG files mean, it can use these new parts, still, the files are usable on every device, just these XT parts don't mean anything to the devices that don't understand XT.
You want to get browser support, but a gap-filler can keep you from waiting forever to reach 99.9%. And sometimes bandwidth use is more important than clunkiness.
https://uprootlabs.github.io/poly-flif/
As soon as you have some usage browser vendors will then be much more willing to support it.
https://www.w3.org/Consortium/Patent-Policy-20040205/
> The goal of this policy is to assure that Recommendations produced under this policy can be implemented on a Royalty-Free (RF) basis.
So are creation tools such as Adobe’s.
New formats are basically only relevant if they get blessed by Google/Chrome, and even then you often need fallbacks for a long time.
If we can agree on a standard (hard, I know) we can star using it pretty quickly.
I looked at alternative image formats for a photo site I was working on and never convinced myself that the benefits of switching formats was worth the trouble.
https://www.sisvel.com/licensing-programs/digital-video-disp...
Sisvel did the same thing with VP9 and a couple of other recent video codecs.
See also: https://en.wikipedia.org/wiki/JPEG_XT