JPEG XL

It was not just broken. It was constantly crashing, freezing, on quite basic websites.

It might be hard for you to hear this news about something you used to admire

but a lot of sites like blogs and stuff, for me I tested a lot on boorus, it was fast

I used to admire mozilla too

But they changed

a lot

they really changed

They didn't change, they just try to survive

ah yes, them with their ludicrous amounts of money they refuse to allot to firefox xD

If they changed, they would just go with Blink as Opera and Microsoft did

They definitely changed. They are not the mozilla I used to admire, at all. They don't even resemble it at all anymore. It's a bunch of corporate vampires sucking the brand name dry

baker had a 5.59m dollar bonus last tax return, they arent struggling at all

Because it's just cheaper to use Blink that develop your own engine

OMG, just stop counting money of other people, OK?

no, they cant do that, because they would lose the last bit of the facad they have left lmao

mozilla used to fund video/audio codec development like xiph.org and opus and av1 developers

Well, switching to Blink helped Microsoft

they fired everyone there too

It also helped Opera

no? we have the right to do that, thats very explicitly why tax information for non profits is availible

Mozilla is not non-profit organization, lol

he have the right to know where the money we donate goes, and it's not to firefox

mozilla is a non profit that owns two for profits

Your money, really?

yes. I donated to them before

I used to be a large supporter of mozilla

They literally fired every talented engineer working on world-changing projects like AV1, Opus, and Servo

Bigger than Google? 🙂

that's not true, most of the cuts were in non-Firefox teams, Mozilla still has hundreds of people working on Firefox.

I mean, donations is just small percentage of their income

Well, in servo's case, it was "potentially" world changing

im not sure about hundreds, the offical firefox team page lists about 30 or so

And they spend obviously much more on important projects than they receive donations

that's also not true. Gecko improved to the point in terms of modularity that they could work on improvements directly in it, instead of as separate projects.

They don't have any important projects anymore other than a skeleton crew to maintain firefox with bare minimum patches and updates

but judging from the tax info from mozilla they are in no way struggling, they could easily afford to hire more firefox devs but choose not to, this is why so little people still care about firefox

never forget too short staffed to review merge patches the communtiy provides xD

Development is not like "if you hire twice more developers, development will go twice faster", it doesn't work like this

community literally implemented good JXL for firefox, and mozilla has too little resources to review and merge the patches

https://tenor.com/view/wheeze-wheeze-laugh-nikkal-kundhal-nikkalkundhal-gif-23223780

Yes they are literally too short staffed to click the merge button

hitting the merge button is not how it works

it's very obvious that neither of you know anything about development

well you do have to review it first, but still

The time it took to write the vaguely apologetic reply would have been enough time to review the changes in the patch

considering it would be literally nothing extra for maitnence, yeah big DOUBT from me

They cut a lot of side projects. Servo is not Mozilla, Rust is not Mozilla anymore, etc. They had to cut side projects to keep enough developers on Firefox. And they are actually doing great with Firefox.

and it was a bugfix patch

> considering it would be literally nothing extra for maitnence 🤦‍♀️

well it was three patches IIRC, but all of them were fairly simple to review

Fixing the color bugs, transparency bugs, and animation support.

the patches didn't really add anything outside of what they are currenty already maintaining afaict

yeah, three patches

And of course they're not gonna bother merging new JXL stuff when they've decided to not care about JXL unless it becomes popular.

and hence, how mozilla has changed

I don't remember mozilla enthusiastically merging webp support just because someone else did the work or whatever

> and hence, how mozilla has changed They changed around 2000 then 🙂

webp is still a buggy mess, today, implementations like gdk-pixbuf still have issues. JXL is in a far better state now then webp ever was

For example, they removed MNG around 2003 also to avoid burden of supporting an unpopular image format.

but in the past, firefox was the people pushing new technologies, if it was something genuinely cool, firefox was behind it 100%

Before they were more willing to get random stuff into browser, but quite early they started to think what they add and support. When Firefox was released, it was already not just a bunch of geeks that were ready to add any random thing into the browser just because it looks cool.

this started to slow down around 2010-2015, but it wasnt until around 2020 or so that their momentum took a nose dive off a cliff

mozilla started killing off a lot of their cool projects, and thats when mozilla died to me

Not for webp, but I do remember them doing that for avif

hence how they have changed

they were no longer the same company that I felt good about donating to

the mozilla I used to know actually hired a team of developers to develop mozjpeg in response to webp

tbf, avif really should have been simple to implement, why moz took so long for animated avif... yeah...

> Not for webp, but I do remember them doing that for avif It's just because AVIF is easier to support, it shares most of code with AV1 that is already there

and yet it still took moz ages to support basic animated avif support xD

donations never went to anything developed by Mozilla Corp anyway...

Well, count how many years took Google to support APNG 🙂

Maybe 10 or more?

Im not sure why this matters, mozilla is mozilla is mozilla, mozilla non profit owns mozilla corp, well I wanted my donations to go to firefox anyways. but clearly mozilla doesn't care about firefox as much as we did xD

Different priorities, popularity of a technology, etc etc etc., so probably animated AVIF wasn't that high priority if it wasn't supported for long time (but I guess it's not 10 years)

money legally can't flow from the non-profit side to the for-profit side

I was surprised to see how fast Mozilla implemented Nested CSS, for example. The spec writers were dancing around expectations of Blink devs, trying hard to meet their expectations and limitations of their CSS parser. At some moment the Blink devs implemented it as an experiment, and quickly it became "production". I thought that it will be required to redesign a lot in Firefox to support it, but Mozilla implemented it just in a few months and even better than in Chrome (similar level of support is coming to Chrome soon too).

again, im not really sure why this matters when mozilla corp is supposed to benefit mozilla non profit,

Well, the company spends much more on important projects than it gets donations. All the extra money from deals with Google and others. I see no issues if a manager gets huge salary in such case. This huge salary is definitely not from the donations.

I have an issue with it when firefox is in the state it is in

there is no reason that firefox should have resource limitations

they have the financial means to expand firefox development efforts and don't that's the issue I have with mozilla

the ONE thing that avif is actually useful for

it's also worth noting that mozilla does accept donations for projects under MZLA corp

Except that they probably didn't need to create a whole separate file format for it

It becomes faster and better every year. The only issue of Firefox is shrinking market share. And it requires money not for development, but for marketing. And even with money for marketing it is not clear how to spread it when you don't own a popular OS.

Firefox has not gotten faster or better for years now

lmao no. no amount of marketing will ever increase firefox's market share

It hasn't changed hardly at all for more years than I'm willing to count

Yes, I told this, because Mozilla doesn't own a popular OS

firefox for me, is still slower on lower powered gpus and ram limitations

I disagree that this has literally anything to do with it

Actually it has gotten better at tab unloading in recent years, I think. about:unloads is useful

for instance, on my chuwi hi10x running arch, chrome is leagues faster. on android, chrome is also faster

It's just stats. If you don't own a huge platform that could promote your browser for free, you immediately loose.

chrome being faster doesn't mean firefox hasn't gotten faster

on my lenovo N22 with 4gb of ram, chrome is once again faster

firefox has definitely gotten faster over the last year or two

(source: I have 20000 tabs open)

Firefox has always been the best for having 20,000 tabs open

like for the past 10 years or so

nah just a few years ago it was really bad at it

tabs doesn't really matter too much since both chrome and firefox do a good job at dealing with many tabs, mind you, I only have around 200 or so tabs on both

I remember when I could barely have 2000 tabs open

on the same hardware, mind

this is ofc, completely irrelevant to the vast majority of users

The only time Firefox was actually gaining significant market share was the time when Google was promoting Firefox.

the front page of the internet telling you to install chrome has probably helped chrome on desktop more than anything

You probably don't remember it, but there were even little Firefox ads on google.com that were displayed in IE

what? firefox absolutely dominated in the windows xp days, it wasn't until google started really pushing chrome on google that chrome took off, ofc chrome did some shady shit, but it's not like firefox was really starting to struggle until a bit after that when google started to sabotage firefox

other browsers were a lot worse, though. I know, I tried.

I also had literally thousands of tabs open

firefox was starting to recover from that, but they stopped innovating

Sorry, but Firefox never dominated, Windows XP days were days of Internet Explorer

And other browsers have always been worse than firefox at that thing

It gained some market share because (thankfully) Google was pushing it for some time, but no more since Chrome was created

And then Chrome squeezed out both IE and Firefox

The highest market share that Firefox ever had was around 25-30% (in 2010). IE had around 95% in XP era.

IE was only popular because it was preinstalled on everything firefox had around 20% or so market share until chrome started picking up stream, which was actually significant because it was something people went out of their way to install, firefox maintained 20% market share which again, not an insiginificant amount, until chrome started to sabotage it

for anybody who actually needed something more then email, searching and a couple forums, firefox was the go to

everyone developed their sites on firefox, and IE,

Well, Firefox consistently looses its market share since 2010, and the reason is: Google stopped to promote it. That's it. It's that simple.

no, that's not the case, it's because firefox was genuinely a worse experience and it *never* recovered from that

even today, chrome is still a much more reliable browser, and preforms better on lower end hardware

Yeah, of course, worse experience 🙂

By the way might I interest you in Pocket? No? Please, here, have Pocket anyways

Well, there were always features that I didn't use in any browser

no? fine but take VPN instead

Don't see any issues. If you don't need a feature, just don't use it. You can even remove the button using standard means if it bothers you.

Otherwise, you probably had to stick to IE6

Because tabs are also not a required feature

they can spend money on implementing pocket and VPN, but not on reviewing a PR that wont even be added to stable

https://cdn.discordapp.com/emojis/758892021191934033.webp?size=48&name=kekw%7E5&quality=lossless

VPN might generate some money for further development

the VPN was already developed by someone else anyway

yeah but money to do what with, pay baker a better raise and not spend on firefox

xD

so it is basically free money

You are not even trying to be constructive

constructive in what way?

I don't see the point in supporting firefox, which not even mozilla is willing to adequate support

You have some expectations. OK, I understand. But you also need to meet your expectations with reality.

They are not spending the money on development.

the reality that firefox is somehow understaffed despite mozilla having more then enough funding for it?

They are literally just giving the corporate execs raises and firing devs

They are developing a major browser engine.

That is up to date.

If they didn't spend money, it would quickly become obsolete.

Try to use Presto that was abandoned by Opera.

Your claim just contradicts obvious reality.

And development an engine is an extremely expensive thing.

They are spending less and less money on developers and more money on corporate salaries

Im not saying they arent spending money on it

That's the facts

It's opposed to what you're saying and believing

im saying they arent spending enough to make it actually progress instead of just skim by

Microsoft with their billions gave up, but Mozilla still tries to do their thing.

those corporate salaries are still very tiny to their overall development costs

sure, but thats not the point

so while I personally think the executive suite salaries are silly, cutting them to zero would not make much of a difference

I see the result. An impressive result. I don't care how much money they spend on management if I see that this management really moves forward the project I like.

the reality is that they have less money to spend each year than they used to

pretending that mozilla is somehow struggling in finances despite the CEO getting a multimillion dollar bonus is just lying

True, the salaries are kind of a red herring, but they are still spending less and less on development

judging by the tax returns I read, I didn't get this from that at all

and if that is the case, they can stop donating to frivolous and unrelated stuff

what are you talking about?

It hasn't been moving forward though. How has firefox significantly changed in the last 10 years?

And they cut some of experimental side projects like Servo

please tell me this is a joke

Exactly because of this

I am still using firefox to this day

I've been using it for decades

The past 10 years it's hardly changed at all aside from minor patches here and there

well there have been a couple signifcant changes like quantum, but it does feel very insignificant

> It hasn't been moving forward though. How has firefox significantly changed in the last 10 years? 10 years ago Firefox 24 was released. Try and compare it with Firefox 120.

I see a lot of progress in both user experience and developer experience

A lot of it has just been firefox going with the flow however, there hasn't been too much of actual innovation

As a developer, I'm quite impressed how much they really do as a single company

well, mozilla is 1 foundation and two companies https://cdn.discordapp.com/emojis/867794291652558888.webp?size=48&name=dogelol&quality=lossless

I think the biggest real thing mozilla is doing right now is common voice

really wish they kept the other one, that coqui took over? coqui hasnt really been doing anything good with it

https://commons.wikimedia.org/wiki/File:Firefox_29.0_on_Windows_8.png

A single page could just freeze the whole Firefox 29 easily

The whole browser was single threaded

They kept it till 57 because of compatibility with old addons that expected synchronous access to tabs

goana my beloved

And yes, I remember days when I was waiting for a minute while my Firefox was starting

Back then, it was single threaded.

But chrome had already been out for a long time.

I still wait for nearly a minute for firefox to start lmao

Because I like to open hundred tabs and Firefox was struggling to start with them

But these days it starts in a second even with around 800 tabs

I wish that was me

https://cdn.discordapp.com/emojis/720670067091570719.webp?size=48&name=cheems&quality=lossless

Chrome was multi-threaded from day 1 and it took a glacial amount of time for firefox to react

They optimized a lot of not so popular edge cases

not as multithreaded as servo

in servo even the webpage rendering is threaded T.T

Yes, STILL not as multithreaded as servo

> Chrome was multi-threaded from day 1 and it took a glacial amount of time for firefox to react The main reason was extensions.

Servo was created as a reaction to chrome

And then mozilla cancelled it :)

at least someone hired igalia to work on it

They were considered the main power of Firefox. And Chrome extensions were just a joke in comparison.

But those extensions were also the curse of Firefox.

damn, Mozilla spent 200 million $ on software development in 2021

Mozilla created servo so they could have a multithreaded browser engine to compete with chrome, that was faster and more multithreaded than chromium was

yeah mozilla is actually quite the sizable foundation and corps

servo was never intended to replace gecko

it was designed to be merged into gecko, and to be a standalone embedded engine

Then they said "nah, we should only focus on making as much money as possible, and improving the browser beyond the bare minimum doesn't matter, people respect our brand name and history"

looks like they got 7 million in donations

yeah, donations are no longer a large part of mozilla, they actually havent been for a very long time

They almost never were.

Maybe during very early days only.

Before Firefox.

So they rest on their laurels, and so does chrome now since they have no competition now

Mozilla Suite times

it was meant as a place to develop new stuff, and a lot of that new stuff was in fact merged into Gecko. But the standalone embedded plans were mostly from people outside Mozilla being interest in that use case, and AFAIK those plans were dropped long before they dropped Servo in general.

even during firefox, the winxp days it was not insignificant

in 2005, 95% of Mozilla's income was from Google

Most of the money were from Google search.

no, this was very much an active goal for servo, samsung played a large part in servo during the early stages

https://blog.mozilla.org/en/mozilla/mozilla-and-samsung-collaborate-on-next-generation-web-browser-engine/

But Mozilla did not have their own such plans

they did, hence the post I just posted

No, they had outside collaborators who wanted to do it

Mozilla doesn't have any embedded products or anything

> At the same time, we will be putting more resources into Servo, trying to prove that we can build a fast web browser with pervasive parallelism, and in a safe, fun language. We, along with our friends at Samsung will be increasingly looking at opportunities on mobile platforms.

they did at the time, mozilla was actually pretty active in developing for mobile

from the firefox OS, geckoview

and a lot of fun stuff

That's not a new embedded browser, though.

?

That's Gecko, the thing they never intended to replace.

As far as I know, Mozilla never had any plans to use Servo outside of research themselves

They intended to use Servo to improve Gecko

And they did

But other people were interested in using Servo, and Mozilla was totally open to that.

again > **At the same time, we will be putting more resources into Servo**, trying to prove that we can build a fast web browser with pervasive parallelism, and in a safe, fun language. We, along with our friends at Samsung will be **increasingly looking at opportunities on mobile platforms.**

that doesn't contradict anything I said

yes it does, it directly contradicts it as mozilla is stating that they wanted to put more resources into servo and looking at expanding it into mobile platforms

"putting more resources" super obviously doesn't contradict anything I said

it was a project at Mozilla Research, why would putting more resources into imply anything about replacing Gecko or whatever?

It would be lame for them to waste money on development of 2 browser engines. Microsoft couldn't manage with one.

And the second one doesn't mean a whole lot

It's just not possible to develop a competitive browser engine from scratch in reasonable time.

again, they never wanted to replace gecko, gecko makes a terrible embedded engine

they wanted to merge servo parts into gecko, which they succsessfulyl did to a large extent, and have an embeddable offering

And they continue to improve gecko.

you can see how close they actually got with servo here by loading the legacy layout https://wpt.servo.org/

since it largely hasnt been worked on since servo has been abandoned

servo was actually shapping up to be a very competent web engine, so competent that people have sponsored igallia to actively work on servo to continue the efforts into making an embeddable web engine

what exactly do you mean by "an embeddable offering"

the bones for servo, are increadibly strong, it was never just a dumping ground for new tech, it was a very well thought out project

webview, low end stuff etc.

a lot like webkit is

what I mean is, it's hard enough to make money from a web browser even when you control defaults like search engine

so a plain embeddable component would have been extremely hard to justify

unless there happened to exist some partnership, such as the one with Samsung

It was being justified until the managements changed, and nothing in the tax forums suggested a major financial issue, it was just a turn in leadership

it was the shift in direction, it's as simple as that

them having an embeddable offering doesn't really make sense, but if some external entity like Samsung wants to dump money into it, it makes sense

so it seems more likely that they simply lost that external funding

probably samsung figured it was easier to just use blink or webkit

Servo could become something like Sciter (https://sciter.com/). It's exactly embeddable HTML/CSS/JavaScript engine.

But it is not suitable for rendering web.

It's just for UI of apps.

Extremely limited support of standards isn't a huge problem when you need to run just one UI of a program that was developed exclusively for this engine.

As I recall, beyond simply needing to make Gecko modular enough, by far the biggest time consumer in integrating Servo components into Gecko was simply making sure they worked correctly for every single website ever.

Like for years after they were available in Firefox via about:config, they couldn't be on by default because they couldn't handle everything yet.

So yeah, using it for apps is very reasonable.

We could replace Electron with something Servo-based.

Servo is in development again, maybe one day it will become usable for UI of programs

But I'm quite skeptical if it will manage to keep up with web standards even in 15 years

It's a moving target, while you develop features that were added last 10 years, another 10 years of features appear. Unrealistic goal.

And I'm surprised that Mozilla manages to keep Gecko up to date with modern standards

Because Blink is developed by many rich companies, not just Google (even though Google makes key decisions).

https://chromium-review.googlesource.com/q/author:*.microsoft.com+AND+status:merged

For example, commits into Chromium from Microsoft

it made a lot of sense, because mozilla was always about open technology

Amount of development going around Chromium is just mind blowing if you try to follow it

developing and pushing open and innocative technology

And Mozilla is keeping up literally alone

servo fit directly into their mission statement perfectly, servo is still a revolutionary webengine

It's a revolutionary web engine just in plans and words

In reality it's just a failed experiment with a few useful bits (that are used in Firefox)

as we now know, embedded usecase for web engines is a massive market, it plays a very significant roll as electron has showed us

bullshit, if it failed no one would be spending money to hire devs to work on it

Somebody wants to continue experiments, it's fine

Some people and companies can afford experiments for some time

Mozilla also did, some of the experiments were very successful (e.g. Rust), some not so successful (Servo)

not experients, people want to use this in production for frameworks like tauri

because servo has really solid bones, it is a very well thought out webengine

That is the use case mentioned above that doesn't require full web compat

you are right, it doesn't, but it does need a significantly large portion of it.

keep in mind this embedded use case is in cases literally running full websites through it

people use Tauri to render arbitrary web pages? 🤨

yeah, see stuff like lemoncord

it uses wry

https://github.com/japandotorg/LemonCord/blob/main/src/main.rs

Out of curiosity, downloaded the latest Servo. Well, it's improved a a lot: it doesn't crash when I just click something 🙂

But even if you try the demos on demo.servo.org, Firefox performs better

probably, servo isnt up to date on deps like webrender or anything yet, so there is a lot it needs to catch up on

For example, the dogemania is terrible from the first dog in Servo, and Firefox is smooth with hundreds

webrender, mozangle, spidermonkey etc, servo needs to play catchup on a lot of these updates, and it takes a decent amount of work to update them

https://github.com/servo/servo/pull/30323/files

The point is, Servo never was impressive. I remember playing with all these demos 5 years ago, and it was almost the same situation

well no shit, considering for nearly 4 of those years servo hasnt been developed on at all outside of the odd contribution xD

I'm talking about 5 years ago

not to mention they changed to the new layout engine by default which put them back a lot of progress

5 years ago Firefox and 5 years ago Servo

ah sorry, it was nearly 3 years, not 4 my bad

Servo didn't look promising even in their standard demos that in theory had to impress

but yes, including the web engine being changed, servo has a lot to catch up on

Again, I was excited about servo as you were, just because of reading the marketing materials Mozilla published

But whenever I tried to test it every now and then, it always disappointed

I didn't see anything promising except marketing claims

And I'm talking about all the years of active development of the engine

then you didn't test it well enough, ofc servo is far behind firefox in terms of perf, especially now when it is nearly 2-3 years behind in development time and dependencies. but on large webpages when a lot of items needed layed out, servo was really fast

not to mention, servo only really started in 2012 or something like that, and only passed acid2 in 2014

Meanwhile... Dogemania demo. Servo completely froze at 180 doges. Firefox shows 1000+ doges and it is still smooth.

probab;ly

but here are some old benchmarks from back when servo was actually being developed https://www.phoronix.com/news/MTgzNDA

And since then Gecko was optimized and borrowed best parts of Servo

yep, but it's not like servo inhereited everything, they realistically inherited what they could with their rushed schedule

but it was a significant portion that was pulled into firefox, things like webrender were originally a servo thing iirc

Gecko became faster. And it continues to improve. Goal achieved?

some goals, mozilla never did get an embeddable web engine

Why would they need it?

They don't have an OS where they could offer that engine

I explained earlier. but for an actual embedded use case that moz wanted servo for, IIRC firefox reality was one project they had used it for

yeah they did

firefoxOS

That failed dramatically

well kaiOS

Yes, they tried

but even without that

it's not like firefox needed it

even without having their own OS, linux would have picked it up, and the interest for android was healthy

and ofc you had projs like firefox reality and leap

OK, whatever, it's too late here, I'm going to sleep. See you 🙂

then again, that was also back when mozilla was pushing webxr hard

cya, havea good sleep fren

https://cdn.discordapp.com/emojis/1160370117302046780.webp?size=48&name=pepeHappy&quality=lossless

KaiOS is currently used in hundreds of millions of phones, and Mozilla collaborates with the developer to help keep its embedded Gecko up to date.

the very fact that they have to collaborate with the devs is telling, and yeah I completely forgot, but mozilla did have examples of them using servo for embedded use cases, the firefox realtity for the MS hololens was one such project

the webxr stuff was actually a decent chunk of servo dev direction

but being a proper electron replacement, right now is a decent driving force behind servo atm, which does need a good chunk of webcompat, because of this we are seeing servo flourish, it has a large uptick in activity, so it's really nice to see servo see its potential of what it was once aiming for

KaiOS is iPhone-level locked though.

Only one phone jailbroken so far.

And only web apps.

Developer modes enablement are few and far between across phones.

Certainly not in US phones.

I realized this isn't the place to discuss this though.

Oh, huh, apparently this lists Jpeg XL support even though the page looks from the 2000s https://meesoft.com/Analyzer/

possibly JPEG XR?

but it does support Windows 11, so I guess it's current (and the website uses https :p)

Speaking of JXR, djxl can't export anything to it yet, not even HDR photos that can be used as wallpapers in recent Insider builds of Win11 (not sure about 23H2).

Windows is the only use case for JXR, and even then the files Nvidia's Photo Mode outputs show incorrect mapping in the Photos app

Cool software. Especially the plugins.

The feature preserving resize is something I've only seen as a proof of concept in 2000s blog pages, unfortunately didn't work well on my images since I try not to leave much room in the first place

https://ptgui.com/versionhistory.html

> Version 12.24 (5 November 2023) > Added support for DNG files with embedded JpegXL encoded data, as created by recent versions of Adobe LightRoom and Photoshop.

support is spreading

Is it similar to context-aware scale?

They just need to add OS level support like Apple already did what-feels-like ages ago

Actually JXR seemed really nice, with the way it handled film grain, and extremely dark textured shadows, things that JXL struggles with in 0.8.2

JXL, at least the latest release, is very blurry and blocky in default lossy mode. Actually in my opinion “lossy modular” mode looks subjectively better on average to my eyes. Even though it’s not optimized or “supported”

It does have 'support'. It displays in file explorer and the photos app, but for some reason in completely the wrong format

Does it even display properly as a HDR wallpaper on Windows 11 Insider builds?

Wait what? Windows has support for JXL?

not without jxl_winthumb

JXR

as mentioned its JXR but oddly in one press release ms wrote JXL ("Fixed an issue where it wasn’t possible to select .JXL files for your wallpaper slideshow.") . worth rechecking and pushing for tbh, windows supporting this out of the box would be huge for apps, games and photography

https://blogs.windows.com/windows-insider/2023/08/10/announcing-windows-11-insider-preview-build-23521/#:~:text=JXL

really seems like a typo

actually I'm almost certain it's a typo

it's the only thing that makes sense

Hmm... "L" isn't close to "R" on the keyboard, but maybe the article writer was using Dvorak keyboard layout.

writter is a JXL fan keke

I mean who knows, they might've been looking at it recently

You think? They're not the fastests in adapting image formats, they still didn't enable avif in Edge

For preservation's sake

I was just going to say, L and R are not on the same row, but they are adjacent on a Dvorak board

Oh wait they are on the same row actually lol

I use Dvorak and I thought L was somewhere else (I avoid pinky typing)

Also Apple finally added support for Dvorak layout on the iOS keyboard. Weird they took that long considering their history with promoting Dvorak. But it offers little advantage on a touch screen. It was designed for touch typing, not touchscreen.

I still use qwerty on my phone screen

Nah their ego is too big to abandon JXR.

Updates: November 21: 13.92% of Cloudinary image requests came from a user agent supporting jxl (2.8 billion out of 20.1 billion images served that day) December 5: 19.52% of Cloudinary image requests came from a user agent supporting jxl (3.1 billion out of 15.7 billion images served that day)

<@794205442175402004> do the jxl team have a plan to convince the pdfa to add jpegxl to pdf ? do you have some lobyyist in the place ?

I will ask Leonard from Adobe next time I see him what the plan/timeline is. Last thing I heard from him is that it is planned to be included in a future version of PDF. But no idea what exactly or when this is happening.

here (https://pdfa.org/pdf-moves-ahead-with-hdr/) they spoke of both avif and jpegxl but as "possibilities"... It would be sad to miss such a chance for support 🙂

~~wow they're adding jxl support to tumblr?~~

more efficient losless mode than in avif may be a selling point for jpeg xl in pdfs. lossy mode higher target bitrate also seem to fit pdfs more

The key sentence I noticed was about emphasis on backwards compatibility, so as others mentioned the JPEG transcoding is a good selling point

Indeed, then I wonder if they'll be considering JPEG XT, but then there's no mention of that on the page

Ideally they'd want a format already with a moderate presence, there's no point in using JPEG XT when nothing can encode it, AV1/AVIF is already dominant, and JXL is on Apple with Adobe strongly pushing and even forcing software to adapt to it

JPEG XT's full encoder library is under GPLv3; there's an ISO-licensed one but it has no support for loselessly encoding images.

And it's very irrelevant.

There's no serious support for it.

And the reference repo for it only stands at 287 stars at GitHub.

but neither JPEG XL nor AVIF is backwards compatibile in the sense that existing readers (maybe apart from web browsers which implement avif) would be able to decode it. Or maybe they'll explore Google UltraHDR extensions

I only expect UltraHDR extensions to be explored to be honest.

It already has Android backing.

The permissively-licensed variant of JPEG XT library can't do anything meaningful besides HDR read/write.

And alpha channels.

does it even matter? pdf readers keep extending their support across time and remain able to decode even the most ancient pdf specification versions. if acrobat reader adds decoding for those formats, the other apps will follow - including browsers' own pdf decoders (ie mozilla's pdf.js). Not that a browser shouldve been decoding pdfs in the first place but that was once a strategy to maximize adoption by leveraging file associations

I was mainly referring the sentence "Addressing backward compatibility will be a key objective.", which jonny pointed out in the article. If that is "key objective" I wonder if they'll go with some existing HDR-backwards-compatibile JPEG extension, or if they'll come up with something entirely new.

images lacking hdr data in the first place need actual replacements, gainmaps cant be generated from thin air. afaik print only accomodates sdr-level data and hdr-ready displays may mislead with unrepresentative previews

animation <:Hypers:808826266060193874> https://patchwork.ffmpeg.org/project/ffmpeg/list/?series=10238

Yes, I am considering a case of new pdf document and legacy reader. Someone may want to create a pdf in which he'll include HDR image that he would still want to display in ie. older ebook reader that won't get an update

but I'm merely speculating, as there are no details on this in the article

some movement in Midori, a minimalist browser now based on gecko/firefox. https://github.com/goastian/midori-desktop/commit/cc99564b32b785c6bf5651a7e7af1e3078e3307a

> adds JXL to some inactive part of code for completeness sake. oh alfonso, never change

afaik hdr images display on non-hdr displays as if they were sdr all along but that wouldnt be the issue here - an old reader not receiving further updates would need to be compliant with the newer pdf spec version in the first place. unlikely unless pdfs bundle some decoders like they can with fonts but fallbacks do exist (rendering everything else, presenting a jxl thats actually a reconstructed jpg so just not the most efficiently compressed)

And there comes basic solution like JPEG XT or Google UltraHDR, or some other new extension they may come up with, which allows to add HDR data to standard JPEG that will be decodable (in SDR mode) with an exististing jpeg decoder in legacy reader, while being decodable in HDR mode in upgraded reader. Also JPEG 2000 is allowed since PDF-1.5 (year 2003), so they may come up to use with a solution to push HDR through it, but I'm not sure what is the status of the decoders for its HDR mode in existing readers. It's interesting for me how they'll solve this, because eventually the solution they'll choose will be implemented (or maybe ignored if it will be completly transparent for SDR-only mode) in many readers.

not playing along and supporting ultrahdr doesnt make apps "legacy readers". even extended with hdr data, that original jpeg with a non-standard extension has the same issues new formats are supposed to solve and the sdr intent is just a plain lossy jpeg

by legacy readers, I mean the ones which are just plainly no-longer-updated, or old versions of maintained readers, but as HDR displays aren't yet a commodity I suspect any way it will be implemented, the HDR data decoding will take time to adopt in new releases of readers. Do you have any bets on what will be selected?

for me it makes no doubt jxl would be the lead pick by far. chromium/avif scheming to preventing its adoption is of no consequence to almost all other use cases, especially on reguar computers and workstations

as for physical printing of hdr images, it takes a certain mix of adapted paper and printers to even accomplish and nobody but serious photographers (pros and studios) will produce such prints. hdr is primarly a meant only for monitors

if a central authority existed, perhaps gainmaps could be fetched for certain documents in high circulation like ebooks (from their creators, photo licencing archives or what would be a distributed shader cache for public documents) as a way of providing added data to documents that have not been updated since they were produced

probably specific lighting is needed as well

I doubt that you can stick an HDR print to a random wall in a hallway and expect it to look good

oh lol I recognize this it's almost 1 for 1 directly from a pull request I submitted to Waterfox https://github.com/WaterfoxCo/Waterfox/pull/3262 I honestly don't mind not being credited that much since the reason I contribute anywhere in the first place is to improve things and not to get my name out there and if others find what I've done useful then great.

he probably pulled it from whatever floorp did since them and floorp are in a partner ship

iirc Floorp is scared to pull in anything from Waterfox (I forget the exact reason). And after checking both repos this seems to be Midori specific. Red lines are something from me, Blue lines are something just from Waterfox in the image I sent (there are some other examples but they are on different commit pages).

honestly I don't mind since I probably would have tried submitting some of this stuff to Midori or Floorp at some point anyway

iinm waterfox was at one point controlled by an ad company (system1) but is independant again waterfox.net/blog/a-new-chapter-for-waterfox/

interesting, I should tell alfonso that he needs to keep the authorship since this can lead to issues in the future

they are free to use anything I do for Waterfox or whatever since it makes thing easier for me since I don't have to submit pull requests to multiple places lol

but yeah doing this sorta thing could very likely lead to issues in the future with other people and such

Also CMYK support is nice for PDF. And lossless JPEG recompression could be applied to existing PDF files too, I guess.

<@794205442175402004> do you know if Cloudinary currently or in the future will have the option to recompress JPEGs as JXL for requests from browsers with support?

Currently no, in the future maybe. For most use cases, you are serving downscaled images so the "original pixels" are available and it makes more sense to deliver them as full JXL (or AVIF/WebP) than as a recompressed JPEG, so this is not currently a high priority thing.

Ah, I see

Indeed, CMYK support which AVIF lacks is a very good point. I hope the pdf association members will take it into consideration.

tbf this sometimes happens I wrote the pixbuf loader for FLIF and it was derived from lgpled code, and when I brought this up and one of the authors refused to relicense it, the JXL one had to be rewritten from scratch.

As long as it's not a recurring pattern and instead just an accident

https://www.pixelmator.com/blog/2023/12/13/pixelmator-pro-3-5/

seems to support JXL import but not export, unless they didn't list all the export formats

import: > HDR photos captured with iPhone, RAW, ProRAW, and ISO HDR formats like HEIC (10-bit), AVIF (10-bit), JPEG XL (10-bit), PNG (16-bit), and TIFF (32-bit from Photoshop and Lightroom). Also, OpenEXR, Radiance HDR, and photos with Smart HDR from Photomator. export: > ISO HDR formats like HDR HEIC, HDR AVIF, and HDR PNG. Also, HDR JPEG, HDR Still Image Video, and OpenEXR.

adoption is ongoing but oddly discrete. take zoner for example, to this day it never advertised anywhere but in the list of supported files it reads and writes jxl since mid 2022, not even its changelogs. the jxl code might need an update now that spec is frozen though

checked caniuse data for some countries. interestingly in the US, united kingdom, canada and japan, jxl support is around **16%** for tracked mobile devices - likely due to the really high usage of iOS devices there (globally tracked mobile is below 8%)

https://tenor.com/view/buttsmarnn-gif-18440715

Got round to messing with Adobe's DNG converter again Turns out it *can* use lossless JXL, at about 13% smaller than the old lossless on e7, although they also have presets that make little sense... The "High quality lossy" uses d0.1, which ends up even larger than lossless

how did you get it to do lossless?

is it a recent update that added this?

Turns out using the application in the command line exposes all the JXL related settings https://helpx.adobe.com/content/dam/help/en/camera-raw/digital-negative/jcr_content/root/content/flex/items/position/position-par/download_section/download-1/dng_converter_commandline.pdf

Although, the order of the commands, full path names and random missing file errors seem to be common with no documentation or printout of what went wrong

> -lossyMosaicJXL Uses Lossy JPEG XL compression with Bayer images.

sounds like an interesting setting…

-somewhat-raw

lossy raw compression is not really new (see Canon’s C-RAW or Nikon’s https://photonstophotos.net/Emil%20Martinec/noise-p3.html#bitdepth:~:text=An%20aside%20on%20%22lossy%22%20NEF%20compression%3A ), but I’m not fully sure of how mosaicised data would interact with lossy JXL (wonder if it’s VarDCT or lossy modular?)

The converter previously used an adjusted form of jpeg for lossy compression, since the addition of JXL it's been completely replaced with no way to use old jpeg compression. Resulting in some software incompatibilities outside of Adobe software (Namely LibRaw isn't updated)

The new JXl settings seem to allow lossy mosaic data, regular lossy, lossless and various quality/effort settings for each

Thank you for mentioning c-raw, I thought it is similar to s-raw (i.e. not raw at all), but found that it is real bayer raw, just with "less precision". P1 backs also have such option, however I've never used it, obviously

What if set DNG version to 1.6 or lower?

Using any of the JXL parameters sets the DNG version to 1.7

Yes, I meant "no way to use old jpeg compression", reducing version should be the way

Ah, I see. Hmm, good idea

Only just got round to it, but yes, setting the DNG version to 1.6 does use normal JPEG compression instead of JXL once again

I use this trick (but with dng version 1.1) to cook "super resolution" data into main data, and used it to cook flatfield corrections into data, instead of having it in opcodes 3.

Netsurf 3.11 released with jxl support https://download.netsurf-browser.org/netsurf/releases/ChangeLog.txt

do you happen to have another update for January this year? Sorry if I just missed it

At around 20% now. Still continuing to go up, likely because we get relatively speaking a lot of traffic from North America and Europe so our iOS market share is larger than what worldwide statistics suggest.

Also interesting: around Christmas day we had a temporary peak of around 25% jxl support. Probably because fewer people are using their Windows computers then, and more are using their iPhones...

their *new* iphones 🙂

Btw, while here is some silence, is it possible for admins\owners to give a user RO in discord?

RO?

readonly

here's the chart of amount of `image/jxl` support we are seeing in our traffic. The peak is on Dec 25 and 26. I'm curious how it will further evolve.

About 2/3 of the mobile Safari requests we see now support jxl, and just under half of the desktop Safari requests. So there is still some room for growth....

may I repost that image elsewhere?

sure

on caniuse, the *tracked mobile* % almost doubled in the markets mentioned in december. almost 30% in us, uk, canada, japan

how do you get non-global numbers on caniuse?

next to filtered | all, theres a cog icon. select the countries you want, or add your own analytic reference (keep in mind analytics are way behind, misreport a lot and dont tally adequately app-only web services and report them all as chrome)

ty

I believe it's more or less correlated to the mobile device usage during holidays

Yes, it's basically US+EU going back to work after the Christmas holiday, using their iPhones less and their mostly Windows work computers more.

i would imagine it's just more of a peak during shopping season

the ezGIF website now supports JXL. https://twitter.com/ezgif_com/status/1745445885960708233 https://ezgif.com/jxl-maker

Ah damn, they have JXL JPG conversions, but it's just standard lossy

Also a size limit of 1200 x 675 which I think is new...

Is there any website or app (other than cjxl) that does lossless conversion?

Does anyone have a clue when Interop 2024 decisions will be announced? Should be this month, but do we know the date?

<@456226577798135808>

no clue, although yeah it is planned to be sometime this month but I don't see a specific day listed anywhere https://github.com/web-platform-tests/interop/blob/main/2024/README.md

Thanks.

Yes, possibly squoosh.app or jpegxl.io but I don't remember which, in the advanced options?

XL Converter does lossless conversions in both directions. also adapts encoding effort (to original or final resolution?)

By lossless they mean Jpeg transcoding

yup, also reversible without loss of metadata

(I'd assume based on my mention of ezgif)

when source is jpg lossless transcode is always preferable. default behaviour of apps like xnviewmp is doing a complete recompression that ends up way bigger

xl converter rebuilds seems to reconstruct once progressive jpgs to non-progressive jpgs so unsure if it could be considered fully lossless or just a chunk reorganization

Last time it was Febuary 1st that the issue got closed, worth keeping an eye on the new one (And maybe adding any new news since last) https://github.com/web-platform-tests/interop/issues/430