JPEG XL

Must a blast of the years of Aquors

been watching bob ross

haven't quite figured out how to replicate his brush technique for the trees and bushes but the smudge tool works really well for his technique to get the water and the sky right

https://www.twitch.tv/mechafest

I wonder if it's possible to make a compression resistant image: like I wonder what the most difficult image to compress would be. Probably a question math ppl would know about.

random noise is the most difficult thing to compress typically

since compression relies on the concept of information entropy

information entropy is how much information something *actually* contains

like if I type the letters `Th`

there's a high probability that the next letter is `e`

just based on what's near it

hmm, so could you do that backwards then? Like un-expect the expected?

Not quite

If someone told you "he dropped an apple and it fell" you go "well that's what I thought would happen" and they provided minimal information

If someone told you "he dropped an apple and it didn't fall" you'd say "well that's specifically what I thought *wouldn't* happen"

so there's still a general idea of what you'd expect and what you didn't

if someone told you "he dropped an apple and stuffed moose toy" then that probably caught you out of left field

When you have data that comes from independent samples of a uniformly random variable, there's absolutely no way to predict what comes next (by design of the data source)

and absolutely no correlation of the data

basically, each bit of information contains 1 bit of information

so in the event of `Th -> e`, that `e` contains *less* than 1 character worth of info

but a `J` would contain *more* than 1 character worth of info

data compression relies on encoding things based on how much info they actually contain

if that makes sense

but in a random sequence, every bit of info contains 1 bit of info

which makes it essentially incompressible

wait, how can any computer data be less than a bit?

the idea is that it contains less information than 1 bit worth, even though it takes up 1 bit of space

like suppose I have 30 thousand 0s in a row

and the next bit is also a 0

this is very unsurprising

yeah

the actual information content can be roughly described as "how surprised are you by that result"

just like if I tell you something obvious and you think to yourself "well that didn't actually tell me anything interesting"

despite me providing you a whole sentence of info, info you may not have known, if it's something you'd probably have expected, then you didn't learn much

information entropy is that "how much did I actually *learn* from that sentence"

how much true knowledge is embedded in it, not how many words or letters

Like how much new information?

yea, basically

This sentence has a word in spoiler tags so you won't be ||able|| to read one of the words.

Read the above sentence, and try to guess the word I spoiler tagged ^

then click it and see if you were correct

I thought it was a swear, lol

nah, this is an experiment. Let me do another one, read it and guess what the tag is

If you can guess correctly which word is hidden ||from|| view, then you will see what I mean.

hmm, I'm guessing "from".

woo

So I guess difficulty in compressing varies by how much information something contains. In that case, I should rephrase my question as what the maximum information density something of a given resolution could contain (I stick with 1080x1920 bc it's been my monitor size since forever). Like a solid black image would be easier to compress than a black and white checkerboard pattern.

I wonder what the "noisiest"/highest information image I could make for 1920x1080 would be then.

Or if at a certain point, the two noisy images are exactly as difficult as each other.

if we allow lossy compression (which we frequently do), human can't perceive two random noises as distinct so random noise can actually be a worse choice

since this implies that *any* sort of random noise can be replaced with other matching level of random noise, as jxl often does

A theoretical jxl encoder could swap noise for generated noise, but that would be tricky to pull off in general in practice

For lossless, pure noise is definitely a worst case scenario

Is this getting into the whole "how does a computer generate true randomness" thing?

It doesn't have to be true randomness to be visually the same

I guess for lossy, the worst case is random but visually somewhat structured high frequency signal. Something like dct coeffs that are chaotic but not uniformly distributed, so it doesn't look like noise

there is this concept of "kolmogorov complexity" that you might find interesting

conceptually, "what's the shortest program that would generate this output"

interestingly, any output of a PRNG, a black image, or a checkerboard have pretty much the ~same k. complexity (I think, it's not like we can *compute* it)

I kinda think I've found a possible bug in WOFF2...???

so I'm trying some hand-crafted font for my (for now, secret) project, and I think the current implementation rejects what should be allowed according to the spec

hand-crafted compressed woff2 bitstream?

is this like one of those gzip quine funsies

mostly related to https://discord.com/channels/794206087879852103/806898911091753051/885978276844167298

what I've done for last 3 days I think

tl;dr: I've extracted the Brotli preset dictionary from the web browser, possibly for the first time

https://imgsli.com/MTE1NDE5

credits to kali

oh he's not here <:kekw:808717074305122316>

HMMM

Toggle dark mode off in individual apps in dark mode settings.

Wave Function Collapse dead me from life for a month

when 10 years ago I drew a pixel pic, but in 2022 I finished it by making it 2 times larger

then I found people with 64 gb of RAM and generated WFC images

i also experement with some parameters

and with the use of filters it can look a bit scary

and i also compiled the original source for new .NET 6, just to be

and, this is a seamless picture

https://youtu.be/nmMRMIE3MGw

Looking forward to compressing these full res images

https://cosmic.mearie.org/tmp/20220713-epyt.html ahhahahha I did it

that'd be my entry to JS1024 this year, 1023 bytes plus 2D canvas shim

neat 👍

https://twitter.com/NASA/status/1546625228691935233?t=wN-sx11E3mEBcWpkAo_PGA&s=19

Anyone compressed it with lossless jxl?

16,679,845 bytes with effort 8

https://webbtelescope.org/contents/media/images/2022/031/01G77PKB8NKR7S8Z6HBXMYATGJ?news=true Got this one down to 87,031,743 bytes, did take nearly 27 minutes to do that though :p

Sounds like time for mega slow settings in 6 hours, because then I won't be home for 90 minutes <:thinkies:854271204411572236>

I legit thought the Full Res images would be in the order of gigabytes

Hm, 121,615,720 bytes with optimized PNG (can be slightly smaller, but that'll take long af XD)

I used just `-d 0` and it came out to 88,196,301 bytes

Works out to 100,927,289 bytes for me with just `-d 0`, but I'm also using cjxl_ng so there might be differences in defaults

But it also took 9 seconds total so that's clearly better

Oh yeah, the default effort for ng is 3, I think old cjxl was 7

Is cjxl_ng the new one? Sorry, I am not that familiar apart from the basics yet

Yeah, uses the library via the normal API, so it'll match what other apps will be able to do

From where can I get cjxl_ng linux prebuilts?

https://github.com/libjxl/libjxl/actions/workflows/release.yaml might work

That's where I get Windows binaries from

how come there is no djxl_ng in there 🤔

but it does not have ng

djxl was replaced by djxl_ng, but the file sizes are the same so I'm not entirely sure what's happened there

``e 9 -I 1 -E 3 -g 3`` <:KekDog:884736660376535040>

Is there a jxl viewer on android already? Coudn't find one in the play store yet

But to be fair, I didn't look all that much yet

djxlng doesn't wok

also cjxlng didn't neve woked

use only cjxl and firefox nightly

thee's no ipoveent on decoding speed

cjxlng is same as cjxl in file size

and you cannot use the features

as they started changing 4 days ago

Sad

mpv on android might be able to do it if there's a recent enough build

it isn't, and I don't know the answers to those questions

https://twitter.com/js1024fun/status/1547183423239897089 this game is now officially submitted, I'm happy with that

There should be a 0.7 release relatively soon

Does anyone play Go (Board Game) here?

https://online-go.com/

OGS is a good place in case anyone was interested

||ahem...chess but better||

Definitely interested but never had enough time for it. If I had, I'd probably want to study the 9x9 variant. I once read a book about it and it caught my interest. ||and I agree about chess||

https://myanimelist.net/anime/51092/Yurei_Deco this has nice style

good OST too

(continuation of discussion from <#824000991891554375>) When I said "international copyright law is a mess" what I meant was "the international copyright law situation is a mess". The European tradition of *authors' rights* may be compared to the American *copyright* but they are very different in spirit. The authorship law of Germany (and also Poland and probably more European countries) gives all authors some additional moral rights that are deliberately irrevocable and nontransferable. For example: *§42 UrhG.* gives authors the right to retract a work from publication if they don't agree with it anymore. The law doesn't allow for signing this right away under any circumstances to give authors the right to a change of heart (but they have to pay for it). As long as an author has the right to retract a work from publication it can not be public domain. You say `"protecting" American business interest over the interests of its own citizens` but that's not what happens. There is no distinction between countries made. All authors get those moral rights. You seem to imply that Germany should rid American authors of these rights?

How does CC0 help then if the author can still retract the work?

In this case: It doesn't. I was bringing this up to argue why in some jurisdictions works can not truly be published into the public domain. Some moral rights can be signed away in some countries but this has to be done explicitly. CC0 does so. A more practical example would be the introduction of new rights like the recent *ancillary copyright for press publishers*. Regulatory bodies are currently pondering the introduction of regulation regarding the use of copyrighted material in machine learning as well as establishing new rights to machine learning output data. By releasing under CC0 these new rights should also be covered. Creative Commons has some general info about the motivation behind CC0 (see "The Problem"): https://creativecommons.org/share-your-work/public-domain/cc0/ My real point is that a simple statement such as "I hereby release this work into the public domain." may not have the desired effect. But "I hereby license this work to the public under the CC0 1.0 license (https://creativecommons.org/publicdomain/zero/1.0/)." might fit the intention better.

PS: Personally, I think the current legal situation (national and international) is deliberately harming the spread of works and doesn't benefit the majority of authors. In that sense: Ceterum censeo lex authoris esse reformandam. (Copyright should be reformed.)

https://twitter.com/TamithaSkov/status/1548380370038444034

o

> gives all authors some additional moral rights that are deliberately irrevocable and nontransferable This sounds like it actually hurts authors because there's an irremovable risk that licensed work can be unlicensed at will

so it's a major disincentive to commission work

since at any point if the author doesn't like you they can unpublish it

the ability to release your rights to something provides a guarantee and security, something that's important for contracts

Might be only the right to get your name dissociated from the work, not the right to unpublish

I play. I was a supporter for OGS under the old ownership when it was correspondence-only. I liked the tournament feature.

Well it's not that easy. The author has to pay the licensee a fair compensation and at least the licensee's current expenses if they had any. Additionally if the author ever wants to relicense the work to someone else they also have to offer it to the original licensee at a fair price.

*§34 UrhG* also allows for retraction of licenses (but without compensation) by the author if the exercise of the license by the acquirer cannot reasonably be imposed on them in good faith. For example if I lost my family while fleeing from the fictional country of Obristan and later an Obristanian state-controlled company buys the company where I published my book, I may retract the license because I can't bear the thought of Obristan profiting through my work. The thought here is that authors still have some degree of control about what happens with their work because it's still their work. While US copyright can be fully sold and transfered, European style authorship can not. This also manifests in how works created in employment get handled. In the US if an employee creates a work for their employer the copyright is granted to the employer (as in *the company*). In the EU the rights to the work are granted to the employee instead but the employer gets an automatic unlimited and exclusive license to it. Both cases are in practice not that different but there's still a difference in ideology. I've read that in the US companies can be the authors of works. In the EU only humans can be authors. ||By the way, I have a clause in my employment contract that for the code that I write I only grant a non-exclusive license. That's ok because that code is usually licensed freely anyway.|| An example from another country: In Poland authors seem to have the right to renegotiate the licensing fee is the use by the licensee earns more than expected. This made the use when the Polish author of the books behind the the international successful "The Witcher" video games sued the game company for more money. While this might seem unfair for the one who bought the license, it's apparently fair game in Poland. The central idea here seems to be that the use of the work is only by grace of the author and as long as the work is protected by copyright the author has some say in it.

But in practice the use of such authors' rights (license retraction, price renegotiation etc.) happens so rarely that I don't think they play a role in licensing price negotiation.

off-topic of off-topic, but that handle looked too familiar to me

(because Daniel Micay previously worked on Rust)

If at least one of the Vanadium/GrapheneOS project members has the right jurisdiction that could be an interesting court case. (Provided they are willing to bring this to court and have the funds/insurance for it.) But as the project has already declared they won't press it, it probably won't happen.

https://www.phoronix.com/scan.php?page=article&item=rembrandt-windows-linux&num=3 I didn't know libjxl was used by Phoronix for benchmarks now!

Like the version of libjxl

I can't update for every language

Wikipedia forbid that

Or even only the spelling

so, the context is that I actually had been an active user and administrator in Korean Wikipedia, though that's now more than a decade old

People do

So what you know

Koreans though are more strict

while you may not feel like that, generally people do assume good faith

No. The rule applies everywhere

The question is Who will update the version number

but that good faith is because you don't know anything about a particular user at the first glance

I can't edit spanish more

Because of many flaggings

That put my account at risk

My article got deleted because of errors in translation. I'm native speaker of that language.

If i do at mass

as you continue to give a bad signal, regardless of the actual wrongdoing, that alone can be seen as a reason not to assume good faith

So who will fix the GitHub spelling in es Wikipedia

and that's an inconvenient truth of Wikipedia; there are too many bad actors so good actors who are not necessarily good editors can be penalized

Also will happen to italian and spanish translation

an article hoarding is one such bad signal

Ok but answer my 2 questions

bad translation can be also seen as a way to hoard the article, as in "I don't know much about this language but still I want to 'own' this article anyway"

I can't

I should stress that this is how you are seen to others, not how you actually think

Can you explain this part a bit more

Yes i sae

so there are some sort of the first mover effects, the initial article contents can affect the final article more than other edits

That's the part in which is said

and there are plenty of bad actors who want to leverage that in all relevant language editions

which can't be easily distinguished from just overly enthusiastic foreigners

I didnt started the JPEG XL and av1 Italians articles

They were started by bots

And then i added my autistic shits

yes, but first few edits can still set the direction

so my advice is: calm down, not having the best possible JPEG XL article in these languages is not the worst thing

https://it.m.wikipedia.org/w/index.php?title=AOMedia_Video_1&oldid=128050718

Read here

That's what i did

Better don't waste time in writing Wikipedia articles.

2000 characters

20 minutes of time

Then like 160 commits in av1

With also 2 autjors

yeah, one lesser known way to contribute to the Wikipedia is to write your own article *outside of Wikipedia*

because Wikipedia is a tertiary source

But what you think of what i did in this link

Translate it in English

Is spam what i did here

if you can write a great secondary source (that is, summarizing the primary source in the useful form) that's much better than wrestling on the tertiary source

Do people wanted my change

https://it.m.wikipedia.org/w/index.php?title=AOMedia_Video_1&oldid=128050718

What you think of that

Read adozione

Is complete garbage non sense

honestly I don't know Italian...

Or something that has good luck

if you think the article is definitely wrong you can also make a new discussion

Adoption The AV1 codec is considered a next generation codec. Each generation of codecs is bound to reduce the bitrates of the streams by 35%, 50%. It is also focused for screen content as well as including grain synth functionality. 10-bit profile support. Software the decoder used by web browsers for example when playing a video is dav1d developed by Videolan, but much of the code was contributed by TwoOriolies, a proprietary software company. Another decoder included is libgav1, however it is only used for decoding native software AVIF, found as of Android 12 versions. Both decoders have a royalty-free license. However the encoding and decoding of the codec are quite slow and working with a dedicated fixed-point hardware helps parallelization and multi-threading; The encoder that uses Google or Youtube, however in VCU (Hardware) forms, with DeepMind AI probability that can solve chess with machine learning. It is aomenc. The latest version is 3.4.0 svt av1 developed by Intel, Netflix and Cidana Developers prefers VMAF, the latest version is 1.1.0 but we are already at 1.1.0-70. The encoder is suitable for the Meta company (formerly Facebook), which intends to do rav1e from Xiph and some Mozilla \ employeers. The latest version is 0.5.0 mediainfo ffmpeg mpv, vlc media player, handbrake for both svt av1 and hw intel arc (qsv), official support from Intel for both, documented in a spreadsheet. eve-av1 (for a fee), only real-time cisco web-ex but not encoding. Also AV2 is in development but it has been kept secret to make HW implementations and decoder optimization flourish, however according to a Youtube engineer it will be necessary to have multi-thread and desktop configurations and the hardware is necessary but in a video it is specified which are possibly worth no more than $ 50.

the discussion bit is more lenient than the article (in general, I can't guarantee that for all language editions but I don't see why not)

read

is descriptive

sounds neurotypical

add something new and intesting

ok bye

I don't get the context why you keep saying neurotypical, do you mean that the *original* article shows that characteristic?

do i sounds autistic

in this

https://discord.com/channels/794206087879852103/806898911091753051/1000769159593992303

I don't know?

this was the fist addition to wikipedia

is any good

I do see that you have many typos but that's irrelevant

what I can infer is that you seem to rush

in what sense

I think you really, REALLY want to do a particular thing (in this case editing a particular wp article in some way)

yes

i wanted to make like my banned youtube video

and you seem to try everything you can do, which in my opinion is a very bad signal to most WP users

but not like eddit to desxpitive

o benchmay

isn't that a direct opposite to what you actually want to achieve

https://it-m-wikipedia-org.translate.goog/wiki/AOMedia_Video_1?_x_tr_sl=auto&_x_tr_tl=en&_x_tr_hl=it&_x_tr_pto=wapp

that also what i did now

adding many links

on a pc that's horrible to read

if you think you just added many links and got banned (I can't tell whether that's true or not!), please ask around

because administrators *can* make mistakes

no in italian i'm respected

the aticle

probably because you speak perfect Italian and not-so-perfect Spanish?

this isn't spanish

you can appeal to administrators

spanish was jpeg xl only

english nothing

and spanish was automatic tanslation

done with wikipedia without contextualing

i've posted the flags with sceens

if you feel better with that, that's what you should do

why google shows the av1 italian wikipedia

no clue

that's because my edits are respected

because youre italian

or i can have banned wikipedia accounts everytime

before a counts of char it didn't

you spoke good Italian and the resulting article looked fine, in Spanish WP it didn't and that triggered a bad signal, in my view

now that has moretable of contents.

so is this

and why we talked so much

you are korean

the bad signal thing is honestly not really an optimal solution and caused lots of problems, but if you do want to contribute to Wikipedia that's difficult to ignore

I speak from my experience dealing with actual trolls and non-trolls that looked like trolls

what you feel will be the eaction of pc uses when seeing this

https://discord.com/channels/794206087879852103/806898911091753051/1000770404274675752

I tried to give my advice to maximize the probability that you are recognized as a genuine editor

whether to take my advice is up to you

what you feel will be the eaction of pc uses when seeing this https://discord.com/channels/794206087879852103/806898911091753051/1000770404274675752

how they react

also are you korean

yes, but I didn't edit Korean Wikipedia for years

so you said i changed totally the direction of italian av1

how??

with the 2000 char

https://discord.com/channels/794206087879852103/806898911091753051/1000769159593992303

no, I said that the first few edits *can* change the direction of future edits, not always

but a vote?

how is?

rushed

wdym by "a vote"

like is something useful

interesting at least

no

nothing matters

is bad to see for pc

too linking

the 11 was deleted by an user and i readded it

are in english synonymous for re added it

so an opinion on that

like my details and my rushed way can be falsified or totally rewrite from scratch and if some user don't like what i did

or it has to be discussed

why spanish blocked me with no discussion only on jpeg xl arricle

bastards

i think you should step away

https://www.macmillandictionary.com/thesaurus-category/british/ways-of-adding-extra-information

https://music.youtube.com/watch?v=F_p7afMlHVU&feature=share

I mostly agree to your sentiment, I have been an advocate for more decentralized encyclopedias for years

no, but fab said one is a native speaker so I guessed so

So Windows 11 doesn't support legacy hardware, right?

It means Image Coding and Quality, it's one of the subgroups of JPEG

There are USB floppy disk drives you can get today. Presumably they should work like any other external drive as there should be a controller translating the USB storage calls to whatever the drive expects.

Have you ever seen USB 5,25" floppy drive? This is a real 5,25" floppy drive connected directly to the mainboard. 3,5" USB floppy drives are still pretty easy to find though.

Anyway, that's kinda funny they updated both floppy icons, although there's no way to connect 5,25" floppy drive to "officially supported" hardware, and they forgot to update the Bluray icon.

https://lwn.net/Articles/902410/ wow

oh wow

I see the rationale but it seems... far-fetched given that many other licenses other than CC0 do not concern trademarks or patents appropriately anyway?

someone from HN pointed out that if you declared some work is in the public domain and used CC0 as just a safety guard (like me), there is no legal issue in some American relicensing it into 0BSD or MIT-0 (Fedora seems fine with them)

are those two licenses like "public domain with a patent grant"?

I guess if you really want to CC0 something you probably need to include an explicit patent grant.

copyright and patents are two quite different things. You can have a public domain implementation of patented technology (so you can distribute the implementation however you want, but if you use it you may have to pay royalties) and you can have a proprietary, closed-source implementation of a royalty-free technology (so you cannot freely distribute it but you can use it freely).

the classic example of this was LAME, an mp3 encoder that could only be distributed in source form for years

since distributing binaries of it violated mp3 coding patents, but in source it was legally classified as "exchange of ideas" which is protected free speech in the US

in countries like germany without an american free speech protection, software patents are invalid so it didn't matter

nearly all copyright licenses just don't say anything about patents, since those are two different things altogether.

AFAIK only (L)GPLv3 and Apache 2.0 combine a copyright license and a patent grant (with defensive clauses)

that's also the main reason why GPLv3 is not compatible with GPLv2

most FOSS licenses are just a copyright license, ranging from permissive (BSD,MIT) to restrictive copyleft (GPLv2), but they don't say anything (at least not explicitly) about patents

both libaom and libjxl went for a BSD license (which is a permissive FOSS copyright license) plus a separate PATENTS file giving a patent grant with defensive clauses (similar to what is in GPLv3 or Apache 2.0: basically everyone gets a patent grant except patent trolls)

I think it makes a lot of sense for a distribution to aim to avoid distributing patent-encumbered software, but this will not help much for that. You essentially can never know for sure that something is not patent-encumbered. Even if you use a license like Apache 2.0 or GPLv3 that includes a patent grant, it _only_ gives you a patent grant from the patents owned by the authors of the software. There could very well be other patent holders that come and sue you if you use that software.

(the hope is that the defensive clauses will scare patent trolls, but there are no guarantees, see e.g. https://www.sisvel.com/licensing-programs/audio-and-video-coding-decoding/video-coding-platform/license-terms/av1-license-terms

The last Windows version I used was NT 4.0 back in 1998, in a dual boot setup with slackware linux iirc

the preinstalled apps thing and all the "crap" is only a thing on windows home edition

but that's to be expected

just checked a clean install behind me and it doesnt have anything i would want to uninstall immediately

I was using MS Edge for couple of years. It was pretty OK except once a while it wanted me to reset to the "recommended setting" with Bing search and so on. The problem is it got very slow and laggy. Especially when opening multiple tabs, it completely hangs for several seconds. And it's getting worse the longer the OS is running. Restarting the browser doesn't help, the whole system needs to be restarted.

meanwhile firefox

i really liked the old edge (pre chromium)

it was the best pdf reader ive ever used

current edge is not as good but still the best among all browsers

Firefox is not the best but for me it doesn't have such performance issues.

Edge's PDF reader kills my battery life

It uses so much energy for some reason compared to everything else

I get invalidation errors like this in the Edge PDF reader

Selecting some of the text fixes it for the line

I had a different issue with PDF in MS Edge. It was loading some PDFs partially. It loaded a few pages and then stopped.

when I don’t need color management, I use Okular as my PDF reader on Windows

Math equations look blurry and rendered wrong on ms edge. Many people reported that bug, but they did not care to fix it.

https://thephd.dev/c23-is-coming-here-is-what-is-on-the-menu > “Angry” is quite frankly not a strong enough word to describe a process that can make something so necessary spin its wheels for 5 years. It’s absolutely bananas this is how ISO-based, Committee-based work has to be done. To all the other languages eyeing the mantle of C and C++, thinking that an actual under-ISO working group will provide anything to them. > > Do. Not. > > Nothing about ISO or IEC or its various subcommittees incentivizes progress. It incentivizes endless feedback loops, heavy weighted processes, individual burn out, and low return-on-investment. Do anything – literally anything – else with your time.

hey uhh can someone ban and delete all the messages from <@456226577798135808>

Poor duck person, he somehow fell for a Nitro scam <:KekDog:884736660376535040>

<@&807636211489177661>

spam in every channel on this server

Yeah, kicked and deleted most of thr messages

but I don’t understand why his messages weren’t deleted everywhere

the few remaining ones where probably left to give context since without them some other peoples messages responding to them seem out of place

if I was left with the option I would probably delete the scam messages and any messages related to them so they don't feel out of place

It looks like the only remaining message is in <#805414133788180491>.

(which is probably muted by everyone, hence it was probably overlooked)

it seems to be perfectly on-topic for <#805414133788180491>

Done! (I think)

I actually used the "delete messages in the last 7 days" feature of banning, but it didn't delete them all ?!?

<@263309374775230465> if you want to read something more embarassing read that

https://it.weeksnews.org/it/AOMedia_Video_1

2 months ago this was public in wikipedia

and 1000 person saw this garbage

it's embarassing

I see.

I have a feeling I was the only sober person ~~not like I didn't drink alcohol, but stupid ass genes in my family prevent me from getting drunk easily <:ReeCat:806087208678588437>~~

https://github.com/lxgw/LxgwGuiKai/releases new font for reading fast

https://github.com/cssobral2013/Argentum-Sans good update, amazing for notepad, even better than alco sans from kreative square, legible at small sizes, good for windows explorer.

Nice font

at the moment is a copy of that

a bad copy

ah LxgwGuiKai i'll update

i'll try to record

Question. **view at full resolution/100%/Open original**: https://glennmessersmith.com/images/gamma220.png Are the lines... A) Red, green and blue? B) Cyan, magenta and yellow? C) All gray? D) yes.

looks CMY to me, but at the pixel level the left one are horizontal stripes of cyan and red, the second one stripes of magenta and green, the third one yellow and blue, and the final one black and light gray

That's neat trick. I like it.

that's a cool way to check the angle someone is looking at it

I don't see much of a difference when rotating, but it depends on the image rendering I guess. When I open it in a browse and set `image-rendering: pixelated` in CSS, rotating it does indeed switch between RGB+white at steps of 90° and mostly-grayish-CYM+gray at odd angles.

my initial reaction was none / green / none / white; i can see the red and blue ones squinting but they're way less apparent

I see CMY and RGB, wonder if the sharpness on my monitor is slightly off

https://entropymine.com/imageworsener/

Worsener 🤣

<@794205442175402004> You can convert latex to html and then convert that to a normie document

Latex <:NUT:714771929420005426>

Tech

Yes, the main thing is how to maintain as much of the layout as possible... HTML is meant for a reflowing layout. Then again, in the end ISO wants Word only to convert it to NISO STS XML which is then used to produce PDF, EPUB and HTML versions...

> normie document <:kekw:808717074305122316>

In 11:30 minutes I read the 21375 added lines of libjxl tiny

With latest junicode font

Wonder how is like to read the full jxl

I'll send a zip file with jxl

I would like a fonf with esa but it doesn't exist

Where might one find the original latex document?

I am still partial to the idea of converting it to images and giving the stupid dipshits a word document with embedded images for each page

Can tex not be converted directly to sts xml?

Probably, but they want Word, because they don't have a process to ingest sts xml directly, only via Word.

Something I don't see often. PNG on Facebook <:KekDog:884736660376535040> https://www.facebook.com/photo?fbid=760818925356125&set=a.726295875475097

Oh nice, their previous cover was also a PNG <:KekDog:884736660376535040> https://www.facebook.com/photo/?fbid=4264176246949599&set=a.726295875475097

Tf, most of their past covers are <:thinkies:854271204411572236>

should i log stats

i was more concerned about how much effort it would be to add

https://www.youtube.com/watch?v=oPv0smUbdrI

https://gutterpink.bandcamp.com/track/tortoise-dragon-fortune-and-unfortune

https://twitter.com/jerrynixon/status/1561879544247762944?t=OJekt1NMDpm3l2GeSqpx2g&s=19

needs more EBCDIC

https://www.instagram.com/p/ChYmFcPrxmF/?igshid=YmMyMTA2M2Y=

https://github.com/fabiorug/Coprica-font/releases/tag/1.0

New font for you

Latest i installed was gulax

But i didnt enjoyed it

Lol on 14th august i got 34 cloners

I'm happy you liked this font

~~wait. Is that your font?~~ nvm.

I pinned some things on GitHub

There is also an onlinewebfonts encoded Suduwe font

Fork of los Angeleno Sans

Called Los Angeleno Grotesquw

The original is called use only for personal use

And is on da font

Not dafontfree

Ok.

As I said on AV1 Discord i'm gonna switch to

https://github.com/silnrsi/font-daibannasil

When 5.0 Is out

That looks like a serif font.

https://github.com/fabiorug/allfonts

https://sturmen.github.io/posts/hdr-jpeg-xl-2022/ This page crashes Android chrome.

Linux image viewers don't support hdr jxl images yet.

And some of them don't have any color management yet.

I assume they don't support hdr images period...

mpv can at least tonemap hdr video to sdr

https://redd.it/wy7405 Is mainstream opinion for new image fomats really that low?

I cannot edit this image with my outdated software. So it must be trash format!!!

I've had completely non-technical friends complain about it as well, issue is sites using it as a general JPEG replacement instead of just for thumbnails or the like

Another trick is to open the image in a new tab, and instead of right clicking on the image itself to download, right click on the background

or right click and copy

most UIs will copy to clipboard as PNGs

at least that's how X11 and Windows both work

a PNG in BMP/dib

or bitmap in BMP dib

I always couldn't understand the reason why BMP has PNG and JPEG embedded, is it kind of an adapter?

I remember when I came across this piece and thought for ~20 seconds “hm, maybe I could try to learn this” https://youtu.be/pHlqEvAwdVc

and then “… yeah, no”

From what I understand, BMP is mostly just a loosely documented version of various internal windows graphics APIs.

As far as I know, PNG or JPEG in Windows bitmap structure can't be used in BMP files. It works only in memory.

Windows handles bitmaps in curious way. BitmapFileHeader structure is present only in BMP files on disk. Bitmaps in memory, bitmaps embedded as resource data in executables etc. don't have file headers.

ICO files consist of bunch of bitmaps, those don't have file headers as well. Since Windows Vista, ICO may include PNG images too, however those are regular PNGs just embedded in ICO.

the windows clipboard requires a dib header

There are two headers in BMP. There is a file header followed by DIB header. The file header is present only in actual BMP files. The DIB header is always present because it's needed to interpret the bitmap data.

When Hamelin came across that piece, he though "why not play two other Chopin études at the same time.." https://www.youtube.com/watch?v=OhgOh8mmefQ

https://twitter.com/tim_language/status/1563918931747676162?t=Sj6nxnLND5ifwMCvPnnkpw&s=19

i saw this on reddit just now. disappointment..

& <@224363555074342912> You'd be surprised. I monitor Twitter for mentions of the various formats. Many average users are very angry when they encounter a WEBP or AVIF as there is little to no OS support. They can't preview it, sometimes have no idea how to open it, can't right click and change metadata, etc. They are valid issues. Even Windows 11 doesn't know what a webp is out of the box. You need a third party plugin to get it (at least partially) working. They just want things to work just like they do with JPG and PNG. Once JXL comes out, I will do my best to help people navigate it until the OSes add native support, but people will still be mad they have to do more than lift a finger.

There's installable WebP and AVIF plugins, but they're a bit odd

The AVIF one is actually a generic HEIF plugin, that then depends on the HEVC/AV1 decoders (Which are also separate) to actually function

WebP is supported since Win10, AVIF is partially supported too, it's limited to 8bit or 4:2:0.

There's still a 4:2:0 limit, but it supports 8/10/12 bit images now

Might be a licensing thing? It uses dav1d and I don't know what the devs would think of it being included with windows

given dav1d is BSD-2, I doubt that licensing is a problem (and IIRC - but IANAL - the devs don't have much say :P)

heh, should have checked that before guessing

The Photo App could be smart enough to suggest downloading the plugin from Windows Store.

The video app is at least

Nah, that's all on the many HEVC patent pools

They probably got money from me anyway from my GPU

HEIF container, AV1 and WebP are downloadable for free.

In the case of HEVC, I guess, MS took the easiest path.

also x265 is an encoder

what do people use for hevc decoding?

at the same time x265 isnt free free

like how it's not automatically in ffmpeg

Many GPUs already can decode HEVC.

i think the main problem is it being a legal minefield

The HEIF container itself is not free of risk. Nokia claims patents on it but doesn't offer clear licensing terms last time I checked. Apple clearly sorted it out somehow, but I dunno what Microsoft's opinion on it is.

another example is firefox only supporting h264 main profile since it uses openh264

openh264 is only used for webrtc, playback uses the main OS codecs, which are also just quite limited

iirc they hardcode the profiles you can use

I've seen an MP4 file used as a small silent, looping clip on a webpage, so I guess animated AVIF is not even needed. A simple AV1 video clip can be used instead.

Yes. Just Chrome refuses to allow video containers in an img tag for some reason

Safari allows mp4 in an img tag for a while already

So let's use video element then.

Maybe for webrtc, but I'm not sure about that for normal video. Like the decoder on Windows can only do 8bit 4:2:0 high profile at best, not much to actually limit there

yes, but that's not always an option. not everyone has control over the html generation.

the trouble of having to deal with two different code paths (one for still, one for anim) is probably one of the major reasons why GIF is still a thing

but can we put images in video tag

to me the major difference between animated images and video is a presence of controls and audio. so I believe it should be equally possible to put video to img or image to video, since it's more about the user interface than formats themselves.

Websites have to check file types, so users won't be uploading nonsense, so the sites can generate two different HTML codes as as well.

There are many videos without controls, look at ads, for example.

Allowing video files in img element might be useful, but using video files exclusively in video element seems to be a cleaner solution.

One issue I saw raised is that a playing <video> can cause the browser to pause the screensaver/display sleep, but that's not the case with <img>

<video> is mostly interactive, while <img> isn't

deprecating GIF / APNG / animated WebP in an img tag is not going to happen, even if it would be 'cleaner' to have only still images in an img tag

so if we accept 'animated images' anyway, then it would make most sense to also allow video, just treated in the same way as GIF: muted, autostart, looping, no playback controls.

That's fair. But what about other places, where images can be used, like background image of an element?

Those behave similarly to img.

there too, if you can use GIF right now, then why not allow a format that actually compresses?

But you wouldn't allow GIF in video element?

I'm wondering what's the logic behind assigning the file formats to different elements. I suppose it's based on mime type.

You know, GIF is always image/gif, PNG is always image/png, but MP4 is either video/mp4 or audio/mp4