JXL support is enabled in Chrome 155 by default, JXL on 4chan when?
This place still blocks AV1 despite every single current major browser seamlessly supporting it.
>>109998195It might get support on umigalaxy. It has support for WEBP at the moment.
4chan only implements obsolete formats.
>>109999122Your retarded .webm don't even load on my computer.
>>109998195Right after WEBP support
>>109999132Sigh, is this another file vs data format lecture at hand?
>>109999332we can skip this one
Image format technology peaked with Animated PNG.
>>109999837Why? Webp seems to have the lowest decode of all modern image formats so isn't it still good for poorfags?
>>109999837What's wrong with that? Wouldn't it be nice to no longer have to transcode all those WebP images you find on the web?
>>110001236No HDR
>>110001380You think poorfags can afford HDR displays?
>>110001387Considering every phone made in the last 10 years, from Klarma-invoking flagships to burner shitter budgets have HDR-capable screens, yes
>>110001399>The Samsung Galaxy A15 5G does not have an HDR-capable display.>It has a 6.5-inch FHD+ (1080×2340) Super AMOLED panel with a 90 Hz refresh rate and up to 800 nits peak brightness.>Samsung does not list HDR support in its display specifications, and independent specifications confirm HDR support: No.https://nanoreview.net/en/phone/samsung-galaxy-a15-5g?utm_source=chatgpt.com
I'm still on 149 for Ublock
Webp on 4chan when?>b-b-b-bu-bu-but-but-but i-ii'm SCAAARED of that file format!!!! im super super scared because i've never read a whitepaper in my life and someone told me it contains spooky black magic and google is gonna rape me and found out what i masturbate to!!!!
>>110001380>>110001387I have an HDR display (OLED) and I don't even use that "feature". HDR is fucking retarded and unnecessary.
>>110001420>>110001387>implying normies would still be using a 3 year old beater phoneConsidering how retarded American spending habits are, pretty much everyone has some form of HDR display. So why would we stick to a format that can't take advantage of these displays? To shave 1/50th a second off decode time? Which realistically occurs off screen anyway?>>110001816What a shame, I love HDR. It's like listening to music in stereo, instead of mono.
>>110002038AFAIK, zognald still hasn't given everyone $5,000* 10% interest loan yet*must be a working adult and be donating at least $100/month to Israel
>>110001380The worst fucking thing is scrolling some brainrot social media in the evening before bed and getting my eyes blasted out because someone uploaded their fucking HDR iPhone video
>>110002082>trannie seethes about Trump apropos of nothingYou lost, tranny.
>>109998564Might be more of a security concern to write an imagemagick condom.https://gitgud.io/zuse/secmagickH*ro cannot code
>>110002146What phone? Because mine barely gets brighter if I'm in a dark room. I have to crank the brightness really high to get any HDR headroom (confirmed via developer options)Also serves you right for browsing brainrot before bed, if the story even is true.
>>110002038>I love HDR. It's like listening to music in stereo, instead of mono.Nice cringepost. I got some platinum HDMI cables to sell you to really get that warm picture you're looking for.
>>109998195ur a JXL
>>109998195Avif won.
>>110002653Weird interpretation but okay, you enjoy your bright objects and specular highlights having the intensity of a word document, and I'll enjoy those being actually somewhat bright.
>>110000000
>>110003233>>110002653It doesn't even matter if one hates or likes HDR. The shift away from the 8-bit status quo is one of great significance. WebP is standing in the way of this.
>>110001380>HDR>to post wojak memes
suck my dick
>>110006545why tho?
>>109998195Fuck yeahFinally a non-slop format is getting support.
>>110007449You mean the one that had to have its decoder rebuilt from the ground up because it was slow as fucking balls to decode on 90% of CPUs and had like 6 millions lines of code that would have taken 100 years to properly audit for security vulnerabilities?
>>110007763>new format's basic reference decoder was slow and unoptimized...so? this is literally always the case with every new format, av1/avif was unusable before dav1d came years later, there's nothing really inherently wrong with that.wait till you learn that half of jpeg-xl's toolset hasn't even been implemented at all in libjxl yetit's a format meant to be as future proof as jpeg was, none of these "problems" will be of any actual relevance 10 years from now.
>>110007834JXL was 100% SLOP and you can't prove otherwise.I could make 100 different co.parisons where WebP decodes 5-10X faster and is 80-90% close of JXL compression.But you would still call it "non-slop", wouldn't you?
>>110007893>JXL was 100% SLOP and you can't prove otherwise.you are 100% A NIGGER and you can't prove otherwise.>I could make 100 different co.parisons where WebP decodes 5-10X faster and is 80-90% close of JXL compression.with quite a bit of cherry picking? sure, but what would that prove?btw yeah you really suck at text comprehension, just discarded my entire argument
>>110007834Provide JXL wins, that is. Which it might not, considering that it's avif that's being developed from Google's (and everyone else's) money and it's literally up to google and nobody else whether or not they want to make it a thing.
>>110008043this makes absolutely no sense1. google literally developed most of jpeg-xl, they have probably contributed as much to avif (people think aom is just google around here for some reason), they don't have any reason to "make their format win", especially when both have the same level of support and both are fully open source/royalty free.2. avif encoders/decoders have reached the limits of what is possible with the bitstream (and that just about competes with jxl in its infancy with barely any feature properly implemented), the only way to improve avif is to introduce a new format which uses av2 as a base, and even then it's highly improbable it can cover the vast amounts of use cases that jpeg xl can, and will it be able to keep improving over time like jpeg xl will? probably not. we already know an av2 based avif will be at best 20-30% more efficient, jpeg xl with half its toolset still sitting unused and the other half still not properly implemented can easily tie that gap in the next decade, once again, it's meant to age well on the long term (that's literally what the L in xl stands for) and not necessarily impress now.besides, jpeg xl already found use as a shooting format in smartphones and is part of the new pdf spec, avif might "win" the web and that's about it... the complexity you're complaining about is exactly what makes jpeg xl stand out and what gives it the possibility to gain market share outside of web use.
Did they solve the extension problem? There's no way to manage my 30.000 hentai collection without extensions. No, I'm not going to use metadata or go into every damn file to figure out what it's about, and no, the size doesn't tell you anything about whether it's lossless, lossy, or animated. Put a damn extension on this shit so I can organize this crap in Explorer.
>>109998195Does it work on Windows 7, Photoshop CS6, and Firefox 56?
>>110008612It's time to let Firefox 56 go.It doesn't even support WebP ffs.
firefox doesn't support it
>>110008612wondows 7 and whatever old firefox version you use are completely irrelevant because they're not suitable for normal work in 2k26 due to security and compatibility stuff.CS6 holds up really well and is perfectly usable , but it'll never get jxl support. It did get webp though so maybe it's possible.
>>110008895Sup fixLIVES, I wanted to try to shill you the umigalaxy altchan with WebP (both lossless and lossy btw) again. Void, the owner there has been redoing the site and fixing more performance problems and changing the UI to a more familiar imageboard style. Hope you give it another try, I'd be cool if you could join us there.
>>110008043>>110008226AVIF definitely has a huge head start because of the hardware decoding thing. Since this thing really is just a still AV1 keyframe you don't need any snowflake custom hardware decoder, you can just use the one for AV1.JXL doesn't have this and it's going to be particularly difficult to gain momentum when giant 100 megapixel AVIF images are being decoded in like 100 milliseconds and using up 0.001 watt of power to do so on phones that have said hardware decoding for it.
>>109998195When xitter allows JXL because this site is moderated by xitter faggots
Daiz lost.
>>110009016Who is this bald cunt anyway? You keep mentioning him as if we all know who this knob polisher is.
>>110008987>>110008942>>110007893you lost daiz aka pixdaiz, ss2, jonsneeders, fixlives, https://desuarchive.org/g/thread/108742889/#108746907
>>110009055What about you? Would you like to peruse the new and improved umigalaxy altchan? Void has actually been listening to user feedback there and I don't find it impossible for him to add JXL now that both Chrome and Firefox support it DESU.
I'm serious you nerds, what's the point of these threads when you all know full well that due to the ape's continued dictatorship you're not going to get JXL, you're not going to get AV1, you're not going to get AVIF, you're not going to get WebP.Basically you're not going to get shit. I don't know if umigalaxy is THE best place for all of us to move to but to me it feels like a promising option DESU. WebPfags get to tinker around today and JXLfags now have hope for a place that might give them support.
>>109999122Then why can't I upload pcx?
>>110009565Wait your turn man, the ape is slaving over a hot stove as we speak cooking up support for jpeg2000.
>>110001236webp is a garbage format, and it's just one of many Googles attempts at taking over the internet.jxl allows for lossless recompression of lossy jpgs (ie. take a lossy jpg, turn it to jxl, image data stays 100% the same but file size is lower). this alone makes it deprecate webp instantly.
>>110001810webp didn't even had a proper whitepaper when it came out, they just got the source code of the reference encoder and said "this is the spec".it's an abomination of a format. it's not even an image encoder, it's just the vp8 video encoder making single keyframes.
>>110009624>>110009603Theo actually made a video about WEBP and despite all the downsides, it achieves a compromise because of low CPU usage.Both AVIF and JXL are much better but they come at the cost of higher CPU usage.https://www.youtube.com/watch?v=pFNmS_HZ2ZsIf you're gonna shit on WEBP at least be informed about it. Shill JPEGLI instead, it's a modern JPG encoder project that achieves similar compression to WEBP.If I were a WEBP shill then the last thing I would ever want is for people to find out about JPEGLI.https://opensource.googleblog.com/2024/04/introducing-jpegli-new-jpeg-coding-library.html
>>109998195use case?
>>110009868You get to DDOS poorfags by sending them a link to a 100 megapixel JXL now.
> This lead to widespread condemnation and the development of a royalty-free re- placement format called PNG [51] — officially “PortableNetwork Graphic”, but it is pronounced “ping” and also considered to be a recursive acronym meaning “PING Is Not GIF”.who the fuck calls png ping and not pee-en-gee
Is "attack surface" really that big a deal for codecs and formats that are so well-established? For instance, is there realistically any chance someone could manage to put something in a jxl file that could cause the thumbnail generator or something to execute code?
>>109998195Never, just like on most other websites
>>110010875Yes, it's actually a pretty big deal with everything, and there have been examples of vulnerabilities. There was one in webp too. Or remember when 4channel got hacked? That was because of a vulnerability in a PDF library for generating thumbnails.It's a constant cat and mouse game. If you have a big C or C++ program, it WILL have vulnerabilities. You only need 1 to compromise a whole system. Now AI makes finding exploitable stuff like that take weeks to months instead of years to even decades. A huge advantage of Rust is that it eliminates the most common problem that causes this, but it isn't a magic bullet.
Why is Daiz allowed to use the internet when he's mentally ill? Note that he's posted this image over 5000 times, 4800 times with a different hash and image dimensions. Why does he flip flop being a shill for WebPiss and Tel Avif? https://desuarchive.org/_/search/boards/g.desu.meta/width/1488/height/1488/
>>110008897It does, you have to turn on a flag. I think if will be on by default on the next update.
>>110009868So I can post more lossless images here, and skip having to convert them to lossy jpegs.
>>110001816>I have an HDR display>OLEDLol
>>110004081https://files.catbox.moe/dvf6jq.jxl
>>110011839nigga no one wants to see your repost of a repost low quality encoded shit anyway, post something original
>>110012231I used to, but posting on Twitter is more profitable.
>>110007893The slow decoding speed doesn't really matter. In practice JXL is fast enough and you won't notice a difference while browsing the web. Decoding takes around 300ms for 8K x 8K images.
>>110014050It doesn't matter if you have an iPhone 20 max pro cocksucking edition(tm).However for everybody else who isn't a fucking retard with their money it's going to be a huge pain in the ass. It definitely has room to improve but as of right now JXL is nowhere near ready for gaining any adoption.
Thread related, I asked void if JXL is something he will add.>>110015074>>110015074>>110015074>>110015074
>>110001236>>110001361Lossy webp was already obsoleted by HEIF even for muh hardware decoding spastics and lossless webp is just inferior lossless JXL.
That's cool, but I'm still uploading my files as PNGs just to take more space on your servers.
>>110017252AVIF uses the HEIF container.
>>110017252>>110017505I don't think I'm getting my point across. New codecs = better compression, right? but new codecs = require more CPU resources, right?If you want the lowest CPU resources possible obviously it's going to be JPG but it offers the most awful compression. Maybe this 5 star ranking thing can get my point across:JPG:__CPU decode * * * * * Compression * - - - -WEBP: CPU decode * * * - - Compression * * - - -AVIF:_ CPU decode * * - - - Compression * * * - -JXL:__ CPU decode * * - - - Compression * * * * *Basically WEBP sits in this goldilocks zone, especially for poorfags. It's dogshit at compression relative to other newer image formats but not compared to JPG and is nearly as decode friendly as JPG on low end hardware.
>>109998195source?
>>110001236>forced 4:2:0webp was obsolete the day it released
>>110009851>low CPU usage.literally not a problem.
Time was, you downloaded an image from a website and it was .jpg or .png or if it was animated it was a .gifthen webp started showing upthen videos replaced gifsthen avif started showing upnow people want jxlfuck it bring back bitmaps
>>110017799>-sharp_yuv>Use more accurate and sharper RGB->YUV conversion. Note that this process is slower than the default 'fast' RGB->YUV conversion.https://developers.google.com/speed/webp/docs/cwebpAlso 4:2:0 isn't a problem on retina displays, a giant computer monitor sure. On an M4 MacBook Air with a retina display, not really.
>>109998195Webp meme faggot anon will be seething. JXL CHADS STAY WINNING. GOOGLE BTFO
>>109998195>>109998564>>109999332too many unpatched exploits
>>110017931jpeg, jxl and avif all support 4:4:4, 4:2:2 and 4:2:0. Also 4:4:4 is the default on all of themgenuine fuck you for supporting this dogshit format
>>110018026>Also 4:2:0 isn't a problem on retina displays, a giant computer monitor sure. On an M4 MacBook Air with a retina display, not really.
>>110018077yea dude, I know you're a lost cause
>>110018105Do you have any proof that 4:2:0 images are a problem, specifically on devices with retina displays?AFAIK 4:2:0 is terrible on giant computer monitors but not on devices with retina displays.Prove me wrong else picrel.
>>110017931>accurate and sharper RGB->YUV conversion>4:2:0 isn't a problem on retina displaysAre all webp shills this retarded?
>>110018077You are genuinely retarded.
>>110018754I can tell when a screenshot was taken from a video because every video on the web is 4:2:0 and 4:2:0 looks "blurry" compared to 4:4:4that guy has a shit display if he can't see the difference
>>110018762>>110018802Do you retards really not know what a retina display is?>the absolute state of /g/
>>110017799This is one of my biggest gripes with it. It is a designated shitting format.
>>110019063see >>110017931>inb4 your money ass also doesn't know what a retina screen is
>>110017684>I don't think I'm getting my point across. New codecs = better compression, right? but new codecs = require more CPU resources, right?Wrong. We have hardware video decoders and new formats use video iframes for stills to leverage this faster execution path. Phones and low power PCs even have hardware JPG. Smaller data set + fast decode path = less resource use, even if the hardware accelerated decoder is a little more complex. It still beats keeping the entire input path lit up longer for a larger input set.
>>110023026I've heard about this but this is only possible on newer phones/computers (for webp/avif/jxl/whatever), right? Everyone else, say a /g/ loser living in their mother's basement still using a Thinkpad T420 has to decode everything with his throttling/overheating CPU.I mean shit man, there must be a reason why google shoveled so much energy into jpegli. What other deranged reason would there be to squeeze out ever ounce of compression efficiency out of an ancient lossy image codec from the 90s?https://opensource.googleblog.com/2024/04/introducing-jpegli-new-jpeg-coding-library.html
>>110023081No it's pretty much every phone with a touch screen. They give free phones to homeless people with hardware AV1. Anything new enough to be on 4G has hardware webp minimum.
>>110023557So why did google develop jpegli then?
>>110023734Why did they spend 10 figures making CunnyAI? So they could hire people to do it who wouldn't be working for competitors most likely.
Losless JXL is magic
>>110024040It's sometimes worse than lossless webp. Lossless webp is basically always smaller than png, but lossless jxl is sometimes bigger than the original png, which is fucking retarded.
>>110024063Try using effort=11
>>110024063>>110024119It takes forever, but it's worth itcjxl image.png --distance=0 --effort=11 --allow_expert_options out.jxl
>>110024063Over a fairly huge sample set of doujins lossless JXL is about 2% better than webp on average. The magic part is in animation support and non-integer color spaces.
>>110024234That's incredibly dogshit when you take decoding into accountSee >>110014102
>>110024537Nobody cares about CPU decoding speed
>>110024119>>110024183>dude just do this retarded thing that's not viable in real-life scenariosAnd let me guess, also install le gentoo xD
>>110024234That's literally worse than webp then, because webp is a lot easier to encode and works on everything anyway. And it's still probably less flexible than webp for lossless if you want to avoid your images getting bigger.It's kind of a shame. I like jxl for lossy. Still a lot better than avif lossless, which is so bad kt might as well not even exist.How is the animation support? Is it better than apng or is it half assed?
why do you want to share 4k backgrounds on 4chan?I swear to god the people who aren't satisfied with the current file size limit, jog, and png support are legitimately just braindead zombies. idk maybe its a cult thing, newer filetypes cause more CPU overhead, they want 4chan to be slower and worse for people on cheap hardware
>>110024806>And it's still probably less flexible than webp for lossless if you want to avoid your images getting bigger.Idk why you would think that. There's always a possibility with lossless schemes that you have negative results, but a simple output sanity check can control that.>How is the animation support?It's the best frame by frame format. Differential encoding, lossy or lossless, any color space.
>>110024692>>110024870look at >>110014102 again because there's 2 JXL decoders being testedA) libjxl: a complete piece of shitB) jxl-rs: not as dogshit but still terrible compared to djpegliIn fact wpd (the new rust WebP decoder) is a huge fucking deal because it will allow for even faster WebP decode speed on low end/shit devices out there.https://github.com/halidecx/wpdIf nobody gave a fucking fuck about decode then wpd would have never come into existence. Anyway we might not like it but WebP is just going to keep on getting more popular because of this.
>>110024063I always get 40 to 60% file size reduction when converting png to lossless jxlyou're doing something wrong
>>110024692correctthe time to transfer an image through network is a bigger bottleneck than decoding time
>>110024996If that's true then why was FLIF a completely and total fucking failure? It outperformed Webp lossy and Webp lossless so why aren't we using it today?
