[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / r / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/g/ - Technology

Name
Options
Comment
Verification
4chan Pass users can bypass this verification. [Learn More] [Login]
File
  • Please read the Rules and FAQ before posting.
  • You may highlight syntax and preserve whitespace by using [code] tags.

08/21/20New boards added: /vrpg/, /vmg/, /vst/ and /vm/
05/04/17New trial board added: /bant/ - International/Random
10/04/16New board for 4chan Pass users: /vip/ - Very Important Posts
[Hide] [Show All]


🎉 Happy Birthday 4chan! 🎉


[Advertise on 4chan]


File: 1758165387672.png (66 KB, 1258x409)
66 KB
66 KB PNG
I don't mean to start a conspiracy theory here but is it possible that 4chan's dev team members have shit for brains?

How can these technical merits be ignored?
>dual encoding modes (lossy and lossless)
>alpha transparency in both lossy and lossless modes
>20-30% lower file size for similar quality compared to older JPG encoders still used today
>lossy encoding mode restricted to 4:2:0 ensuring 50% less RAM is used compared to 4:4:4 JPG
>supports animations also in dual encoding modes (lossy and lossless)
>operates in both RGB and YUV, no weird in-between conversions required
>90% of web browsers, image editors, image viewer support
>Open source and royalty free
>VERY GOOD replacement for GIF
>GOOD replacement for PNG
>OK replacement for JPG (cannot operate in 4:4:4 in lossy mode)
>>
>>106824389
That's like buying music in aac
lmao
>>
webp doesnt work on windows, which means its broken freetard garbage and not worth anyone's time

simple as
>>
File: 1756244405120908.jpg (81 KB, 960x717)
81 KB
81 KB JPG
I forgot to mention:
>20-30% lower file size compared to PNG
>50-80% lower file size compared to GIF

The part about replacing GIF is what I find so mind boggling desu. Normally with GIF the only way to reduce filesize is by reducing color accuracy (ie use 64 colors instead of 256) which looks like ass and still takes up huge amounts of space. With webp you just make the animations slightly more blurry and still end up saving 50-80% in filesize. The bandwidth savings from this alone are pretty wild since GIFs tend to be 5-10 MB on the webm.
>>
>>106824425
So we should ignore that webp does work with 90% of web browsers, image editors, image viewers because...???
>>
>>106824474
webp cant even double click
linux ahh format
>>
>>106824463
we already have webm
>>
>>106824591
Webm doesn't operate in RGB and more importantly does not support alpha transparency, Webp does.

Therefore Webm = half baked replacement for GIF
Webp = full complete replacement for GIF
>>
no thanks nigger
we already got mp4 solely because this retarded wanted easier access for twitterposters to copy paste shit from twitter here
>>
>>106824474
webp works in all major browsers
you have over 95% global coverage. stop making stuff up
https://caniuse.com/webp
>>
File: webp.png (3.54 MB, 3008x2000)
3.54 MB
3.54 MB PNG
I've noticed something really interesting about lossy webp even with photographic images: at medium quality it's often able to create files smaller than what would normally be possible with JPG. picrel is a ~80KB Webp converted to a PNG, which scored 59.6 SSimulacra2 compared to the original JPG which makes it officially medium quality.

>Scoring
>The score that SSIMULACRA 2 outputs is simple: a number in range -inf..100. According to the developers of the metric, for image quality assessment, SSIMULACRA 2 scores correlate to subjective visual quality as follows:

>Very high quality: 90 and above
>High quality: 70 to 90
Medium quality: 50 to 70
>Low quality: Below 50

https://wiki.x266.mov/docs/metrics/SSIMULACRA2

catbox:
https://files.catbox.moe/lbcizc.webp
>>
File: dogshit.jpg (99 KB, 3008x2000)
99 KB
99 KB JPG
The smallest JPG file size I got is 100 KB and looks extremely awful.
>>
File: test.jpg (1.86 MB, 3008x2000)
1.86 MB
1.86 MB JPG
Picrel is the original source which itself is already a JPG. So we're talking about a 95% file size reduction when going from JPG to webp, yet somehow you get medium quality and not low quality which is kind of batshit insane when you think about it.

Like I know both avif and jpeg xl would do even better but webp is still pretty impressive imho.
>>
>>106824970
wow, wtf, i can barely notice a diff
>>
>>106824947
>>106824938
cool, but i have a better connection than a 56k modem.
We are living at a time with multiple MB javascript blobs, so whining about 100kB imaged is idiotic.

Most images on the web are PNGs because nobody wants compression artefacts. No matter how small they are.
>>
>>106824970
>the original source which itself is already a JPG
alright, so you compressed lossy to lossy and lost quality

Quality loss is not acceptable.
>>
>>106824983
the original jpeg is 1860kb which he reduced to 80kb, that's a -1780kb savings, not "wining about 100kb". learn to read
>>
File: a462YY4B_700w_0.jpg (55 KB, 648x960)
55 KB
55 KB JPG
>>106824972
It'll depend on what your screen resolution/size is but yeah the memes that webp is only good for low quality brackets is objective/subjectively false.

>>106824983
Please tell ebay that. All the images I upload look like absolute fucking dogshit.
>>
>>106825000
But the original JPEG is the highest quality one. So i will take it.
>>
It's cursed
>>
>>106824999
That's why I used Ssimulacra2 to make it acceptable. Or are you saying that scores between 50 and 70 should not be labeled as medium quality?
>>
>>106825001
I dont save images from ebay. I don't distribute images over ebay.

Arguing for a quality loss is stupid.
Justifying it by "BUT BUT MUH OTHERS RECOMPRESS AND DO IT HARDER" is dishonest.

(i didnt even check if your claim is true, you might be talking about fucking thumbnails, with the original being available)
>>
>>106825015
>to make it acceptable
Quality loss is not acceptible. You lost quality.
>>
Looks like lossless webp is about 20% smaller than png, which is nice and all but png has better support, maybe if a format comes along that offers better compression but webp doesn't offer anything meaningful, for larger animations just go with video
>>
>>106824389
>is it possible that 4chan's dev team members have shit for brains?
Is this a rhetorical question? They can't update the software dependencies a single time in 10 years and you think "The Team" cares about anything but standing in the way of more capable anons until this place rots away?
>>
>>106824389
Not only it's possible, it's true. If you look under the dome there's WYSIWYG. And what you see is shit, so that is what you get
>>
File: 1759607099452063.jpg (133 KB, 1024x1024)
133 KB
133 KB JPG
>>106825022
>>106825024
Lossy is a necessary evil because bandwidth isn't free, it cost actual real money in real life. The real practical question is what quality brackets should be prioritized now that we have better image codecs. At the same time we've never had a good way to accurately gauge these quality brackets and now we do (ie ssimulacra2).

60/medium quality seems reasonable for the web but I understand if you disagree with this since it's a subjective thing.
>>
Another reason why I think people don't like webp is because of the name and extension name, a three letter name would have been more marketable

Webp makes it sound like someone is pissing over the web
>>
>>106825072
Cool, what if I want lossless RGB video with alpha transparency?

Should I just save 6 million individual PNG files instead?
>>
>>106825107
>Lossy is a necessary evil
...for thumbnails

If you only talk about thumbnails, then say that.
And please argue why we should have ten different image formats specifically for thumbnails on the web.
>bandwidth isnt free
It is getting cheaper every year.
We aren't on 56k modems anymore, so why should we compress even harder than 25 years ago?

Losing quality is bad. Lossy to lossy reencoding is completely unacceptable. This is non negotiable.
>>
>>106825124
Usecase?
>>
>>106825133
No stupid, I mean bandwidth as in the resources required to redistribute your image to many people at the same time.

80KB shared between 100 people = 8MB
2MB shared between 100 people = 200MB

Now scale this up to 1k users, 100k users, and potentially even 1 million users. Are you starting to see the problem here?
>>
>>106825134
VFX/CG/CAD/Blender and so on and so on.

>inb4 just shit it up with 4:2:0
nah, I reuse this stuff
>>
>>106825124
JPG2000
>>
>>106824425
>webp doesnt work on windows,
it does. why the fuck are you lying.
>>
>>106824655
>transparency
Nobody gives a shit about transparency. It can even be an issue depending on the theme and the dumb transparency being used.
>>
>>106825150
Use the format professionals use: JPEG2000
It is what digital cinema uses and can do lossless.
Or do you think studios share movies in webp or webm?

You are creating a problem where there is none. Professional applications all have their respective lossless formats already.
>>
>>106825133
Images are also becoming higher resolution as well. Lossy compression might be redundant for a DVD screenshot but a UHD blu-ray screenshot kinda needs to be lossy to share it on the web desu.
>>
>>106825142
And 25 years ago, when we had 56k modems, nobody served images on websites?

Quality loss is not acceptable.
>>
>>106825176
if a website doesn't allow you to post in original quality, that website is unusable for sharing images

NOTHING should ever be lossy-to-lossy reencoded, unless its a thumbnail.
And i dont give a fuck about thumbnails and you are pathetic for spending years of your life arguing about thumbnail-formats on 4chan.
>>
>>106825196
>no true scottsman fallacy
>>
>>106825196
You think ssimulacra2 60 which lines up with subjectively medium quality brackets is actually thumbnail quality?

Alright so what ssimulacra2 score do you think should be medium quality then?
>>
>>106825196
>NOTHING should ever be lossy-to-lossy reencoded
bluesky-
>AHAHAHAH YOU USE BLUESKY?
i just follow artists, man. anyways, bluesky does some funky shit to images that often causes them to balloon in size. maybe they're using jpegtran for JPEG images but aren't checking if the new image is smaller than the original idk.
>>
>>106825205
You don't even know what that term means.

>we should not accept a regression of technology
is not a no true scottsman fallacy
>if i cant share an image on a website without losing quality, it is unsuitable for sharing images
isn't one either
>>
>>106825206
>subjecticely
It is objectively losing quality in the lossy-to-lossy reencoding usecase you shill.

And it loses quality EVERY SINGLE TIME this happens.
So if you do it often enough, it becomes subjecticely lost quality as well.

Thats why it losing quality really is unaccaptable and why it is non negotiable.
>>
>>106825215
Bluesky is a platform for child porn and leftoid political drivel.
If you use it to share images, you are an idiot.

Meanwhile 4chan is an imageboard, image is in its name.
And Discord is private chat app that allows you to share files. This would be unusable if it fucks around with the files. So it offers an original-file download, even when its a 10 MB monster of an image.
>>
File: 1758995395821z.jpg (86 KB, 719x957)
86 KB
86 KB JPG
>>106825235
No you tard, I mean the official medium image quality qualification in Ssimulacra 2 is a score between 50 and 70, so ideally around 60. Are you saying that 60 is too low? Are you saying 70 is more ideal?

Or are you one of those fucking nuts that believes anything below Ssimulacra2 90 = thumbnail quality?

>Scoring
>The score that SSIMULACRA 2 outputs is simple: a number in range -inf..100. According to the developers of the metric, for image quality assessment, SSIMULACRA 2 scores correlate to subjective visual quality as follows:

>Very high quality: 90 and above
>High quality: 70 to 90
Medium quality: 50 to 70
>Low quality: Below 50

https://wiki.x266.mov/docs/metrics/SSIMULACRA2
>>
>>106825288
I don't give a flying fuck about any stupid score, because they become unusable once they are used as a metric and don't account for generational loss.

Lossy to lossy reencoding loses quality and this is unacceptable.
>>
>>106825181
Dam so you've never had to pay rent or a phone bill in your life? I'm jealous desu, I'm like 2k in the red under the table because I've had to borrow from friends/family to cover rent/bills. Maxed out credit cards to boot as well...
>>
>>106825303
Okay, so you're a fucking 90+ Ssimulacra2 nut. Got it.
>>
>>106825310
I have more bandwidth for less costs now than i had 25 years ago.

So why should i serve images in lesser quality than 25 years ago?
>>
>>106825320
No.
100
Exactly 100.

If you want me to reencode existing images, just for the sake of changing format, it has to be 100.
Every pixel has to be identical.
Otherwise you introduce generational loss and a conversion that messed up 10 pixel the first go, will have 10000 the 100th time.
>>
>>106824389
KYS jewglejeet. This is a rare RapeApe W.
>>
>>106825462
Why would you defend him, even ironically...
>>
>>106825490
Why would you defend jewg.. oh wait your H-1B depends on it.
>>
>>106825528
Ah yes, the open source royalty free image codec is going to make google rich.
>>
>>106825567
Your coworker invented it and your employer pays you to shill it while turning a blind eye to the shit they bloated the web with.
>>
Is RapeApe a nocoder?



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.