contact my nuts bitch

Stop uploading your images with metadata, faggot.

>>4812226
Thanks for raising this question here. Image compression was a kick in the balls for me. What the heck? Now I'm wondering if they came up with some kind of script that would immediately open the ORIGINAL images.

Compressing images without warning is really paranoid shit.

File: 1P6PB3Pk_o.jpg (892 KB, 2337x3504)

892 KB JPG

>>4812226

My tests show the image returned by putting ?dfd on the URL is not the original uploaded image either. Its a slightly larger file ??

901901 Feb 4 19:01 1P6PB3Pk_o.jpg Original file
912952 Feb 4 19:31 1707091367689325-dfd.jpg File returned by adding '?dfd' to URL
872499 Feb 4 19:30 1707091367689325.jpg Default image returned

>>4812253
>Now I'm wondering if they came up with some kind of script that would immediately open the ORIGINAL images.
Check the Desuarchive link in the OP, I personally haven't tried it, but tell me if it works for you, if you use it.

>>4812262
4chan strips images of any metadata when they are uploaded, so that is liable to change the file size. Why it got slightly larger in size, I don't know. The problem addressed in the OP is the Cloudflare compression that was turned on last year.
Also note, you don't have to type exactly "?dfd", it can be any random letters you input after a question mark is added at the end of the image file URL.

>>4812262
Can someone dignify AI her.

>>4812439
What are you talking about? She's covered up much more than usual. :P

>>4812226
I noticed file sizes were fucked and assume it is steganographic fuckery. They were already fucking with images to remove exif since forever.

>>4812450
>steganographic

that's an interesting idea. since 4chan does not allow VPNs or proxies any images posted anonymously to 4chan could be tagged with the IP address of the original poster with steganography. Then if the images were re-posted on other sites they could be traced back to an IP address.

It’s over…

likely in response to the Taylor AI backlash

File: Absinthe_Fairy_illustrati(...).png (6.6 MB, 3200x1800)

6.6 MB PNG

>>4812226
800x800 dpi
3200x1800 pixels
16.7 M / 24 bits/pixel
282,658 unique colors
6.60 MB (6,924,853 Bytes) on disk

File: 1707283953538153_stats_01.png (1.89 MB, 1920x1080)

1.89 MB PNG

>>4813339
>picrel

The stats are the same, except that the dpi has been stripped out and set to undefined.

The file is exactly identical, including the DPI is no longer stripped out and remains the original 800x800

And fuck 4chan for flagging my post as spam and disabling the conversation.

File: 1707283953538153_stats_02.png (146 KB, 1200x1080)

146 KB PNG

>>4813358
fuck 4chan and it's gay fucking broken spambot detector; I was going to do some more shit, but fuck 4chan for being fucking gay

>>4813339
>>4813358
I did screw up one thing, though. The Size on Disk for everything I downloaded did change. No matter which version (I downloaded 5 copies from 4chan and 4plebs), it has been reset to:
6.60 MB (6,924,832 Bytes)

The original uploaded image was 6.60 MB (6,924,853 Bytes), 20 Bytes larger than all the versions that download from 4chan/4plebs servers.

Also of note, 4plebs uses the identical URL for its stored image (see previous post since I'm forbidden from using URLs in my posts today for some crackhead reason). Over time, that may change? Once the 4chan archive expires, 4plebs may roll over to using a different server ..? Dunno. Have to wait for 30 days after the thread goes to archive & then expires.

It will take more experimentation to determine exactly what is going on, but I'm out. 4chan successfully shit on me enough to make this too fucked for me to care about untangling it. My suspicion is that *.png files are a best case scenario of an already optimized & losslessly compressed file tat doesn't have much room for further alteration. Unoptimized PNGs and other lossy formats such as JPEGs may get different treatment.

>>4813366
>20 bytes larger
I’m guessing they’re just stripping out some metadata. That’s not compression.

>>4812239
They're incommunicado.

>>4812857
What would be the point of that? I'm sure they maintain hashes and logs anyway.

At least there is something I can do, that is the fresh cherry on the shit cake.

>>4812450
>steganographic fuckery
What's that?

>>4813339
That is some amazing trad artwork.

>>4813358
>The file is exactly identical, including the DPI is no longer stripped out and remains the original 800x800
I'm aware its "lossless" compression and won't effect the image quality, but the image file size is what is different (smaller). Is this from the DPI info being stripped?

>>4813359
If 4plebs archived the correct versions, there wouldn't be as much of a problem. You could always get the proper file there.

>>4813366
>Have to wait for 30 days after the thread goes to archive & then expires.
For boards that do not have active 4chan temporary archives (like /b/), say you left a thread open in a tab and its now 404'd off the board, and you can still access the thread itself in that tab. It takes a bit of time for the image cache to purge from 4chan's servers. You can still access the usual image URL with the compression that has been added to the image for a few hours, however the trick using "?randomletters" will result in a 404 right away.

>>4815133
>some amazing trad artwork
You will probably appreciate the relatively recent Olivier Ledroit thread where that made an appearance, along with a lot of his other work:
https://archive.4plebs.org/hr/thread/4625075/

And, agreed. One of the more amazing artists in the world today.


>image file size is what is different (smaller). Is this from the DPI info being stripped?
I don't think so. If you do the ?arglebargle hack on the image URL, the version that preserves the 800 dpi is the same 20 bytes smaller as the images with the dpi stripped from them.

What I was posting was proof of concept & confirming the basics that OP presented:
1) the images are altered
2) for a short time you can hack out the original, unaltered image from the 4chan servers
3) 4plebs (I think the only archive for /hr/) saves the altered file, not the original

Beyond that, I gave up further experimentation (such as with different compression JPEGs) to try & determine how severely the images are altered. I only began the journey with an optimized .png specifically because I didn't think there was *ANY* room for compression. Which was "sort of" validated since 20 bytes on a 6.6 MB file is effectively very nearly zero alteration. For purists (like me) though, even that small alteration is disturbing. There was no metadata to strip off the original file since I edited & saved it myself without any metadata at any point in the creation of a new image file.

If 4chan wasn't preventing me from posting URLs as part of the investigation, I'd be happy to fuck around with some other files and try to fine tune it. But, fuck 4chan for making the task ten times more tedious than it needs to be. Accusing me of SPAMming just pisses me off & I don't care to eat a ban for something I'm not doing and would never do.

>>4815285
At least 4plebs archives the images. /wg/ (i.e. archived.moe) doesn't even do that.

>>4815752
https://archive.palanq.win/wg/

>>4815788
Oh, didn't know about this archiver. Thanks, anon.

>>4815752
at least up until about 2014. otherwise the thumbnails are there, but trying to view or save and it 404's.

I just discovered that thebarchive is archiving the wrong, compressed images just like 4plebs is doing. It's one of the most active boards on 4chan and alot of original content gets posted there. This sucks.

I'd like to give this thread a bump to spread the message regarding this image compression problem.

File: hingis.jpg (875 KB, 2000x1625)

875 KB JPG

>>4818371
>still thinks removing metadata is compression after being told several times already
Zoom zoom. 866737.

>>4818390
Thank you for contributing nothing of worth to the thread and being an ignorant ass. Was well aware of the metadata stripping 4chan has been doing for years. If you bothered to read the thread, we are talking about the compression being done by a Cloudflare feature 4chan starting using mid last year. It's been confirmed by several people already, what is happening. You're wrong, but nice try trying to gaslight by framing it as "after being told several times already".

>>4812226
Thank you for letting us know OP

>>4819876
You're welcome. Please spread the word when and where you can, linking the /trash/ thread in OP as well as this thread.

>>4819876
>>4820181
>>4812226
retard here... people
I've seen various counter arguments to this and am not sure what to believe
compressing is okay but only if lossless (real lossless bit for bit) not visual lossless aka lying
filesizes are a meme
has anyone ran a pre-upload vs a post-upload and overlaid them in an editor in diff mode to see if there's any photographic/image based differences in the pixel content? or maybe this is just discarding color profiles/exif or something or just lossless compression added somehow

either way if this isn't some big fake news lie this is a pretty big happening IMO

>uplooad something
>have filesize limits
>they alter it anyway
umm
at that point why not just eliminate upload limits if they're gonna cloudflare them down?
if this is some jpeg to webp to jpeg autism shit I will seethe but haven't seen much evidence so I'm not sure if these posts belong on /x/ or if they are valid at all quite yet
if anyone cares to explain with examples (like pngs on imgur supplied, where we can upload on our and see 4chan fucking with them firsthand) that'd be appreciated

>>4820528
>compressing is okay but only if lossless (real lossless bit for bit) not visual lossless aka lying
In my opinion, compressing is not okay even if its lossless compression. It's also frustrating that it was implemented without any announcement from site staff. Users just slowly started to realize what was happening, and the real crime here is that alot of people don't even know it's happening. Not sure what you mean by "lying".
>filesizes are a meme
No, they are important with regards to 4chan archives and how the "View Same" image feature has been rendered somewhat useless because file sizes keep changing when a saved file from 4chan with the added compression is reposted. If you cared to read the information in the OP and the rest of this thread, you'd know what is going on.
>has anyone ran a
Yes, right here in this thread if you cared to read. I've also seen posts on other boards about it.
>or just lossless compression added somehow
You really just blazed in here without reading a single thing, didn't you? It is lossless compression.
>either way if this isn't some big fake news lie this is a pretty big happening IMO
Why would you think its a lie? Like seriously, where is your head at?

>>4812226
Has anyone investigated if this is also happening to webm & gif & other file types? Animated gifs same as static? Etc? Any different behavior from different boards? Or at different archive sites?

4plebs has a page with some of the various archive sites
here https://archive.4plebs.org/_/articles/credits/

Are there any others not listed on that page?

>>4812231
>>4812239
>>4812242
please stfu and kill yourselves

>>4821775
Gookmoot is a real skekel-grubber lately, maybe the lawsuit against the site has something to do with it (over the Buffalo mass shooting IIRC). Legal fees aren't cheap.

>>4823534
>Has anyone investigated if this is also happening to webm & gif & other file types? Animated gifs same as static?
This is easy enough to test yourself. Simply find one of those file types to save, and save one version of the file as you usually would and save it again with the URL gibberish trick shown in the OP. You don't even need to use more than one letter, just add "?[letter]" to the end of the image URL to get the non compressed version of the file.
That said, I recently saved a couple of animated gifs, I saved them both twice using the method described. One showed a difference in file size. The other did not, both files were the same size (not compressed). It's very odd. It's like once in a while, a file is missed and the extra compression is not applied to it.
>Any different behavior from different boards?
As far as I can tell, it’s the same on every board.
>Or at different archive sites?
Now that you mentioned it, I made another sad discovery. Archiveofsins is another archive that archives the compressed versions of images (the wrong ones). I checked this by going to a couple of the boards that archive covers, saving the same file twice from a thread using the URL gibberish trick, and saving the same file again from the archive. The archived file is the smaller file size, the compressed one.

>>4824095
>maybe the lawsuit against the site has something to do with it
I had no idea this was happening. Is it still ongoing?
I keep wondering one day if 4chan will ever get shut down due to lawsuits.

File: lion_Cameron_and_tiger_Za(...).png (2.1 MB, 1920x1080)

2.1 MB PNG

>>4824580
Yes, I am investigating. I was hoping some other anons might post some of their results. It's a fuckton of work to do all by myself.

For the autism of it all, I posted this >>>/an/4739913

The change in filesize on the post reminded me to check some of the stats. picrel shows:

middle: original img properties on my HDD; I saved it as a progressive 99% JPEG. (the subsampling originally was 2x2, and remains so in the downloads)

right: D/L'd WITHOUT the ?arglebargle hack to see native stats and how it might be expected to be saved to an archive service

left: WITH the ?arglebargle hack

I deliberately stripped out all EXIF info and used the 99% & progressive JPEG settings For Reasons so I could compare later to other settings and other file formats. Surprisingly, the progressive JPEG setting was stripped.

Next, the size on disk was increased by almost 1/6th on both the D/Ls over the original ... maybe due to not using the progressive setting? As far as "compression" goes, WTF the JPEG was INCREASED in size, not decreased. Dunno. Weird. But, the load time changed from 250 milliseconds for the original to 100 ms for the two downloads. The Current Memory Size (Win7/IrfanView) remains identical for all three files.

Also note, my upload was to 4channel (blue board) rather than 4chan (red board) ... even though the root domain in the URL stays the same. Dunno if this is a factor, but keep an eye on it.

picrel stats on my HDD:
optimized level 9 lossless *.png
no DPI
2.07 MPixels
16.7 million/24-bit color
224296 unique colors
disk size 2.10 MB (2,206,555 Bytes)
current mem size 5.93 MB (6,220,840 Bytes)
loaded in 94 ms

For the home players who want to see what /hr/ did to that image.

What software & systems do other anons use to analyze their images? What stats are available?

We need to stop referring to this as a "compression," though. Smaller filesizes are not always the end result.

Here is a summary of what archives are doing:

Archives that save the correct, non compressed versions of files:
desuarchive, archived.moe, arch.b4k.co, warosu, palanq.win, eientei

Archives that save the wrong, compressed versions of files:
4plebs, thebarchive, archiveofsins, fireden.net

>>4826044
Stop using "compressed" in this context. The files are being altered, but sometimes they are quite a bit larger than the originals being uploaded. Something else is going on here. All you had to do was read the post immediately above yours.

>>4812226
Madelaaaaiine

this has been occuring for years, 4chan uses a bad EXIF stripper that repacks the JPEG file and unoptimizes it in the process
the presented pixels are the same but the way they're stored changes
fun fact, people used to exploit this a few years ago to get images past the filesize limit, since the stripper's repacking would bloat filesizes and there used to be only one filesize check that only ran when you selected the file to upload
for example, the EXIF stripper could unoptimize a very tightly packed and specially prepared JPEG from something like 2MB to 100MB
4chan's devs didn't change the EXIF stripper but they did add an additional filesize check that checks the fully uploaded file to see if the repacked version is past the limit
this is also why JPEG files around 3.5-3.9 MB on most boards and 7.3-7.9 MB here can get stopped by the filesize checker and refuse to post

Weird. Encountering an intermittent issue where the URL gibberish trick doesn't always work, and I'm still getting the smaller file size.
If on a board where the archive is saving the proper images, I can still resave them there, and they are the larger sizes. Overall it's very annoying why the trick isn't working all the time,

>>4826166
>this has been occuring for years
No, you are incorrect, what is listed in the OP and what this thread is all about - saving images and the file size on your computer not being the same to what is listed on the site/some archives - is the problem.

Sure it's annoying in principle.

But you guys are already using lossy compression. Can you even tell the difference between a 710kb jpeg and a 791kb jpeg? There's a reason we don't use pngs and raw files for these things.

>>4812226
this site is long dead dogshit ran by glowing assets and troons that hate you. there are better places to spam your waste of bandwidth celebrity worship bullshit

>>4827546
Every other English speaking alt chan is dead activity wise.

>>4827531
It's not about the immediate visual but the effect on repeated compression/re-encoding. Phones are the leading cause of image degradation alongside phone-centric social media like twitter converting 50kb gifs into 2mb lossy videos, and google converting everything to webp to save pennies on bandwidth.
https://files.catbox.moe/ktoywy.webm

There was a brief "issue" a few months ago where 4chan had started converting .pngs into .webp which stopped a couple hours after.
https://desuarchive.org/g/thread/95875634/

>>4824785
>Next, the size on disk was increased by almost 1/6th on both the D/Ls over the original
That is very odd. I've never had a case personally where the downloaded file from 4chan was larger in size than the original uploaded.
>The Current Memory Size (Win7/IrfanView) remains identical for all three files.
Also weird. But as stated, sometimes the compression misses some files. I don't know why.
>Also note, my upload was to 4channel (blue board) rather than 4chan (red board)
I don't know if that matters, 4channel no longer exists. It was removed recently.
>We need to stop referring to this as a "compression," though. Smaller filesizes are not always the end result.
They almost always are, though, at least in my experience . It is compression, it"s lossless. At least that is how it was explained to me.

>>4827531
It's the fact it's messing up archives and creating new file paths/hashes for the same image making the "View Same" function relativity useless, while also adding more images to their servers because a new copy of the same image is being made. If users keep resaving and posting the altered version of the same file, you could wind up with several different hashes for the same image.

>>4829216
Interesting video you posted, never heard of JXL.
>where 4chan had started converting .pngs into .webp which stopped a couple hours after.
They likely were testing something and rolled it back because people were complaining about it. That would really be the final straw for many users, imo, if they made that change to convert things to webp.

bumo

>>4830355
>That is very odd. I've never had a case personally where the downloaded file from 4chan was larger in size than the original uploaded.
You should pay more attention because I guarantee you it happens constantly.

File: 931B50F3-E586-4FB9-A88A-7(...).jpg (2.84 MB, 2160x3840)

2.84 MB JPG

>>4828524
/ttg/ has their own chan now after they got kicked off /wsg/, it's very much alive (couple thousand posts daily) but rather narrowly focused.

File: 1710631762276.jpg (474 KB, 1664x2304)

474 KB JPG

Theory 1: They want to enhance our anonymity. (Because you can spy on posters by their images.)

Theory 2: They want to disrupt steganography. (Because large images are ideal for steganography.)

Personally, I think that §1 might make some sense 15 years ago; it would be stupid today.

Theory 0: They are cheap as fuck and reducing file size means lower hard disk space, meaning less leased servers. (Because reducing the customer experience puts more money in their pocket.)

>>4828524
>Every other English speaking alt chan is dead activity wise.
I wonder which mysterious law of nature kills every chan that isn't 4chan… If anonymous forums were people, they would be falling out of the windows.

>>4831852
I can remember having this exact problem on /wg/ back around 2009/2010. I'd try to upload a jpg that was ~200 or ~300 KB under the post size limit, and after attempting the upload & solving the CAPTCHA 4chan would barf and tell me the filesize was too big. I never understood why the upload would fail, but now I see 4chan was re-encoding the file into something larger. I didn't know any better, so I never investigated other file formats, I'm just assuming right now I had that problem with jpgs based on the posts in this thread. Increased filesize has been around for a long time now, but I never suspected that the files were being rewritten & altered beyond metadata getting stripped; THAT I knew about but didn't think it changed the filesize.

>>4831854
who is this woman?

>>4831854
What is /ttg/?

>>4832387
Bingo.

>>4833282
4chan is the biggest chan, so it would take alot of things to happen to finally bring it down.

>>4834861
Weird, I don't remember this issue. I wonder if it is still happening?
/b/ has a 2MB limit and there are times I was able to sneak in and post an image that was a few bytes over 2MB. Like within the last year, this happened. Might be because of the compression issue as talked about in OP.

>>4836467
>don't remember this issue. I wonder if it is still happening?
Yeah, been going on for years. Yes, still happening. Pretty recently (1? 2? months?) I tried to post a ~3.95 MB jpg to a board and it failed the 4 MB limit due to "exceeding the size limit." Had to resize it using a higher, more lossy compression to something like 3.7 MB, then it posted at something like 3.98 MB. Had to do that a couple dozen times over the years, 2010 sounds about right for when I first noticed it. Have had to plan around that before even attempting to post.

wheres the tldr?

>>4840182
You're in it.

Summary: original files that are uploaded to *chan servers get "reprocessed" into new files that no longer match the originals. There is no explanation for why or how this is done, and it turns out that smaller file size to save server space & bandwidth isn't the correct answer.

It's bad because original files get deleted and vanish from the Internet, chain of custody and integrity is broken. Quality & integrity of the new files is unknown. Archives for the *chans are picking up the altered files rather than copies of the originals, so once a *chan delets an original file, it's gone for eternity. Which is a dick move.

Unless saved & archived somewhere else.

Is that still too long?

>>4826044
>thebarchive, archiveofsins
nothing of any value there anyway.

>>4840182
I think I understand the whole tldr bit
but nobody has provided proof of anything yet
so I'm not sure if this is schizoposting or not

>>4845015
yes its schizoposting
the metadata from uploaded files gets removed, this results in a small reduction in filesize
it would be easy to show if images were actually being compressed

>>4812226
Ahhh so maybe that's why image hash-based searches are no longer working on the archive sites. I noticed last year or maybe the year before, it suddenly stopped working.

>>4813623
They've been doing that for years because retards kept getting doxxed.

Unrelated, but how do you guys upload such high res images without hitting the file size limit?

Also any tips for uploading optimal webm would be appreciated too

>>4845750
Size limit varies by board moran.

>>4826580
I'd like to update this by saying this trick seems like it no longer works at all on /b/. I have saved several images from there as of late, and they are all a smaller file size than what is stated on the site at the time of download. Their archive also saves the wrong versions of files, so anything saved anywhere regarding /b/ is going to be the incorrect file. It's still working on other boards I visit (with occasional failures). Strangely enough on occasion I get a slightly larger file size than what is stated on the site at the time of download. This happened recently and is a first for me.

>>4841631
>and it turns out that smaller file size to save server space & bandwidth isn't the correct answer.
Then why would they do it, if it was not this reason?

>>4845015
>but nobody has provided proof of anything yet
This entire thread is full of proof. Are you genuinely retarded? Or too lazy to read? Or both?

>>4845668
Correct.

>>4845750
>Unrelated, but how do you guys upload such high res images without hitting the file size limit?
Sometimes it's just easier to upload files offsite to an image hosting service and to post the links in the thread, than to mess around with resizing images to get them under the size limit.

>>4848010
>why would they do it, if it was not this reason?
That's the point of these threads. Why, indeed. Could be a real reason, could just be retarded fuckwits who think they are genius "programmers" and don't even qualify as script kiddies. Seen that stupid shit a thousand times.

>>4846585
It's Moron you fucking pleb

>>4850480
>moran
(You) missed the meme. Welcome, newfren, enjoy you're stay.

>>4812226
>>4849693
>>4841631
>>4834861
>>4831852
>>4826127
>>4824785
Recently posted a jpg that was about 750K on my hard drive, but showed up something like 930K on the 4chan board. Because of this thread I was watching for it. After reading this thread, just more anecdotal confirmation that whatever is going on does not always generate a more compressed file, just a different one that's been re-processed.

lossless jpeg optimizers have existed since shortly after jpeg. You can often get 2-3% just by making the file progressive. PNG optimizers can get even more because most png software is garbo.
I don't begrudge them shaving down files like this so long as they're not actually recompressing. Bandwidth is money & 4chan shouldn't have to pay for adobe's lazy jpeg implementation.
Image hashes/comparison should always be based on pixel data, not the file. Get better software if you're doing it wrong.