Do publishers track raw providers? Images here contain hidden metadata

[TypeDexuAgain]

There's something I noticed a couple of times but I've never seen it discussed. Some images contain what I suspect to be hidden data, encoded in a way that reminds of the old Machine Identification Code used to track printer machines. In short, some images contain some barely-visible series of dots that may be used to track whoever is providing raws. This is most visible in blank space between pages.

Below is an example from page 10 of A Returner's Magic Should Be Special Ch. 138, altered to make the dots visible. (click to see the full size)


Does anybody know what's going on here? Do publishers use that data to track raw providers? Do raw providers add that data to sign their raws? (doubtful, but still) Are translating groups aware that this data exist? Could it be a freak accident that causes JPEG to create those highly-suspicious artifacts at such regular intervals? I've seen that pattern enough times to not take it as a coincidence, but who knows.

 

[TypeDexuAgain]

I looked at a few other series and as far as I can tell, most (but not all) Korean webtoons I tried had the same kind of tracking code. It's a different code every time, encoded in a 9 x 8 pattern repeated a few times at different scales on the same page. That about 72 bits of information. Not enough to store an IPv4 address but enough for a unique ID and timestamp that points to whoever downloaded the image.

I don't know why they would want to track users like that though. They don't need that to shut down sites, so it would only be useful if they wanted to sue individuals, which would be a complete waste of time and money.

 

[Halo]

News from 5 years ago: https://redd.it/3otdq8
Raws distribution got much worse (i.e., "better") since then, so maybe nobody really cares. The dots in your example can be easily removed by adjusting levels in photoshop.

Quick google: drminside.com
They have a number of patents, for example 10-1841192:

Unlike a conventional method that a feature point is extracted based on an image of the entire area, the method of the present invention comprises the steps of: extracting a feature point based on frequency properties of an image; and, based thereon, cross-correlating feature points extracted respectively from illegally copied content and content of online comics...

Spoiler

Not saying it's the same, just looks similar.

so it would only be useful if they wanted to sue individuals, which would be a complete waste of time and money.

They need a government issued ID to read webtoons in Korea. So, simply by having this system in place they can scare off a lot of people even without suing them.

@TypeDexuAgain

 

[pandascepter]

@Halo
Wow, youre so smart.

 

[Halo]

@pandascepter Not everything must be a shitpost, Pando 😔

 

[TypeDexuAgain]

Thanks for the links, Halo. It's not the same system because the one in my screenshot is much dumber than the one from that patent. The patented one seems to use statistical distribution so it shouldn't be as obvious to the naked eye as the dumb dots. I don't think either is very hard to mitigate but they're annoying to deal with considering the whole fan translations thing is done by... well, fans. Koreans really have some weird surveillance-type stuff.

A quick correction about my earlier statement, the 9 x 8 pattern doesn't carry 72 bits. They're just lines of 0 to 9(?) dots, so it's really a 8 digits number that can only generate 100,000,000 combinations. (26.5 bits)

I've just Googled Daum to see if they were suing anybody and... they are. There are headlines about Kakao/Daum winning at least one lawsuit over webtoons. Raws providers need to cover their asses.