>>2521648

Curious. The originals have all been restored.

Anyway, this thing seems to have fizzled. There is a legitimate explanation for all of the spoopyness that we have been examining. It's coming from Medium's weird backend image resizing code. They appear to be using the same funky Java-based encoder library from the late '90s that's only other use was in the F5 stegano demo code, also from the late '90s, which is what PixelKnot is based around.

So I think I might switch teams and instead work on patching all the problems with F5 and PixelKnot. ;)

>>2525792

>time to circle back, figure out where we got off course

Agreed. If we are to continue then we should focus on the SS photo. And we should look beyond PK and F5. Other stegano programs may leave telltale traces on the structure of their output JPEGs that could help narrow the search for the right method.

As far as I know this (pic related) is the original SS file. It was being passed around early and it's size matches the 4Chan screen caps.

SHA1: f1335a1095a3ae15094e0a09e1cb83e5679dda26

>>2529049

We can tell that they are originals if their SHA256 hash matches the links in archived threads (8chan hashes all uploaded files and uses that hash on the back end as a easy way to eliminate redundancy what the same memes keep getting uploaded numerous times). I haven't noticed links to the file_store ever changing, not even what the files stored under a given hash have changed.

>>2525792

I finally quit screwing around and got that C-based cracker I've been talking about to a (unfinished but) usable state. It's a bit faster than the Java code, but not by as much as I had thought. I have newfound respect for the JIT compiler.

(You can probably tell from my folk-code that I'm not a professional programmer. I'm a welder and machinist at a widget factory. All that I know about coding I learned through many late segfault filled evenings. )

https://nofile.io/f/NB4x9EOZYd3/BruteCrackF5.zip

>>2567227

…because we don't know what software was used… or if there is even anything there.

Q linked to the PixelKnot app. PK leaves behind a very distinctive signature in the JPEG file that does not exist in the Sarah Silverman pic. So either 1) it's a modified version of PK made to fix its weaknesses, it's and entirely different program, or there is no hidden data and Q was simply citing PK as an example of the sort of thing that is widely uses on the Chans.

If there is an expectation that were are to crack this thing then we are going to need more clues.

>>2573936

In the previous bread you said you were trying RockYou and some others against the SS pic. Did you finish? And was this only with the last third of the password?