It seems that you're using an outdated browser. Some things may not work as they should (or don't work at all).
We suggest you upgrade newer and better browser like: Chrome, Firefox, Internet Explorer or Opera

×
I took one of your posts to make a sample dictionary as a proof of concept. Here's "meowmx"

graphical even effort maybe. hard. meeting anyways. not {key} hide
and more fun:

gcc encode.c; ./a.out "Oh, I wish I were an Oscar Mayor Weiner!"
do of Hmmm... look. be difficult difficult 0.1% once for deduced had of Hmmm... Hmmm... difficult deduced be automate... hard. is had Hmmm... feels difficult 0.1% i graphical Hmmm... be and read method feel meeting feel far be Hmmm... do deduced feel far is maybe. had Hmmm... be any hard. is for deduced graphical isn't effort messages, be deduced
The dictionary is small, and i didn't do any compression, so the entropy is really high. In a real world scenario, i'd use a much, much larger buffer and dictionary.

EDIT: Source for dictionary:
Again once the pattern is deduced it probably isn't too hard. say if you had blahblah{key}blahblah, then looking for the {key} to decode it with...

Hmmm... Though i do remember coming up with a graphical method to make static and you could hide images within it, though that's too easy to automate... maybe.

Though it feels more like it would be far more effort to read your messages that people will both not decode and read your messages, but not respond to you anyways. If you make it difficult to any degree, only those who feel like meeting the challenge (hmmm less than 0.1% of people) would even look.
Post edited February 10, 2021 by kohlrak
avatar
kohlrak: I knew a guy who had something that uses the odd-even of a particular bit in a lossless format. The difference between 0xFFFFFF and 0xFFFFFE are even indestinguishable to the human eye. Only people knowing there was a crypted message there would bother to look. You could get 1-3 bits per pixel of hidden info by making it 7bit colors instead.
Mhmm... did something similar, though i used the palette. Though i'd done basically 7bit (128 colors) duplicated so the upper or lower bit were the hidden message. Wasn't too hard to make or do, though a little more annoying without Jasc PaintShop Pro installed. Obfuscation. Although there's other tools to hide messages in pictures.

S.S.E on android does encryption, but also includes a stenography option. You could encrypt a message hidden inside, you probably can't even detect there's a message unless you know the key.

avatar
kohlrak: The dictionary is small, and i didn't do any compression, so the entropy is really high. In a real world scenario, i'd use a much, much larger buffer and dictionary.
Doing per letter would not let you do many characters, you might do 2-3 characters. Upper/lowecase is also unimportant so you could drop it.

You could as a side bit, also take it where the 0/starting point is the last point it was last at (so double letters is far less likely to show anything), so it leap frogs. Depends on the range of characters really.

Assuming Base64 as a valid character set, 2 would be 12 bits (4k), 3 would be 18 bits (256k).

Remembering it there was this one old cipher done by a father/son pair for one of the kings of old, and it was almost unbreakable, which they finally broke due to 1 tiny hint. But they did character pairs i believe for the encryption too.

Meant to bring it up before, there's my old crypto thread, i also included a working but very simple cipher. Might be used as one of the passes.

avatar
kohlrak: It'd be painfully obvious for humans to know to copy and paste it into a known decoder, but it'd require an AI capable of actually understanding english to detect it. If you saw the following, for example:
it did look weird but seemed like something you would glance over unless you were really looking for something. So there is that... at just a quick glance at passing it looked like english.
Post edited February 10, 2021 by rtcvb32
avatar
kohlrak: I knew a guy who had something that uses the odd-even of a particular bit in a lossless format. The difference between 0xFFFFFF and 0xFFFFFE are even indestinguishable to the human eye. Only people knowing there was a crypted message there would bother to look. You could get 1-3 bits per pixel of hidden info by making it 7bit colors instead.
avatar
rtcvb32: Mhmm... did something similar, though i used the palette. Though i'd done basically 7bit (128 colors) duplicated so the upper or lower bit were the hidden message. Wasn't too hard to make or do, though a little more annoying without Jasc PaintShop Pro installed. Obfuscation. Although there's other tools to hide messages in pictures.

S.S.E on android does encryption, but also includes a stenography option. You could encrypt a message hidden inside, you probably can't even detect there's a message unless you know the key.

avatar
kohlrak: The dictionary is small, and i didn't do any compression, so the entropy is really high. In a real world scenario, i'd use a much, much larger buffer and dictionary.
avatar
rtcvb32: Doing per letter would not let you do many characters, you might do 2-3 characters. Upper/lowecase is also unimportant so you could drop it.

You could as a side bit, also take it where the 0/starting point is the last point it was last at (so double letters is far less likely to show anything), so it leap frogs. Depends on the range of characters really.

Assuming Base64 as a valid character set, 2 would be 12 bits (4k), 3 would be 18 bits (256k).

Remembering it there was this one old cipher done by a father/son pair for one of the kings of old, and it was almost unbreakable, which they finally broke due to 1 tiny hint. But they did character pairs i believe for the encryption too.

Meant to bring it up before, there's my old crypto thread, i also included a working but very simple cipher. Might be used as one of the passes.
The entropy drops significantly with a sufficiently large dictionary. It could actually be seen as compression since it would distinguish between capital and lowercase, maybe also include typos. English doesn't require acknolwedgement of these.
avatar
kohlrak: It'd be painfully obvious for humans to know to copy and paste it into a known decoder, but it'd require an AI capable of actually understanding english to detect it. If you saw the following, for example:
it did look weird but seemed like something you would glance over unless you were really looking for something. So there is that... at just a quick glance at passing it looked like english.
Of course. Most people would know the people they pay attention to. Anyone who likely would skip over the encrypted post would skip over the unencrypted, anyway, except machines looking for keywords and encryption keys.