unforum

[arksor]

[SOLVED] November 8 posts on paintover

[yanka]

yeah, thanks to astashenko for noticing that first.

[BrianEnigma]

To move over my comments from the other thread and add something new:

If anyone is thinking this is a substitution cypher, here are some of the highest letter distributions for the Unicode chickenscratch (sample taken over all Unicode posts, excluding the ASCII one):

http://metacortex.netninja.com/paintover/letter_distribution.txt

There are 362 unique Unicode characters used, which makes me think this is not the case...

[Ehsan]

I'll start analyzing the first post..

Factoring to:

4468 = 2 ^ 2 x 1117
1814 = 2 x 907
1821 = 3 x 607
18185 = 5 x 3637
17416 = 2 ^ 3 x 7 x 311
3587 = 17 x 211
3333 = 3 x 11 x 101
2311 = 2311
17223 = 3 x 5741
2833 = 2833
1033 = 1033
7195 = 5 x 1439
4368 = 2 ^ 4 x 3 x 7 x 13
6427 = 6427
21508 = 2 ^ 2 x 19 x 283
2374 = 2 x 1187
21021 = 3 x 7 ^ 2 x 11 x 13
5715 = 3 ^ 2 x 5 x 127
5638 = 2 x 2819
5895 = 3 ^ 2 x 5 x 131
21573 = 3 ^ 3 x 17 x 47
4870 = 2 x 5 x 487
7439 = 43 x 173
1054 = 2 x 17 x 31
1562 = 2 x 11 x 71
18177 = 3 x 73 x 83

Notice all the large primes.. only a few have smaller factors.. others can give you a large prime if you divide by 2,3,4 etc.. or a large number like 17x13... Take a look at the first 5 to see what I mean..

Looks fishy..

Edit: Other digits chosen randomly from different posts also have the same characteristics..

More spec: 26 numbers.. hmm.. alphabet? The first post is a guide?
More edit: Nope, they all resolve to differnet primes.. I tried other numbers and they weren't on the above list so it can't be a guide.. I'll try to do a full list later today

[BrianEnigma]

Another odd thing is that the control characters pass through as control characters and are not encoded in Unicode. It seems odd to embed a ^Z character in the message when you could just as easily (or even more easily and less ambigously) converted that to a Unicode sequence like "& # 1A;" (without the spaces). Additionally, there could be character sequences like ^M that we do not see because Movable Type would convert them to a BR tag (giving the funny line breaks in the ASCII post).

I'm really liking that prime angle.

Some more random thoughts: each post (with the one exception) is a string of Unicode. A string of Unicode can be thought of as an array of 16-bit integers. A 16-bit integer is a pair of 8-bit integers. I have been going through the array of numbers, splitting each, then applying various mathematical transforms (and, or, xor, adding, etc) between the low and high bytes. I have also tried the same transforms between the "current" position in the array and the "previous" or "next" one. So far, nothing interesting on that front.

Also, I tried adding the individual digits of the decimal number (like & #4468; is 4+4+6+8==22) and using those as position within the alphabet, but nothing interesting there yet, either.

[Zimbu]

If you look at the source, the charachter codes are all entered manually, so I would assume they (the character codes) are the significant part. (Except in entry 25, where it maybe the chars or sustitution that makes more sense)

[WedgeGold]

Matrix

[zeyd]

I ran the page through http://babelfish.altavista.com/ just for giggles. Some of the characters translate. It might be good to try another translator, and check the results. I submitted the url rather than a block of source.

Chinese to English translation is attaached.

Others were even less informative.

It doesn't look like much to me, but I thought I'd try a simple way while the mentats were crunching the numbers.

[AnthraX101]

Regarding the ASCII post, that first lowercase "t" is driving me nuts. The file appears too have two "parts", one which consists of characters between 0x01 and 0x20. The second is between 0x41 and 0x5E. The only exception to these two is that little lowercase "t" (0x74).

AnthraX101

[Styles]

If anyone needs a Unicode reference, here is a site with every single character you could imagine, in numerical order.

http://www.devever.net/pegercer/akx/unicode.php?ran=69

[BrianEnigma]

I am not sure how many people independently did this last night/this morning, but for the newcomers I wrote some code to convert the Texel messages into integer arrays of the Unicode+control characters (or ASCII+control characters in the one case). The arrays are formatted for Java, but can be used in C/C++ with slight formatting changes. I am sure the format can be massaged around a bit to be used in Perl (and hey, that's what Perl is great for).

http://metacortex.netninja.com/paintover/code/TexelMessages.java

Some assumptions were made on the ASCII segment based on embedded HTML (BR and P tags), but those are noted in the comments.

[AnthraX101]

I've done some more playing around and I beleve that the one ASCII message should be a unicode message just like the rest, but got dumped back into ASCII somehow. By pasting one of the unicode messages into notepad and saving it as Unicode, the plaintext representation (and character distrobution) looks similar to the one malformed message.

Unfortunately I do not have enough experiance with unicode to figure out the conversion yet.

AnthraX101

[SquareKnight]

But the unicode messages aren't all in unicode.

Each unicode message contains at least one character that is not referred to as a number.

[Moatie]

Can someone just basically explain, what unicode is, and more importantly
give an example...

like what does

#1054#1562 mean?

or how does one translate 1 line or several charaters into some type of unicode as an example.

I would like to work on this problem too

Moatie

[BrianEnigma]

If anyone wants go about this in a more visual and less mathematical way, there is an image of Texel's unicode messages rendered big enough to see well and in the proper fonts right here. Be warned that it is a large (500K+) TIFF image. The grey boxes with diagonal hash lines look to be undefined Unicode characters (I spot-checked a few of the ones Texel uses against the Unicode spec).

Personally, I think the solution has more to do with the numbers or the embedded ctrl characters, but I keep being asked about the visual representations.