siegeofangels | (no subject)

(no subject)

process for solving

melannen's manuscript

1. Stare at pages.

2. Pick out "Charlemagne" and "Byzantine" and "priestly."

3. Try to figure out what corresponds to A. And E.

4. Think you've figured out what E is.

5. Wait, no, that can't be "Charlemagne."

6. Does that say "Higgledy Piggledy?"

7. It's

melannen OF COURSE IT DOES

8. But if it's a font, why aren't all of the Gs the same?

9. WAIT.

10. The missing bits from THAT G are in THAT OTHER G. It's not a font.

11. Save image to desktop.

12. Open image in MS Paint.

13. Color pick to set background color.

14. Cut out half of text, overlay on other half, matching illustrations.

15. Decode!

Like I said, I totally cheated, but I am terribly interested in seeing how you managed to halve the letters to begin with, and also interested in seeing what the ease of translation would be with more common words--I think the only reason I got it was that "Higgledy Piggledy" was so distinctive.

Flat | Top-Level Comments Only

There was no "method" for halving the letters - basically I overlaid two copies of the text (with the both complete) on separate layers, set both layers to 50% transparent, and then artistically erased bits of each layer so that all parts of the letters stayed gray, they all had at least a little bit of black ovelap left, and each separate half looked at least vaguely like it could be an actual text, without making any of the words *too* obvious.

You probably need a program slightly more advance than Paint, though.

Figuring out which bits to erase so that it looked right, wasn't a total giveaway, and still worked actually took me several times longer than drawing the illuminations. If I practiced at it more, I could probably learn to paint a version that was both easier to read and harder to decrypt, but I mostly just wanted to see if the concept worked :D And, yeah, I probably could have picked a better cyphertext, all of those repeated bits and long, unusual words weren't helpful, but I have that poem memorized and I don't actually have a copy of the book (

stellar_dust) stole our copy.)

As I think you have realized, you figured out the science fiction method of decryption, hooray! The fantasy method involves treating the two-page spread as a stereograph.

To get the stereograph to work: make sure the image is no more than about four inches apparent size. If you wear glasses, you might have to take them off (and if you're too farsighted to focus on the image without glasses, or don't have binocular vision, you might be out of luck.) Let your eyes drift out of focus, and let yourself go double-vision until the illustration frames "lock" in to place, with a super-sharp, 'popped' frame in the center and two half-there frames on either side.

The text inside the center frame should look all weird and wavery, and like it's not on the same plane (geometric or spritual) as the frame. It's still really hard to read that way, but if you can lock the illustrations together steadily enough, you (or at least I) can decipher the text by concentrating on the letters one at a time and letting each eye's component fade in and out of dominance.

(If you want to unlock this post I will link it on mine for people who want to skip to a solution.)

Edited 2010-11-01 00:07 (UTC)

Unlocked! Feel free to link.

I used to be able to do stereographs, but either they don't work on a computer screen for me or my eyes have gone wonky (am guessing the latter).

If I was being all medieval scholar about it I would probably use tracing paper and a window, but Paint was so much easier. :)

Ahh, I see your method. (Or not, because I'm still rocking Paint. But I get it.) I suppose the trick would be to split each letter by hand so that the individual symbols couldn't be solved by standard decryption.

It's way cool, though, and now I want to make one.

The difficulty isn't avoiding standard decryption so much. It's pretty easy to come up with several ways to split each letter, ways to make one letter look like two, and ways to make different letters encrypt the same (for example, c e o d and q can all be sliced to look like the same letter on one side), and depending on the font you can go a lot farther. With those worked in, it might still be susceptible to some fancy multiple-statistical anaylses, but your basic manual methods will be fooled.

The difficulty is splitting them in such a way that it isn't subconsciously obvious what you're doing before anyone even tries standard decryption very hard. :D (And clearly I didn't do quite well enough, because you figured it out pretty quick, and holyschist is getting a word here and there without having figured out the main trick yet.) English is actually in some ways a hieroglyphic language - for example, if you write a sentence in lower case, and then block it out so that you can only see the "shape" of the word, which letters go above and below the center line - a lot of fluent readers can still read it easily. So you have to make sure both individual letters and whole words aren't recognizable as themselves, and make sure the two halves look like they have notably different text in them.

To make it work as a stereogram I also had to make sure that each letter had at least some overlap, so each encoded letter had to have something like at least 2/3 of the original strokes left; if you weren't trying to make the stereogram work, you could erase more, and it might be easier.

...in other words, you should totally make one! I want one to try where I don't already know the plaintext.

I knew how it worked IMMEDIATELY, but I still can't read it, because one of my eyes is SO DOMINANT that no matter how hard I try, I can't get any of the word-bits on the left-hand side to pop enough to make out anything. *cry*

well... if you want to try to do it anyway, you might be able to download the image, and then use your graphics-program-of-choice to lighten the text on the right. A lot. I did that with some of the test versions when I was having issues with dominance, but the problem with putting it in the final image is there's no way to guess which side will need the fading for any given viewer.

Alternatively, look through a sunglasses lens with just one eye? :D

Edit: Also note, I can't get it to work with my glasses on, at all - I can't even 'lock' the frames. If you can't see up-close enough to try it with no glasses, you may just be out of luck.

Edited 2010-11-01 03:55 (UTC)

... okay, I can just *barely* do it if I take my glasses off and sit far too close to the screen.

I can lock the frames with my glasses on though, but nothing else!

Flat | Top-Level Comments Only