Realm of Twelve - Extracting Kryptos Font

Extracting Kryptos Font

This is an archived document and is no longer being updated (Spring 2009).
The original Kryptos Font and Placement can be downloaded at the bottom of this page.
See the new Kryptos Font at KryptosRevisited.com for additional downloads.

In January of 2005 I proposed the idea that Kryptos uses some sort of steganographic kerning cipher. The most difficult challenge in pursuing this theory was determining the exact typeface and orientation of each character of the copperplate. The problem: the only resource I had to work with was grainy photos.

I worked with 23 photos of the copperplate in an attempt to extract a font that is reasonably close to the original. This process included mapping each image to a three-dimensional model, and then unfurling and orienting the plate to a two-dimensional Cartesian graph. Finally, I took identical letters from different locations and proportionately resized them to perfectly coincide with each other. With these superimposed averages, I set out to begin vector-tracing each one.

Outlining each character proved to be a tedious task. Because of the thickness of the copperplate and lighting conditions from photo to photo, the exact perimeter of the letters was sometimes ambiguous. In these cases, I had to reference several copies of the character in question to average the location of their borders. Patience yielded a wire-frame typeface with which I was satisfied after several days of adjustments.

In retrospect, obtaining the font was the easy part. Placing each letter in their proper position and proportion on the copperplate was something I would rather do again with more sophisticated tools.

Basically I compensated the distortion caused by the angle of the curved plate and adjusted its orientation so that the four corners of the cipher text side fit into a rectangle. With carefully placed guidelines I placed each character on a grid and adjusted the spacing with precision to make up for the curvature of the plate (which appears to be less than half a cylinder face -- or approximately a 135° to 145° arc per section). I continued to tweak the individual characters for minor anomalies that I was able to see only by comparing the kerning from character to character, row to row. After a week of manually aligning each letter, I began to see results that were as good as they could be under the circumstances.

Of course, the figures shown here are a fraction of the size of the working files. Fortunately, the way I constructed the virtual copperplate keeps errors confined to a small area. In other words, any inaccuracies in the spacing from one character to the next do not aggregate. This is important, because the kerning itself varies in identical characters at different locations on the plate.

When I completed the first stage of this project, I realized I was the first to have an accurate representation of the copperplate in a form that Sanborn probably had when designing this part of the sculpture -- something that may aid in finding the solution to K4. With this new way of seeing the copper scroll, the proper tools became available to try a kerning cipher, but in the mean time, I am porting this virtual plate into an interactive 3-D model of the sculpture. My goal is to try various lighting effects to see if any nuance lies between subtle shading and the absence of light.

I have provided two downloadable gifs of the most correct font, positioning, and kerning of the copperplate available. Although these are not perfect replicas, they are probably adequate for use in light-play testing. These graphics are intended to be printed on transparencies and may be redistributed in digital or print form so long as the copyright notice is clearly visible.

Download 125KB: replica-ciphertext.gif
Download 123KB: replica-tableau.gif

Announcement: A new Kryptos Font is available at Gary Phillips' new Kryptos site - Kryptos Revisited.

> About: The Author of this Site