What you are looking at
An Autoglyph is not a picture that was drawn and then saved. There is no image file. Each one is built from a single starting number by a short piece of code that lives on the Ethereum blockchain, and it is rebuilt from that same number every time you look at it. The code is the artwork. The same code makes all 512, and the only thing that differs between them is that starting number.
That starting number is the . The number and the seed are the same thing, two words for one value. For each of the real 512, the seed is a long hexadecimal value, the kind that begins with 0x. The code the seed into a value it calls , and every choice about how the glyph looks, which symbols it uses, how dense it is, and which way it mirrors, falls out of a. Below you can run it yourself, watch why every glyph turns out symmetric, and read the table that decides what each one is made of.
From a number to a glyph
Type in a starting number, or hit randomise, and watch the glyph appear. Everything about how it looks, the symbols it uses, how dense it is, which way it mirrors, is decided by that one number. Nothing is stored. The picture is computed on the spot, by the same code that made the original 512.
- a = hash(seed)
- The hash turns the seed into a. 1123581321 goes in, 0xb10817…8e68be comes out. The same seed always gives the same a, and it cannot be run backwards.
- Symbol scheme
- a % 83 = 28. The slot 20 to 35 belongs to 2, so this glyph uses .
- Density
- = (a % 11) + 5 = 9 + 5 = 14. Higher mod, more blanks.
- Symmetry
- a % 2 = 0 and a % 3 = 2, so the grid folds top to bottom: horizontal symmetry.
Every glyph you make here is brand new and has almost certainly never existed, because it is not one of the 512. It belongs to the wider Latent Universe of glyphs the code could make.
What the number decides
Almost everything you see comes down to three decisions the code reads off the number. Here is each one worked through with real numbers.
The symbol scheme
a % 83 lands in one of ten slots, and wider slots are more likely. Example: a % 83 = 28 falls in the 20 to 35 slot, which is scheme 2, so the glyph is drawn with .
The scheme is the biggest thing your eye notices.
Symmetry
Two checks on a fold the grid: a % 2 = 1 folds left to right, and a % 3 = 2 folds top to bottom. A third fold runs corner to corner, across the diagonal: it swaps rows and columns, so cell (i, j) pairs with (j, i), and there a bar flips (a | becomes a -) while a slash stays. Example: a % 2 = 0, a % 3 = 2 gives a top to bottom fold, so the glyph mirrors horizontally.
This is why no Autoglyph is a random mess.
Density
mod = (a % 11) + 5, a number from 5 to 15. A cell inks only on the low values, so a higher mod leaves more blanks. Example: mod = 14 is high, so the glyph is sparse.
This is the difference between a heavy woven glyph and a near-empty one, and what makes the rarest glyphs rare.
Every glyph is symmetric. Here is the proof.
There are no random-looking Autoglyphs. Every one mirrors itself, because the number folds the grid before it is filled. Fold the glyph along an axis and watch the ink land back on itself. Load any glyph by number, or try a random one.
#28
- Structural
- All
- the grid is built as a mirror
- Visual
- All
- whether it still reads mirrored once drawn
Fold axis
Folds onto itself.
The mirror that drawing hides
Every Autoglyph is built as an exact mirror in its characters. Drawn, though, a slash or a bar flips as it folds, a folded / landing as a \, so some glyphs read as broken even though the structure is perfect. Autoglyph #51 is a flawless top-to-bottom mirror, yet drawn it looks off: fold it and the flipped strokes turn red. Switch to Data URI and the raw characters fold onto themselves exactly. That is the hidden mirror.
Structural Horizontal, drawn reads broken
Fold axis
The strokes flip when folded, so the drawing reads broken. Switch to Data URI to see the raw characters fold exactly.
The ten schemes
The scheme is chosen by where the number lands in a range of 83. Wider slots were more likely to come up, which is why some schemes are common and some are rare. Designed is the chance each scheme had; actual is how the 512 really turned out, set right beside it, with the raw count alongside. Those actual figures are not published anywhere else; this site computes them from the glyphs themselves.
The sparsest Autoglyph ever minted, and at a glance it reads as a creature. We call it the bear.