Autoglyph #149
- Symbol scheme
- #9
- Structural symmetry
- Vertical
- Visual symmetry
- Vertical
- Density
- 238 marks (5.8%)
- mod
- 15
- Seed
- a
How this glyph was made
New to Autoglyphs? Start with Understand.
Every Autoglyph is built on-chain from one number by a short, fixed piece of code. No image is stored. Here is that code run with this glyph’s own numbers, step by step.
- 1
Start from the seed
Every glyph begins as one number, its , fixed when it was minted. This glyph’s seed is 0x0000000000000000000000009d1ed7a0aa1579fa3c2da515e8c8de8183a1afac. Change the seed and you get a completely different glyph.
- 2
Hash it to a
A seed is just a chosen number, and on its own it would let someone steer the result. Hashing it with , a standard scrambling method, turns it into a wild, evenly spread number that no one can predict or engineer. That scrambled number is : 0xa5649816e843d19d3def673f754dd20578a703dd. That is why two nearby seeds produce completely different glyphs.
- 3
Pick the symbol scheme
The decides which characters the glyph is drawn with. Here, a leaves a remainder of 81 when divided by 83 (written a % 83 = 81). That number, 83, is a row of slots shared among the 10 schemes, and a’s remainder lands in one of them: 81 falls in scheme 9’s slot. Scheme 9 is the single-mark scheme, drawn only with , and it is rare because its slot is narrow, about 2 slots out of 83, so few seeds ever land there.
- 4
Set the density
The sets how often a cell inks: a higher mod leaves more blanks, a lower one fills more in. Here mod = (a % 11) + 5 = 10 + 5 = 15, the top of the range (5 to 15), which is why this glyph is sparse: 238 of its 4,096 cells are marked.
- 5
Set the symmetry
Before the grid is filled, it is folded so it mirrors, which is why no Autoglyph is a random mess. Two more reads of a set the folds: a % 2 = 1 folds it left to right, and a % 3 = 0 adds no top-to-bottom fold. So this glyph is vertically symmetric.
- 6
Fill the grid
Only now does the code draw. It walks all 64 by 64 cells, combining each cell’s position with a to pick a character or leave it blank. The same arithmetic produced all 512 glyphs; only the seed changed.
A curiosity: every glyph’s seed was chosen by its minter and passed to the contract, where it is hashed, so the seed cannot steer the result; the scheme, density and symmetry all fall out of the hash. Most minters used a throwaway number, but a few chose seeds with meaning. Autoglyph #337 was minted with the seed 1123581321, the start of the Fibonacci sequence (1, 1, 2, 3, 5, 8, 13, 21) written as a single number, and #198 with the round number 200. The look still came from the hash, not the chosen seed, so a meaningful seed is a signature, not a steering wheel.
Symmetry
Fold the glyph and the ink lands back on itself. Switch to Data URI to fold the raw characters instead of the drawn strokes.
- Structural
- Vertical
- Visual
- Vertical
Fold axis
Notes & provenance: the bear
Of the 512 Autoglyphs, this is the sparsest minted, and at a glance it reads as an animal seen head-on. We call it the bear: 238 marks on a grid of 4,096, drawn with a single character, and almost entirely empty space.
Nobody designed it to look like anything. Like every Autoglyph, it was generated from a number by a short piece of code with no picture in mind. What follows is an honest account of two separate things: why it is genuinely rare, which we can measure exactly, and why it reads as a creature, which we cannot.
Figurative
The “bear.” Out of sparse, symmetric marks it reads as an animal face, the clearest figurative image in the 512.
The sparsest that was minted
#149 is the sparsest of the 512. It belongs to scheme 9, the single-mark scheme, which is the sparsest of the ten symbol types: a cell inks only on one exact value, so most of the grid stays empty. At 238 marks it fills 5.8% of the grid.
But sparsest of the 512 is not the same as sparsest possible. Sample the whole space of starting numbers and about 1 in 1,000 render sparser than the bear. The true floor is zero, a completely blank glyph the generator can make but no one minted. So the bear sits at the far edge of what exists, near the 99.9th percentile of sparseness, without being the edge of what was possible.
What the marks actually do
Here is what the algorithm happened to produce, measured from the glyph itself.
First, it is exactly mirror-symmetric left to right. Fold it down the centre and every mark lands on its partner. That single vertical axis is the first thing the human visual system looks for in a face or an animal.
Second, and this is the part that does the work, the ink is not spread evenly. Each upper corner carries 9.5% of its cells inked. The space between them, across the top, carries just 2.7%. That gap is 3.5 times sparser than the masses on either side, and it is what separates the top of the glyph into two distinct shapes. They sit wide apart, at the upper-left and upper-right, the way a bear’s ears point out to the sides. Below them the middle of the glyph is emptier than the glyph as a whole, a quiet central space where a face reads.
That is the whole of what the structure supplies: a mirror axis, two wide masses up top, and a clear gap between them. It is close to the smallest arrangement a person will read as an animal face. Everything else, the eyes, the muzzle, the expression, the species, your visual system fills in.
The measured masses: each upper corner against the sparse gap between them, around a quiet central space. Toggle the face schema to see the animal-face layout these fall into.
Seeing animals in marks
This is not a quirk of this one glyph. The human visual system runs a fast, broadly tuned detector for faces and animal faces, one that fires on the barest configuration, two masses above a centre, and is biased toward symmetric patterns. It is deliberately over-eager: better to mistake a shadow for a creature than to miss a real one. When it meets a sparse symmetric pattern, it groups the scattered marks into masses, bridges the gaps into edges, and reads the result against its template. Researchers describe this as a detector that privileges sensitivity over selectivity, which is to say it would rather see a face that is not there than miss one that is.
This is why so few marks can evoke a creature. A handful of well-placed shapes, the panda on a conservation logo, an animal in a few strokes of ink, is enough, because the viewer completes the rest. The bear is that effect, produced by arithmetic instead of a designer.
It has happened in generative art before. The single Ringers piece that resolved into a bird, the Goose, sold at auction for millions, with its accidental likeness named as the reason. The bear is the same effect arriving from the opposite direction, out of extreme emptiness rather than density.
So “the bear” is our name for it: a reading our visual system supplies, not a measured classification, and a label we have found the wider community using. We can measure the symmetry and the gap exactly; we cannot put a number on looking like a bear. The structure gives symmetry and two masses and a gap. The animal is supplied by you.
Related glyphs
Nearest in shape
A computed similarity: the glyphs whose overall composition is closest to this one, measured across a few simple traits (how much ink there is, where it sits, and how broken up and spread out it is). A rough visual guide, not an official grouping.
Same scheme, nearest density
A secondary view: same symbol scheme, closest mark count.