A digital cryptid is a creature that is generated by an AI that has no real-world counterpart that said creature could be based on. The first digital cryptic to be discovered was the crungus. The crungus was a grotesque, hairy humanoid creature created by DALL·E v1 in response to the prompt crungus. (DALL·E v2 does not produce the cryptid when given the same prompt.)
The second digital cryptid to be discovered was Loab. Loab is a middle-aged woman with brown hair and rosacea, frequently in disturbing or gory scenes. The discovery of Loab was a bit more complicated. Rather than give the (unidentified) AI a gibberish prompt, the woman who first found the Loab gave it several negative prompts.
Yesterday morning I woke up with a word in my head: mesker. Just for fun, I fed it into (Imagine v3 by vyro.ai). What I got back appears to be a new kind of cryptid:
The Mesker
The results were surprisingly consistent for a nonsense word: a humanoid creature with pale skin, glowing eyes (usually), and a skeletal nose (usually). I searched the internet for "mesker" and didn't find anything that looked like this creature. When I tried the same prompt with Imagine v4 Creative by vyro.ai, I got back an array of images that were interesting, but not thematically coherent:
Not The Mesker
Imagine v1 and Imagine v4 (beta) by vyro.ai also gave inconsistent, non-cryptid results, as did a few other text-to-image generators that I tried. It appears that the mesker is unique to Imagine v3 by vyro.ai. So why did I wake up with that nonsense word in my head? I can only guess that the universe really wanted the existence of the mesker to be revealed.
Here's what I get if I pluralize the prompt to meskers for Imagine v3:
Since I got such an interesting result from mesker, I decided to see what I would get from using crungus as a prompt. Here's what I got from Imagine v3:
The New Crungus
It's nothing like the crungus produced by DALLE, but it's still a consistent output: a gnome creature with a yellow nose, glasses, and something around its neck (a tie, a neckerchief, or something slimy), and bits of food scattered on the floor at its feet.
When I tried the same prompt with Imagine v4 Creative by vyro.ai, it consistently drew dogs, and sometimes they were wearing accessories:
Not The Crungus
I wasn't done, yet, though. Something about the spilled food at the bottom of the images of the new crungus made me think of muffins. So I asked several versions of Imagine by vyro.ai to draw a muffin man. Here's what it came up with:
The Muffin Man
These illustrations were drawn using Stable Diffusion 2.1.
