Name lab
Generate names in a culture's style, on demand — for NPCs at the table, and for telling whether a culture's name lists are any good yet. Turn the knobs, watch the names change.
Culture
208 names across family, feminine, masculine, feminineNoble, masculineNoble
Name listswhat the Markov steps learn from
- Elanna Gormsley
- Ande Hanell
- Halen Sardallem
- Lumina Hanell
- Venari Stonewick
- Vidin Calmes
- Fare Irleman
- Adremei Bramwick
- Miolo Enderley
- Vere Montantz
- Corve Wyndley
- Vale Brentmoor
- Delme Underfalc
- Faren Hamswat
- Solmir Thingley
- Zolann Ivorrantz
- Eloma Cort
- Deliona Gredfol
- Kaldi Gredfo
- Ulmari Calmeshe
Template
The finished name, with holes in it. Every {name} runs the step called name; everything else is literal, so {first} sul {last} puts sul in the middle of every name. Text introducing a hole goes with it — if a step produces nothing that time round, its lead-in disappears too.
Steps
learns the shape of a name list
lists:
Higher order copies the source list more closely — 2 invents, 4 quotes. Stop bias pulls names short. A Markov chain needs a dozen or so names before it sounds like anything; below that it mostly quotes. Weights on a list mean nothing here — a chain trains on the words.
learns the shape of a name list
lists:
Higher order copies the source list more closely — 2 invents, 4 quotes. Stop bias pulls names short. A Markov chain needs a dozen or so names before it sounds like anything; below that it mostly quotes. Weights on a list mean nothing here — a chain trains on the words.
learns the shape of a name list
lists:
Higher order copies the source list more closely — 2 invents, 4 quotes. Stop bias pulls names short. A Markov chain needs a dozen or so names before it sounds like anything; below that it mostly quotes. Weights on a list mean nothing here — a chain trains on the words.
learns the shape of a name list
lists:
Higher order copies the source list more closely — 2 invents, 4 quotes. Stop bias pulls names short. A Markov chain needs a dozen or so names before it sounds like anything; below that it mostly quotes. Weights on a list mean nothing here — a chain trains on the words.
learns the shape of a name list
lists:
Higher order copies the source list more closely — 2 invents, 4 quotes. Stop bias pulls names short. A Markov chain needs a dozen or so names before it sounds like anything; below that it mostly quotes. Weights on a list mean nothing here — a chain trains on the words.
Paste this into the culture
Indented to sit under - slug: alberatoran in content/*.yaml — the whole of a culture's naming, lists included. This is the only way to keep what you've made: the page never writes to the database, because the next content import would overwrite it.