Map gen notebook

Vanilla canvas experiments — seed + sliders + regenerate.

Noise weights → Dijkstra → dilate → path walkers

Random edge start/end (min separation), shortest path, dilated corridor, then walkers spawn every K steps along the path starting after K (not from the first cell) and wander randomly while carving.

Seed

Grid size Noise scale Noise octaves Noise amplitude White noise Dilate Min edge separation Walker spacing Turn probability Walker max steps

Hierarchical walkers

Parents stamp a 3×3 each step; children stamp a single cell (1×1). Two colours distinguish generations. All parents start at the center. Separate step limits and turn rates per generation.

Seed

Grid size Parents Parent max steps Parent turn % Children per death Child max steps Child turn %

Volatility-of-volatility walker

Turn probability is “volatility”; each step it does a random walk with exponential-sized moves (−ln U) and a random sign, scaled by VoV. Then you turn with probability = current volatility. Each walker has its own RNG stream (seed + id) so the count of walkers does not change other walkers’ paths. All start at the centre; distinct colours.

Seed

Grid size Walkers Starting volatility VoV (jump scale) Steps per walker

Lerp walker

Starts at the centre. Each step, brush radius and turn probability (volatility) are linearly interpolated from start→end over the run. Carves a disk of the current radius, then may turn (with current volatility), then moves one cell.

Seed

Grid size Steps Start radius End radius Start volatility End volatility