This is an old revision of the document!
Site Generation
Site #loc types
| flag | inspector name | actual check |
|---|---|---|
| 1 | plains | None of the following set: sea, highland, swamp, waste, forest, cave, mountains |
| 2 | forest | forest, not sea |
| 4 | mountain | Either mountains or highland, not sea |
| 8 | waste | waste (can be sea) |
| 32 | sea | requires sea |
| 64 | coast | coast (no other terrain type specified) |
| 128 | swamp | swamp, not sea |
| 256 | deep sea | deep sea |
| 512 | cave | cave (can be sea) |
| 1024 | n/a | not EA |
| 2048 | n/a | not MA |
| 4096 | n/a | not LA |
| 16384 | unique | unique |
| 32768 | n/a | generates in caves only, regardless of other parameters - could be used to target coastal caves only |
| 65536 | n/a | sea, not forest |
| 131072 | n/a | sea, not deep sea |
Terrain local modifiers (in the order they are checked in)
| terrain | mod |
|---|---|
| manysites | +25 |
| throne | +20 (stacks with other terrains) |
| farm | -20 (to a minimum of 5) |
| waste | +20 |
| cave | +15 |
| kelp forest | +10 |
| forest | +10 |
| mountains | + 10 |
| highland and UW | +15 |
| highland | +5 |
| deep sea | +10 |
First, generate sites according the the site flags on the terrain (these are 2^(13 + path id)). If multiple are set, lower ID paths will take precedence. For generation with this method, there is a 5% chance that the site is rarity 2, a (19/80) probability of it being rarity 1, otherwise it is rarity 0. If none of these terrain flags are set, skip this step, otherwise make 100 attempts to generate a site, stopping after one is successful.
Then move onto normal generation, which is 8 rolls of the following:
- Roll a d100. If this is less than local site frequency, generate a site:
- 75% chance to be rarity 0.
- 1/40 to be rarity 2.
- Otherwise (9/40) to be rarity 1.
- Regardless of whether or not a site generated, multiply local site frequency by (75/100) and round down, then go back to the start of this list
I now see why I didn't get the exact mechanics on the data analysis method. The rarity distribution was spot on, though.