I've been working on a game option that chooses civs based on the starting tiles and their surroundings for all players whose civ was set to "Random" during game setup. Screenshots attached (9 in one big jpeg), and the current state of AdvCiv 1.05 with this option included can be downloaded from GitHub (unzipped folder needs to be renamed to just "AdvCiv"; and the new option need to be checked on the Custom game screen). Edit (March): Link obsolete as v1.05, incl. the new option, has been released.
I've named the option "True Starts" (TS); that (or TSL for "true starting locations") seems to be the term that Civ 5 and 6 players have come to use for scenarios with historical starting locations. TS is similar to Culturally Linked Starts, which, however, only tries to recreate the geographical distances between civs. (See the "map_enhanced" screenshot in that thread – no desert at all where the "Middle East" civs have been placed.)
The difference from the starting biases in Civ 6 (and Civ 5) is that the TS option picks the most fitting civs from among all 34, whereas Civ 6 picks a civ for each of the, say, 8 players uniformly at random and then looks for the best assignment of the 8 starting tiles to those 8 players. Moreover, TS is not intended as a mere "bias" but as a best effort to match each civ's historical geography. That said, like the Civ 6 bias, TS does not affect the map generation process (for the most part, see below), nor the placement of the starting locations.
TS does swap bonus resources around as a final step after a civ has been assigned to each starting location, e.g. (excerpt from logfile)
Ideally, these swaps should not make the map look unnatural, nor affect the balance of the starting locations. TS also makes it less likely for land to be placed near the equator on Fractal-based maps – to have more starting locations suitable for the many European civs in the game. (As my screenshots demonstrate, this does not rule out maps with a lot of jungle.)
I'm still tweaking things in order to accomplish a more even frequency distribution among the civs over a series of games. My latest stats about that:
I've done a decent amount of smoke testing with unusual map scripts and scenarios, but not much AI Auto Play testing lately, so some issues might occur when trying to actually play a game with this development version of the mod. I haven't yet played a game with the TS option myself yet, so I don't know if the semi-realistic geography really enhances the experience. I did play a game a while ago in which the (AI-controlled) Maya randomly started near the equator and that was quite a frisson to me – which is what prompted me to revisit this idea I had 2 or 3 years ago.
I've named the option "True Starts" (TS); that (or TSL for "true starting locations") seems to be the term that Civ 5 and 6 players have come to use for scenarios with historical starting locations. TS is similar to Culturally Linked Starts, which, however, only tries to recreate the geographical distances between civs. (See the "map_enhanced" screenshot in that thread – no desert at all where the "Middle East" civs have been placed.)
The difference from the starting biases in Civ 6 (and Civ 5) is that the TS option picks the most fitting civs from among all 34, whereas Civ 6 picks a civ for each of the, say, 8 players uniformly at random and then looks for the best assignment of the 8 starting tiles to those 8 players. Moreover, TS is not intended as a mere "bias" but as a best effort to match each civ's historical geography. That said, like the Civ 6 bias, TS does not affect the map generation process (for the most part, see below), nor the placement of the starting locations.
TS does swap bonus resources around as a final step after a civ has been assigned to each starting location, e.g. (excerpt from logfile)
Spoiler :
Swapping bonus resources Silk (38,33) and Wine (49,12)
Removing feature Forest from (38,33)
Moving Silk to adj. destination (48,13)
Removing feature Forest from (38,33)
Moving Silk to adj. destination (48,13)
I'm still tweaking things in order to accomplish a more even frequency distribution among the civs over a series of games. My latest stats about that:
Spoiler :
Number of occurrences per civ and leader on a Fractal map with default settings, regenerated 9 times, i.e. 10x8 occurrences in total.
Most of those no-shows had been occurring before, some even too frequently. Getting Sumer and Arabia to appear more than once in a blue moon is probably a lost cause on this map script. Netherlands and Rome yet to be investigated. The rest I can probably get to work by adjusting civ bias values in my XML files.
Code:
Maya 7
Russia 7
Catherine 4
Peter 3
Stalin 0
Greek 6
Pericles 3
Alexander 3
Persia 6
Darius 4
Cyrus 2
Ethiopia 5
Celts 5
Brennus 4
Boudica 1
German 5
Frederick 4
Bismarck 1
China 5
Qin 4
Mao 1
Khmer 4
Viking 3
Zulu 3
Portugal 3
Carthage 3
Aztec 3
Mongol 3
Kublai 2
Genghis 2
India 3
Ashoka 2
Gandhi 1
Mali 2
Japan 2
Inca 2
America 1
Roosevelt 1
Washington 0
Lincoln 0
Korea 1
Native American 1
-- 22 have occurred, 12 haven't:
Babylon 0
Sumer 0
Arabia 0
Egypt 0
Byzantium 0
Ottoman 0
Rome 0
HRE 0
Spain 0
France 0
England 0
Netherlands 0
Attachments
Last edited: