I did some further testing of Terra maps, and while I still can't tell you more about the placement algorithm of CS's, I had some related thoughts:
Are we overcomplicating things?
I told you what I like about Terra, how there's an option to continue playing peaceful/wide after midgame and how it represents the real-world colonization wave ingame. That's all still true.
But - We have a world divided in 3 parts right now:
- starting continent
- rest of the old world
- new world
It's the middle part that seems questionable to me. Do we need to specify that civs start together on one landmass? Especially if the mapscript doesn't naturally create a cohesive landmass that's large enough?
Don't get me wrong, your code that does the connection is awesome. It creates surprisingly naturally-looking "Panama" or "Suez" settings. And it is without doubt very useful in the opposite role - to avoid "locked Gibraltar" situations. We really don't want large bodies of water without ocean connection on every second map.
BUT:
Very often the mapscript now connects all major old world landmasses anyway. This means, Civs are often placed all across the old world anyway.
Why don't we allow Civs to start all across the old world (on multiple continents if needed) in the first place? The code that turns most of the old world into one landmass seems a bit of a detour to achieve a very similar goal.
Basically, terra games would play like slightly more crowded continents plus games before astronomy. After unlocking ocean-going ships, we would have another part of the map that allows continued peaceful expansion. Continents plus games are working very well anyway, we don't need to artificially depart further from them than necessary.
The timeframe between "Sailing" and "Astronomy" is probably not long enough to justify having another expansion wave inbetween. We can probably barely finish settling our own surroundings (first expansion wave) before we unlock Astronommy (final expasion wave). Why such an effort to create another middle wave? Note that the map script creates many (awesome-looking) archipelagos like the region around indonesia. Civs adjacent to such areas will have a chance to settle them very early regardless.
I feel a bit bad for the effort you had creating those functions I criticize now, sorry for that! I liked them initially, and didn't come to this conclusion before creating a bunch of test maps yesterday. And I admit that I was trapped a bit in the "more complexity is always better" mindset.
A few more detailed thoughts:
Connected landmasses do have the advantage that AI's have a chronic weakness in terms of naval warfare and invasions, so
in theory the connections should help the AI. But we need to boost the AI in this regard anyway for "normal" communitas or continents+ maps. I also fear that narrow single-tile bottleneck landbridges will often cripple their efforts more than crossing water. The AI's would probably often prefer the land way at all costs, where they're often stuck and need embarking anyway. The human could abuse this and attack their forces while they're in water or simply block the landbridge with very few units and a citadel. I think the AI's would be more dangerous when we can't predict where they will exactly attack. I also think they should be better at invading us when they "know" they have to do a naval attack. This makes it more likely they realize they need an escort of warships. If they are accidentally forced to embark due to the narrowness of the landbridge, they probably won't have ships ready.
I generally think the human can much better cope with complex situations like needing two navies (one for the east, one for the west coast). The human would probably try to put a city on the landbridge to connect the two bodies of water and avoid this issue. I doubt we'll have an AI that understands the value of Panama-style channels anytime soon.
Apart from the AI, there's the issue of isolated starts. While no civ would start in a place where they don't meet trading partners before Astronomy, we should probably avoid starting positions on continents smaller than -say- 50 land tiles. Ideally, it should be rare to start on a landmass completely alone (without another major civ). But then again, this is a problem that applies for "normal" Communitas or C+ maps as well. I'm sure there's already a function in existence to avoid this.
We should also reevaluate map sizes when changing the distribution of starting positions. I think that the old world on its own should provide something around 20% less land tiles than on "normal" maps. This crowdedness should be enough to make oversea settling more appealing, but not enough to make every game end in an early world war like on overcrowded Pangea maps.
The new world on the other hand should offer plenty of space to explore and settle. Settling 50 tiles away from home is an considerable effort, and we should be rewarded with a plentiful selection of good spots for this. We also don't want that the first few civs to unlock Astronomy can grab all good spots within a few dozen turns.
Sorry for the wall of text, hope it helps a bit!
Yeah, we should be able to drop the graphics in a game folder somewhere to get it override the art without using mods.
It should be this folder:
C:\Program Files (x86)\Steam\SteamApps\common\Sid Meier's Civilization V\Resource\DX9
The fpk files in there are the big archives that contain the .dds files. I think we can just drop the .dds files between them.
Here's a very similar mod that changed te colour of the ugly vanilla trading posts:
link
The readme in there should describe what we need to know.
