DESIGN: Terrain Values

[Peter Triggs]

Maq, while you're doing this, another thing to look at ( ) is the 'score' parameter. The AI uses this to decide where to place it's cities. The bad terrains have low scores:

tundra - 0
glacier - 0

polar hill - 10
polar mountain - 10
swamp - 10
desert - 10

desert hill - 20
desert mountain - 20

The highest score is given to beach (90) so that the AI will build coastal cities and hence ships. (So I think you should leave beach production at 10. The AI will build cities next to beaches anyway, so lowering their production penalizes it. In fact you could even try increasing it to 15.)

We all know that the AI builds cities in some pretty strange places. It occured to me that one way to prevent this might be to lower the scores for the bad terrains, i.e. we give them negative scores. What do you think?

[Maquiladora]

I was only testing my own performance on this terrain to see which was the absolute best way to play, then alter the terrain more and more to force me to choose a strategy every game, rather than playing the same "plains" game everytime.

Also the AI is lacking in other areas i didnt think it was worth it to look into yet.

My general feeling is that AI should stay well clear of bad terrain upto to a certain amount of turns, just like a human would. So if after searching for 10 turns through Jungle, it doesnt find Plains or Grassland, it should settle anyway where its "seen" the best place, like next to coast or mountains. Doesnt something like this already happen?

Do we know the absolute maximum terrain score you can give?

As for beaches, you certainly have a point i didnt consider, i admit most of the time i was looking from a purely human v human point of view, and the best ways to force different strategies from game to game. The only thing i dont like about your suggestion is that it makes beaches the strongest early terrain again, and i really wanted to eliminate any "super terrains".

Beaches can accept Nets and Ports, so they dont increase in production ever (using current tile imps), so there is a point for increasing natural prod on beaches..... another way is to lower natural production on beaches and give Ports a production bonus...... the latter seems more realistic IMO, and eliminates the "super terrain" factor from beaches. It also pushes the player to improve their own land, which seperates more the good player (placing good tile imps) from the lucky player (getting good start land... plains/beach).

[Peter Triggs]

Doesnt something like this already happen?

Probably, but I'm not sure. I know where this stuff is, and I've glanced over it, but not in any detail. It seems to evaluate either the entire map or a portion of it, and make a list of settle targets. I don't know how often it does this.

Do we know the absolute maximum terrain score you can give?

When it evaluates a location, there's a radius involved and the maximum score will depend on how big the radius is. But I don't know about the details. OTOH, in each default strategy there's a MinSettleScore of 600 and it won't settle in locations that score less than this.

give Ports a production bonus

I like that idea. The only problem is that with the present system I don't think it's possible to get the AI to build both Ports and Nets on Beaches. Hopefully, though, we can tweak it so that it builds Ports on Beach tiles that are adjacent to its cities, but Nets on other Beach tiles.

[Maquiladora]

When it evaluates a location, there's a radius involved and the maximum score will depend on how big the radius is. But I don't know about the details. OTOH, in each default strategy there's a MinSettleScore of 600 and it won't settle in locations that score less than this.

Ah yeah thats it, that sounds like we can tweak that pretty good if we know how big each radius is.... what im worried about is how to control the AI's overlapping of cities, even on the second border in some cases, maybe we can somehow tell it not to count the terrain "score" if its inside another cities border...... I know we can increase the settle distance, but thats too random IMO and wastes alot of good land if its too high, of course this wont matter as much if trade goods give extra resources etc etc... anyway this is all off topic ramblings.

[Tamerlin]

Martin, even if these new values are tested and seems to improve the game do you really think they should be included in the source pack?

As far as I am concerned it would be better to consider them as an optional mod...

[Martin Gühmann]

Originally posted by Tamerlin
Martin, even if these new values are tested and seems to improve the game do you really think they should be included in the source pack?

We want to improve the game and something can be very easily improved by just altering the text files therefore we have to include also text files in the alterd source pack, and by the way Activision's patch also contains altered text files.

Originally posted by Peter Triggs
We all know that the AI builds cities in some pretty strange places. It occured to me that one way to prevent this might be to lower the scores for the bad terrains, i.e. we give them negative scores. What do you think?

Alternativly we could increase the other values, but I think the negative values are a better idea so that we don't need to alter the Settle Score in the strategies.txt if we don't need to do anyways.

-Martin

[Maquiladora]

Originally posted by Tamerlin
Martin, even if these new values are tested and seems to improve the game do you really think they should be included in the source pack?

If they improve the game, why not?

As far as I am concerned it would be better to consider them as an optional mod...

What dont you like about the changes Tamerlin? Or is it just the fact you dont want any terrain changes? Im open to suggestions.....

[Tamerlin]

Originally posted by Maquiladora

If they improve the game, why not?

Because I think that the patch should be as close as possible to the original game. I can only say that I am happy with the SAP2 values.

What dont you like about the changes Tamerlin? Or is it just the fact you dont want any terrain changes? Im open to suggestions.....

I can't say I don't like your terrain value changes as I would have to test them before giving my opinion. I just think that such changes can modify the rythm of the game or unbalance it.

We all know that the real problem is the AI that is not able to use the terrain properly, changing the values does not solve it.

What I really like in the SAP2 and the other mods is that, thanks to modswapper, they are optional. To change the values in the sourcepack is to change them permanently without giving any choice to the end user.

[Maquiladora]

Originally posted by Tamerlin
Because I think that the patch should be as close as possible to the original game.

It depends which # patch we're talking about. Testing these values i had in mind for a #2 sourcecode patch, something like

#1 Fixes Bugs adds new SLIC thingys and stuff that the 1.1 patch included etc etc, generally "under the bonnet" stuff.
#2 As #1 + Fixes all imbalances in gameplay (hopefully) <<
#3 whatever

[Tamerlin]

Originally posted by Maquiladora
Testing these values i had in mind for a #2 sourcecode patch, something like.

Then I can only agree with you.

#1 Fixes Bugs adds new SLIC thingys and stuff that the 1.1 patch included etc etc, generally "under the bonnet" stuff.
#2 As #1 + Fixes all imbalances in gameplay (hopefully) <<
#3 whatever

This is exactly what has been already discussed earlier in another thread and this is a planning I agree with. I would also like the #1 patch to include changes in the AI so that the patched original game is at least as competitive as the SAP2.

[Maquiladora]

BUMP

Okay after "some" testing and abit of thought ive come to the conclusion my changes were far too drastic, far too drastic in fact they totally unbalanced the rest of the game, and would only create even more balancing work in other areas (tile imps, buildings etc), which is bad.

Two changes i would stick with though,

Forests - Were 10/5/10 - Recommend 5/15/5.
Dale (or whoever had the idea for this) was right. For reasons ive mentioned before about the difficulty in improving Forests with terraform for a long time, i think they should contain mainly production and NOT commerce, because production is far more important than commerce is BEFORE you have the ability to terraform forests. Which makes forests a nightmare early game, when they should be a lush natural resource and a good defence.

Beaches - Were 10/10/10 - Recommend 10/0/10
Another one from SAP. As mentioned the biggest impact would be on sea units but i think beaches are too good for what they ARE. They are sand, seaweed, shells, used condoms, broken glass and surf. If we want to improve production on beaches we can make Ports give production aswell as Drilling Platforms, both of which are more realistic too IMO.

My only concern with this is, when to build Nets instead of covering your whole coastline with Ports. ANY production improving TI's on the right terrain (ie mines on mountains) is essential WHEREVER you can put them, which isnt good news for Nets and other sea TI's. Im not sure if this can be balanced by TI cost or not, yet.

[tyrantpimp]

Whatever you do to tiles i hope its an option the player can turn off or on. Not everyones so eager to see that tweaked. . .

[MrBaggins]

Can you explain any benefit to NOT tweaking these, by default, though?

Whats so wonderful about the originally designed settings?

They had definite and absolute downsides... like.. for instance, making forest a very bad initial building site.

The AI is typically the one who gets hurt by this, and it thus makes AI improvement more difficult.

[tyrantpimp]

NOpe cant explain it in anyway but ive never had any problems with terrain. But i still am not convinced of the reasons why it needs tweaked.

[MrBaggins]

Have you played the Apolyton pack? Do you dislike the Apolyton Pack? Are the terrain effects better or worse than default?

If so, can you explain how the changes in the Apolyton Pack adversely effect your play?