Caveman 2 Cosmos (ideas/discussions thread)

[Harrier]

@ Koshling:

OK i am trying to get to the correct # that would show ALL the buildings again, plus defenses and wonders etc etc . . .

But i cant get it to work correctly, in the attached i have what it is set on now: 3 that as you say
"<!-- Use the following to determine what building art is displayed on cities. It is a bitmask of flags which may be added together:
1 = Wonders
2 = city defences
128 = other
We default to Wonders and city defenses (1+2 = 3)
-->"
ok i added 2-128=130 (pic 2) and see results not that clear???
then i did 3+128-131 (pic3) and as you can see "exactly" the same???

For some reason i do NOT believe this is working correctly??

If you only have those three settings you could use:

0 = display nothing. (not sure what that would display!!)
1 = display wonders only.
2 = display city defenses only
3 = display wonders and city defenses.
128 = display other only. (what is other by the way? [probably buildings])
129 = display other and wonders only.
130 = display other and city defenses only.
131 = display other and wonders and city defenses.

No other settings are valid.


So you are right it is not working properly.

Pic 1 seems to show everything (setting 131).
Pic 2 and 3 are identical. Buildings and defenses only I assume. (setting 130).

 

[charlesb96]

Does the rebuild-able wonders have to be all or nothing? If the Golden Gate Bridge was destroyed, I am sure it would be rebuilt, it is too necessary for the area, but rebuilding the Pyramids is not an option. The WTC was rebuilt, but in a different form. Most Wonders are considered Wonders because of age, a rebuilt Circus Maximus is just a casino. We could get a list of suggestions from everyone as to what Wonders they would think could be rebuilt.

 

[charlesb96]

Is there some way for the AI to also consider number of units in a stack when deciding whether or not to attack. I have one unit sitting next to an AI SOD of 50 units but it never attacks, because in a one on one fight I have 100% chance to win. But I could not win 50 times in a row, eventually my unit will take enough damage and die.

 

[Harrier]

Does the rebuild-able wonders have to be all or nothing? If the Golden Gate Bridge was destroyed, I am sure it would be rebuilt, it is too necessary for the area, but rebuilding the Pyramids is not an option.

If the Golden Gate Bridge was destroyed and replaced - the replacement would be just that a replacement - not a wonder. (though I am not sure if the GGB is a wonder! in civ4 terms - it is a tourist attraction for people who are in the area, but I doubt that people travel to S.F. just to see the bridge.)

As for the replacement Pyramids - again it would be a tourist novelty not a wonder.

So I vote no to replaceable wonders in the base game. If included make them an option in bug and a mod addon. Not part of the base game.

I can see the reason for wanting them - the game benefits (extra this or that) - but not realistic imo.

Saving memory is more important.

 

[Thunderbrd]

Is there some way for the AI to also consider number of units in a stack when deciding whether or not to attack. I have one unit sitting next to an AI SOD of 50 units but it never attacks, because in a one on one fight I have 100% chance to win. But I could not win 50 times in a row, eventually my unit will take enough damage and die.

IIRC, the way Koshling set that up is that if there is significant REASON to attack that unit at that time then it very well could. For example, if the unit was protecting some workers or a settler or a a city or a resource position that the AI found valuable to raid, then it might. But it's not going to waste units on attacking a unit that isn't protecting anything of key importance unless it feels it has to. If the unit itself is a global class unit and its an opportunity to kill it then it might also value the attack at that point but if it's something you can just build at any time, it's not going to waste the strength.

 

[BlueGenie]

Maybe reconsider that evaluation slightly if the offending unit can cherrypick the AI's units off one by one. After all it'll just be a few turns until the AI has lost more units via cherrypicking than it would lose via attacking and taking it out.

Cheers

 

[Thunderbrd]

Maybe reconsider that evaluation slightly if the offending unit can cherrypick the AI's units off one by one. After all it'll just be a few turns until the AI has lost more units via cherrypicking than it would lose via attacking and taking it out.

Cheers

Not sure what Koshling had in mind with this kind of scenario or if it was considered - I know it seems to be a weakness. I would think they'd be moving on to a more defensible position unless they're there for good reason.

FYI: This level of combat evaluation is currently far outside the scope of what I've come to understand in the code. I COULD possibly TRY to evaluate and understand the coding and what's going on there and see if there's a bug or something that needs to change but I figure it would take me a whole cycle just to figure it out enough to understand it let alone fix it or adjust it. But I usually end up running across small portions of AI and coming to understand them in small bites depending on the project and perhaps eventually I'll come to grasp this kind of strategic AI stuff... I'm having a hard enough time with just getting the AI to build more siege weapons at the moment.

 

[Koshling]

@ Koshling:

OK i am trying to get to the correct # that would show ALL the buildings again, plus defenses and wonders etc etc . . .

But i cant get it to work correctly, in the attached i have what it is set on now: 3 that as you say
"<!-- Use the following to determine what building art is displayed on cities. It is a bitmask of flags which may be added together:
1 = Wonders
2 = city defences
128 = other
We default to Wonders and city defenses (1+2 = 3)
-->"
ok i added 2-128=130 (pic 2) and see results not that clear???
then i did 3+128-131 (pic3) and as you can see "exactly" the same???

For some reason i do NOT believe this is working correctly??

What is picture 1 (which I assume is what you want?). From what I can see it includes a wonder that is absent from pictures 2 & 3. It should be absent from (2) (130 doesn't include the wonders flag) but it is definitely wrong that it is missing from picture (3). Is this happening with all wonders or is it specific to a few? How did you generate picture (1)?

 

[strategyonly]

What is picture 1 (which I assume is what you want?). From what I can see it includes a wonder that is absent from pictures 2 & 3. It should be absent from (2) (130 doesn't include the wonders flag) but it is definitely wrong that it is missing from picture (3). Is this happening with all wonders or is it specific to a few? How did you generate picture (1)?

OK i labeled each pic with the # pertaining to each, so pic 1 is the "default" we currently have (3). (not what i want)

pic 3 labeled 131 should be all included, 1- Wonder, 2 - defenses, and 128 - buildings, hence 131?? correct, but it is NOT working???? as indicated

 

[Nimek]

@Koshling

You done great job for the Civ community Thank you.

Do you plan anytime soon other improvements in AI and turn times?

 

[makotech222]

Some ideas for C2C:

1. Better handling of obsoleting buildings and units: I'm in the classic age, and i can still build atl-atls, stone macemen, and other units, while being able to build composite bowmen and other advanced units. The old ones should be obsolete as soon as the new ones are unlocked to clean up the massive clutter of build lists.

Same goes for buildings. I'm in the classic age, i shouldn't need to use bark gatherers and berry gatherers, etc. etc.

2. Cleanly separate modules, so that some can be removed without Civ4 freaking out. For example, i want to disable the -Punk modules, and some of the useless buildings like sports and holidays. When i disable them in the XML, Civ4 brings up dozens of error messages and eventually fails.

I like a lot of the extra features in C2C, but there's just so many things that aren't necessary, but bog down game play and turn times.

 

[Dancing Hoskuld]

Some ideas for C2C:

1. Better handling of obsoleting buildings and units: I'm in the classic age, and i can still build atl-atls, stone macemen, and other units, while being able to build composite bowmen and other advanced units. The old ones should be obsolete as soon as the new ones are unlocked to clean up the massive clutter of build lists.

Same goes for buildings. I'm in the classic age, i shouldn't need to use bark gatherers and berry gatherers, etc. etc.

2. Cleanly separate modules, so that some can be removed without Civ4 freaking out. For example, i want to disable the -Punk modules, and some of the useless buildings like sports and holidays. When i disable them in the XML, Civ4 brings up dozens of error messages and eventually fails.

I like a lot of the extra features in C2C, but there's just so many things that aren't necessary, but bog down game play and turn times.

1) Units - They do but only when you have access to all the units thy upgrade to. This "problem" is due to the convoluted upgrade paths of some units.

Buildings
- Once they are obsolete they are not displayed. If you can see berry bushes then you don't have agriculture yet.
- I have tried to get many of the early and cheap buildings like the bark gatherer to be free in new cities. This does not always work because their requirements are not there before the free bit happens. I do want to change the way this works eventually.

2) If you are getting errors when you are turning of an optional mod please report it. But do it one mod at a time. You should be ale to turn each -punk off. If you can't then it is a bug and we will fix it.

 

[God-Emperor]

What is picture 1 (which I assume is what you want?). From what I can see it includes a wonder that is absent from pictures 2 & 3. It should be absent from (2) (130 doesn't include the wonders flag) but it is definitely wrong that it is missing from picture (3). Is this happening with all wonders or is it specific to a few? How did you generate picture (1)?

OK i labeled each pic with the # pertaining to each, so pic 1 is the "default" we currently have (3). (not what i want)

pic 3 labeled 131 should be all included, 1- Wonder, 2 - defenses, and 128 - buildings, hence 131?? correct, but it is NOT working???? as indicated

Pics 2 and 3 appear to be from somewhat earlier in the game. Was there a wonder in the city at that time? If not, that is why it is not showing up...

If you turn on the "all other buildings" value, 128, what I think you will get is some big empty plazas and fewer buildings showing up in the city. The reason for this is that most buildings in C2C do not have any model to put on the map, but the city building layout code doesn't skip the no-model buildings so it picks buildings to show and then shows them - which means it shows nothing but the underlying city texture in most cases, leaving an empty space in the city. It also sees that there are a ton of actual buildings in the city so it adds fewer of the generic buildings from the city set which leaves the city with fewer buildings showing overall (with 128 off most of the buildings you see are generic filler buildings from the city set it is using).

I can see 2 ways to avoid this.
1) Define models for a lot more of the buildings.
2) Modify the layout code to not pick buildings that don't have models defined for them, if it can avoid it. Probably also counting the buildings that have models instead of all the buildings in the city when determining what percentage of the city should be real buildings vs. generic buildings.

That is assuming my vague idea of how the city layout works is correct...

 

[strategyonly]

Pics 2 and 3 appear to be from somewhat earlier in the game. Was there a wonder in the city at that time? If not, that is why it is not showing up...

If you turn on the "all other buildings" value, 128, what I think you will get is some big empty plazas and fewer buildings showing up in the city. The reason for this is that most buildings in C2C do not have any model to put on the map, but the city building layout code doesn't skip the no-model buildings so it picks buildings to show and then shows them - which means it shows nothing but the underlying city texture in most cases, leaving an empty space in the city. It also sees that there are a ton of actual buildings in the city so it adds fewer of the generic buildings from the city set which leaves the city with fewer buildings showing overall (with 128 off most of the buildings you see are generic filler buildings from the city set it is using).

I can see 2 ways to avoid this.
1) Define models for a lot more of the buildings.
2) Modify the layout code to not pick buildings that don't have models defined for them, if it can avoid it. Probably also counting the buildings that have models instead of all the buildings in the city when determining what percentage of the city should be real buildings vs. generic buildings.

That is assuming my vague idea of how the city layout works is correct...

But i thought they were already designed that way??


<LSystem>LSYSTEM_1x1</LSystem>
<bAnimated>0</bAnimated>
<fScale>0</fScale>
<fInterfaceScale>0</fInterfaceScale>
<NIF>Art/empty.nif</NIF>

normal is like this:


<LSystem>LSYSTEM_3x2</LSystem>
<bAnimated>0</bAnimated>
<fScale>1.2</fScale>
<fInterfaceScale>0.9</fInterfaceScale>

 

[Yudishtira]

1) Units - They do but only when you have access to all the units thy upgrade to. This "problem" is due to the convoluted upgrade paths of some units.

Personally I would prefer the other extreme. I hate units obsoleting. Crossbows (and even longbows) are still useful many many centuries after muskets, especially as you usually have a tech lead over many civs. (The only unit building program I conducted in the Renaissance was when bellybowmen - by some loophole - became available.) And town watchmen are preferable to city guards in heaps of cases.

 

[God-Emperor]

But i thought they were already designed that way??


<LSystem></LSystem>
<bAnimated>0</bAnimated>
<fScale>0</fScale>
<fInterfaceScale>0</fInterfaceScale>
<NIF>Art/empty.nif</NIF>

normal is like this:


<LSystem>LSYSTEM_3x2</LSystem>
<bAnimated>0</bAnimated>
<fScale>1.2</fScale>
<fInterfaceScale>0.9</fInterfaceScale>

Cities are built on a square grid, with the size increasing with the population (I think that is the only factor, but I'm not sure). The first of your examples says the building is 1x1 in size via the LSYSTEM_1x1 part. It then puts a model of nothing (the empty.nif file has a couple of information type nodes, but no actual visible object), scaled to size 0, into that 1x1 area in the city. The second one puts a model scaled to 1.2x the actual size of the model into a 3x2 area. It doesn't know what the model is, or pay attention to the scaling reducing whatever it is (or isn't) to size 0. It reserves the specified number of squares of the city for the buildings, 1x1 or 3x2.

But maybe it should check to see if buildings specify that empty.nif and drop them just like if you didn't specify 128 even when you do. Or perhaps it would be quicker to check the fScale value and ignore anything with it set to 0.

 

[strategyonly]

check the fScale value and ignore anything with it set to 0.

Now that makes more sense, and if its easier also.

As long as ALL buildings are shown that is?? with the scaling Koshling designed????

 

[LumenAngel]

With these memory leak corrected, can you add back some bigger maps?

 

[alberts8]

With these memory leak corrected, can you add back some bigger maps?

Sadly no, the biggest was only 293 * 4bytes for every CvUnit created by the game.

 

[Koshling]

OK i labeled each pic with the # pertaining to each, so pic 1 is the "default" we currently have (3). (not what i want)

pic 3 labeled 131 should be all included, 1- Wonder, 2 - defenses, and 128 - buildings, hence 131?? correct, but it is NOT working???? as indicated

What's missing? The wonder in pic 1 seems to be missing in pic 3 (which certainly seems wrong). What else? It's a very straight-forward piece of code that implements that filter so I'm sure TB can easily debug through it given a test case (just look up where the define is used TB - it's very obvious and simple)