Latest Posts

Topic: new tribe: amazons

hessenfarmer
Avatar
Joined: 2014-12-11, 23:16
Posts: 2646
Ranking
One Elder of Players
Location: Bavaria
Posted at: 2019-12-11, 14:32

Ok I put the the formula of the code into an excel sheet to do some experiments. However this shows that we have a great variety of Humidity, Fertility and temperature values in the different terrains. So the values were made to ensure that the trees belonging to a biome are preferred and grow well in the biome. So Ibelieve it is nearly impossible to have one set of values that fits all terrains. This leads to following solutions:
1. have a rare tree of each type for each terrain (which I prefer)
2. skip the whole growth probability thing for rare trees.


Top Quote
hessenfarmer
Avatar
Joined: 2014-12-11, 23:16
Posts: 2646
Ranking
One Elder of Players
Location: Bavaria
Posted at: 2019-12-11, 16:54

for those interested the excel sheet can be found at this post https://www.widelands.org/forum/post/30547/


Top Quote
king_of_nowhere
Avatar
Topic Opener
Joined: 2014-09-15, 18:35
Posts: 1668
Ranking
One Elder of Players
Posted at: 2019-12-11, 17:58

JanO wrote:

king_of_nowhere wrote:

temperature 110

Really?

was a mistake, temperatures of the terrains are from 20 to 180 or so. i was going from memory first, then i found the old file, fixed a few things i remembered wrong, but didn't fix the temperature range. now i did

hessenfarmer wrote:

Ok I put the the formula of the code into an excel sheet to do some experiments. However this shows that we have a great variety of Humidity, Fertility and temperature values in the different terrains. So the values were made to ensure that the trees belonging to a biome are preferred and grow well in the biome. So Ibelieve it is nearly impossible to have one set of values that fits all terrains. This leads to following solutions:
1. have a rare tree of each type for each terrain (which I prefer)

perfectly doable. only problem, someone needs to make all the drawings. and the trees should hopefully be recognizable enough

  1. skip the whole growth probability thing for rare trees.

also perfectly possible. only problem, they will be planted on mountains and deserts.

I also gave a suggestion in my earlier post

  1. make the precious trees with the following values: preferred humidity 0.5, fertility 1, temperature 110, pickiness 0.3 (needs to be adjusted experimentally).

that way, with preferred humidity 0.5 and such a low pickiness it will grow well in any humidity conditions. same for temperature. on the other hand, with preferred fertility 1, it should grow well on all terrains with a fertility at least equal to 0.4, which are the ones where trees grow.

Edited: 2019-12-11, 17:58

Top Quote
hessenfarmer
Avatar
Joined: 2014-12-11, 23:16
Posts: 2646
Ranking
One Elder of Players
Location: Bavaria
Posted at: 2019-12-11, 18:44

king_of_nowhere wrote:

Ok I put the the formula of the code into an excel sheet to do some experiments. However this shows that we have a great variety of Humidity, Fertility and temperature values in the different terrains. So the values were made to ensure that the trees belonging to a biome are preferred and grow well in the biome. So Ibelieve it is nearly impossible to have one set of values that fits all terrains. This leads to following solutions:
1. have a rare tree of each type for each terrain (which I prefer)

perfectly doable. only problem, someone needs to make all the drawings. and the trees should hopefully be recognizable enough

yes that is the problem but they could be the same for all different versions of the trees. Only the three types should be distinguishable namely balsa ironwood and rubber.

  1. skip the whole growth probability thing for rare trees.

also perfectly possible. only problem, they will be planted on mountains and deserts.

exacly and this is why I dislike this solution.

I also gave a suggestion in my earlier post

  1. make the precious trees with the following values: preferred humidity 0.5, fertility 1, temperature 110, pickiness 0.3 (needs to be adjusted experimentally).

that way, with preferred humidity 0.5 and such a low pickiness it will grow well in any humidity conditions. same for temperature. on the other hand, with preferred fertility 1, it should grow well on all terrains with a fertility at least equal to 0.4, which are the ones where trees grow.

no they won't. If you download my excel (please crosscheck the formula) you will see that there is no pair of values that works on all 4 types of lands. BTW current values are hum 0.5, fert 0.5 temp 95 and pickiness 10 which is bad on some terrains although pickiness so low. if we reduce pickiness still further we will end with solution 2.


Top Quote
king_of_nowhere
Avatar
Topic Opener
Joined: 2014-09-15, 18:35
Posts: 1668
Ranking
One Elder of Players
Posted at: 2019-12-11, 20:22

hessenfarmer wrote:

no they won't. If you download my excel (please crosscheck the formula) you will see that there is no pair of values that works on all 4 types of lands. BTW current values are hum 0.5, fert 0.5 temp 95 and pickiness 10 which is bad on some terrains although pickiness so low. if we reduce pickiness still further we will end with solution 2.

damn. the drop-off is too gradual, resulting in too low growth with fertility 0.5 and too much at 0.2. i guess that's unavoidable with a large gaussian.

in this case we are stuck with your 1.

It's just that i have the horrible feeling that there will be some combination of terrains that will screw things up.

I would suggest removing terrain affinity but restrict those trees on buildable terrain. except it will make forested mountain unavailable.

For option 1, I suggest mimicking the values for already-exhisting trees, to avoid accidental screw-ups on strange terrain combinations. it would require initializing a lot of new trees, but the images for all will be the same

On another note, I have a balance concern and suggestion.

Some maps don't have all that much stone, because stone is only used once. amazons require a steady supply for weapons, it may make them problematic on long games on certain maps because running out of stone would be very easy.

For this i suggest giving depleted stone mines a 20% chance to dig stone. enough to get by. stone is normally plentyful and nobody would make an empty stone mine, so i don't think it will make any difference on a normal map


Top Quote
hessenfarmer
Avatar
Joined: 2014-12-11, 23:16
Posts: 2646
Ranking
One Elder of Players
Location: Bavaria
Posted at: 2019-12-11, 20:32

king_of_nowhere wrote:

hessenfarmer wrote:

no they won't. If you download my excel (please crosscheck the formula) you will see that there is no pair of values that works on all 4 types of lands. BTW current values are hum 0.5, fert 0.5 temp 95 and pickiness 10 which is bad on some terrains although pickiness so low. if we reduce pickiness still further we will end with solution 2.

damn. the drop-off is too gradual, resulting in too low growth with fertility 0.5 and too much at 0.2. i guess that's unavoidable with a large gaussian.

in this case we are stuck with your 1.

It's just that i have the horrible feeling that there will be some combination of terrains that will screw things up.

can't be as the rare tree planter should do the same as the forester. (i.e. looking a couple of times according to the attribute saplingsearches for the best spot tom plant a tree) so a mixed spot should be exluded.

I would suggest removing terrain affinity but restrict those trees on buildable terrain. except it will make forested mountain unavailable.

For option 1, I suggest mimicking the values for already-exhisting trees, to avoid accidental screw-ups on strange terrain combinations. it would require initializing a lot of new trees, but the images for all will be the same

as for my answer above I still feel we could get away with 12 (4 terrains times 3 trees).

On another note, I have a balance concern and suggestion.

Some maps don't have all that much stone, because stone is only used once. amazons require a steady supply for weapons, it may make them problematic on long games on certain maps because running out of stone would be very easy.

For this i suggest giving depleted stone mines a 20% chance to dig stone. enough to get by. stone is normally plentyful and nobody would make an empty stone mine, so i don't think it will make any difference on a normal map

ok that sounds reasonable will do that in the next amazon turn.


Top Quote
WorldSavior
Avatar
Joined: 2016-10-15, 04:10
Posts: 2091
OS: Linux
Version: Recent tournament version
Ranking
One Elder of Players
Location: Germany
Posted at: 2019-12-11, 20:55

Nordfriese wrote:

Gardening center: The worker should get an additional dummy program that only performs the findspace program, which is called by the building before consumption. That way the program will fail if there are no suited places before consuming.

setting rare wood eco settings to efficiency profile lets the game crash

The predefined profiles are just placeholders so far, they still need to be (re)written

Okay

hessenfarmer wrote:

WorldSavior wrote:>

Isn't there an easy way somewhere to show instead just "didn't plant balsa tree because economy doesn't need balsa", for example?

no the message is directly compiled from the conditions in the working programs.

That's a pity. And I guess that it's not worth the effort to change this...

no not in the sight of what needs to be done else.

Okay

To be precise: Fortresses are not enhancable. Or are they? Not sure anymore.

Can't confirm they are enhanceable to fortification in my build.

My mistake. I expected fortresses and fortificiations to be like frisian tower and fortress: Not dependend of each other. Maybe because of the Frisian graphic placeholders : P

I've spotted some more bugs in a long test match. - rarecutters cannot cut trees, but they should if no rare tree there or if rare tree not needed. .

the first is not easy to implement but doable the latter could be done easily, if needed, but I am not sure what was KoN's opinion on that as he is the soiritus rector of this tribe.

His opinion was like my opinion.

  • rare trees can cover statistic percentage!

all tribes immovable can do this I believe, not sure whether world immovables (normal trees) are not doing the same.

Maybe they do now. Possibly the letters should be displayed a bit higher again like in build19...

  • the contra-productive rope eco setting still exists.

don't understand this. Please explain and / or point us to a specific point in discussion which I might have forgotten.

Lianas are only needed for rope, so there is no need for a rope eco setting. Exactly as Imperial whine.

  • replays crash the game immediately

is there any message provided?

Windows provides the message that the game crashed face-wink.png

could you provide a stdout. text log from this?

I'm currently not on windows but we'll see what I can do soonish. Do amazonian replays work for you?

And in the test match I reached the point where Eco didn't need any more rare trees, so it was a pity for some wasted resources face-wink.png

good to hear this is possible. However in this case I would recommend setting new economy demand values. should be easy with +10 button.

That's a little bit like there would be an economy setting for meat and you would suggest me to simply increase the eco setting if when the hunter-gatherer stops working face-wink.png

Thanks for the feedback.

You're welcome

However I'll prefer to get the balancing changes done first.

Fine

king_of_nowhere wrote:

On another note, I have a balance concern and suggestion.

Some maps don't have all that much stone, because stone is only used once. amazons require a steady supply for weapons, it may make them problematic on long games on certain maps because running out of stone would be very easy.

For this i suggest giving depleted stone mines a 20% chance to dig stone. enough to get by. stone is normally plentyful and nobody would make an empty stone mine, so i don't think it will make any difference on a normal map

Sounds reasonable. In my test match, stone was a bigger problem than rare wood, and this was on "The Big Lake", which is rather stone-rich.

Imho the exhausted gold digger dwellings are also too unefficient. Maybe they should get also 20% chance instead of 5?

@tree growth problem: Isn't it possibly to set the pickiness so low that deserts are still bad for rare trees, but forest-friendly terrain always great?

In case you don't know it: You can check the probability for each terrain if you

-open an amazon match

-press F1 and select immovables, then the trees.


Wanted to save the world, then I got widetracked

Top Quote
hessenfarmer
Avatar
Joined: 2014-12-11, 23:16
Posts: 2646
Ranking
One Elder of Players
Location: Bavaria
Posted at: 2019-12-11, 21:21

WorldSavior wrote:

I've spotted some more bugs in a long test match. - rarecutters cannot cut trees, but they should if no rare tree there or if rare tree not needed. .

the first is not easy to implement but doable the latter could be done easily, if needed, but I am not sure what was KoN's opinion on that as he is the soiritus rector of this tribe.

His opinion was like my opinion.

ok so let's keep this on the list then.

  • rare trees can cover statistic percentage!

all tribes immovable can do this I believe, not sure whether world immovables (normal trees) are not doing the same.

Maybe they do now. Possibly the letters should be displayed a bit higher again like in build19...

+1 from my side

  • the contra-productive rope eco setting still exists.

don't understand this. Please explain and / or point us to a specific point in discussion which I might have forgotten.

Lianas are only needed for rope, so there is no need for a rope eco setting. Exactly as Imperial whine.

ok understood will be corrected.

  • replays crash the game immediately

is there any message provided?

Windows provides the message that the game crashed face-wink.png

could you provide a stdout. text log from this?

I'm currently not on windows but we'll see what I can do soonish. Do amazonian replays work for you?

not tested as I'll try to use my time for getting some changes in face-wink.png will do some regression testing if time permits.

And in the test match I reached the point where Eco didn't need any more rare trees, so it was a pity for some wasted resources face-wink.png

good to hear this is possible. However in this case I would recommend setting new economy demand values. should be easy with +10 button.

That's a little bit like there would be an economy setting for meat and you would suggest me to simply increase the eco setting if when the hunter-gatherer stops working face-wink.png

ok noted but not on the priority list. sorry

Imho the exhausted gold digger dwellings are also too unefficient. Maybe they should get also 20% chance instead of 5?

why should they be able to mine more gold than other tribes?

@tree growth problem: Isn't it possibly to set the pickiness so low that deserts are still bad for rare trees, but forest-friendly terrain always great?

no unfortunately it is not that easy. as pickiness works on all 3 dimensions of the gaussian space more the less equally.

In case you don't know it: You can check the probability for each terrain if you

-open an amazon match

-press F1 and select immovables, then the trees.

damn I forgot that thanks for reminding.


Top Quote
JanO
Avatar
Joined: 2015-08-02, 11:56
Posts: 177
Ranking
At home in WL-forums
Posted at: 2019-12-12, 11:28

Regarding the tree-growth problem: Did I understood correctly, that the three parameters temperature, humidity and fertility are basically constant values for each biome? Being a farmer myself, this does not make very much sense to me. So maybe the problem can be solved by not editing the trees, but editing the ecosystem. Let's say temperature should not be a constant value but instead depend on level of elevation. Humindity could depend on distance to the next water or marsh, maybe also elevation or water-resources placed in the map editor. Fertility may be the one which is set by the individual terrain. In reality it's the most complex parameter of those three, which makes it very complex to calculate a reasonable value for it in a game.

I have no idea about the current implementation of those tree-growth determining values, but I fear that this means somewhat hard codework and may have noticable impact on old maps, too. But if the current implementation is as static as I think, I wonder why we have three given values instead of only one.


Top Quote
hessenfarmer
Avatar
Joined: 2014-12-11, 23:16
Posts: 2646
Ranking
One Elder of Players
Location: Bavaria
Posted at: 2019-12-12, 13:17

JanO wrote:

Regarding the tree-growth problem: Did I understood correctly, that the three parameters temperature, humidity and fertility are basically constant values for each biome?

no they are defined not per biome they are defined for terrains. But in one biome the range of the values is smaller, so we have typical ranges of each value per biome. Furthermore the excat values for each node are calculated from the 6 surrounding terrains.

Being a farmer myself, this does not make very much sense to me.

see above it isn't as you thought. However we need to keep in mind 2 things. First we are doing a game not a simulation. Second as far as I remember this was introduced for a special purpose. After the union of the former 4 terrains into the one world design (which means you can combine all terrains in one map) we needed a mechanism to ensure that the trees designed for that biome are preffered in the their related landscapes.

So maybe the problem can be solved by not editing the trees, but editing the ecosystem. Let's say temperature should not be a constant value but instead depend on level of elevation.

it kind of does as mountain terrain has always lower temperatures then normal lands.

Humindity could depend on distance to the next water or marsh, maybe also elevation or water-resources placed in the map editor. Fertility may be the one which is set by the individual terrain. In reality it's the most complex parameter of those three, which makes it very complex to calculate a reasonable value for it in a game.

If you look into the terrain definitions at data/world/terrains you could see that for summer and winter terrains there are some variety in the values which fit logically to the outline of the terrain. However a map designer could use them completely free. (e.g beach terrain at top of the mountains.

I have no idea about the current implementation of those tree-growth determining values, but I fear that this means somewhat hard codework and may have noticable impact on old maps, too. But if the current implementation is as static as I think, I wonder why we have three given values instead of only one.

cause with 3 values we have a gaussian space instead of an area or a line, which gives us more options.


Top Quote