Topic: DF v0.47.01+ Worldgen Cookbook Thread (Read 107614 times)

PatrikLundell · « **Reply #270 on:** February 03, 2021, 04:53:32 am »

With PSV worlds it's possible to mostly lock civs in place by separating them by terrain they can't build in, and if these barriers are very evil and very savage it works better still.

By repeatedly generating worlds you can get RNG situations where goblins don't dominate. If you change the raws so they start only on a particular (set of) terrain you can use separation corridors to mostly keep them from expanding too much.

vjek · « **Reply #271 on:** February 03, 2021, 10:51:47 am »

Quote from: horribleflesheater on February 03, 2021, 02:17:22 am

...
Can I create a sinister, evil world without the inevitable goblin conquest creating a world of 75% goblins?
...

For this one requirement in particular, I have found limiting the Number of Demon Types to exactly 2 tends to make Goblins less 'world dominating' while still allowing their civ creation.

horribleflesheater · « **Reply #272 on:** February 04, 2021, 04:17:15 pm »

That worked far faster and more effectively than generating over and over hoping they're locked in the ice sheets, thanks!

Crystal · « **Reply #273 on:** February 12, 2021, 05:37:28 am »

I need a bit of help generating a particular world. I am trying to get one with all civs including a tower on the embark screen. Along with the fort being located in a heavily forested area with deep metals, shallow metals, flux layer, sand, clay, fire clay, and very deep soil. Additionally, it should have a magma pipe close to the surface with a river near by and the cavern at least 25 away from the surface. Savagery would be preferred with untamed wilds having a higher priority over good/evil areas, but it is not a focus of mine. Maybe just so that good regions are in the world so I can trade for some items from those areas, but its fine if it isn't.

Hardest part I found is trying to generate enough towers to find a location. I bumped up the secrets but that didnt really seem to change much and I mostly only end up with around 1 to 3 towers. The other one was trying to find a magma tube without having to embark. Don't remember how to look for that. I did try bumping up the rainfall and drainage to generate more forests but I seem to be getting more shrub lands instead.

PatrikLundell · « **Reply #274 on:** February 12, 2021, 11:43:32 am »

Longer history -> more towers, and many human civs tends to cause human vs human wars that drives people to search for immortality. The DFHack Embark Assistant plugin allows you to search for magma pipes (as well as many of the other criteria you're looking for).

BeardedShrimp · « **Reply #275 on:** February 15, 2021, 01:06:38 pm »

For v0.47.05+

The simple goal was to generate a small world where 1) no civ was isolated on their own little island, 2) goblins have not overtaken the world in population, and 3) necromancer towers were present but not literally everywhere. Humans dominating the world's population was not surprising. The unexpected part was a giant Joyous Wilds Sand Desert+Badlands smack next to Haunted and Terrifying Temperate Grassland regions in the east:

Spoiler (click to show/hide)

Worldgen Paramters:

Spoiler (click to show/hide)

javascripter · « **Reply #276 on:** February 20, 2021, 07:31:53 am »

I'm very late to the discussion and my addition is also more relevant to fortress mode and probably known about, but I figured I'd share anyway incase either it isn't known about or the issues I have with it are known better than I know them.

I've been annoyed for a while that no worldpainter lets me paint good/evil.
I found by using the "desired good/evil tiles in small/medium/large subregion" combined with the world painter (I guess you call PSV here) it was fairly easy to make a world with lots of evil while actually being able to place some controllable civilizations. (for example, evil tiles in small subregion -> max; evil tiles in medium subregion -> min)

I found water and high mountained (filled in triangle) both block civilization influence, while low mountains (open triangle) seem to not, and that there is still a lot of variance in the maps even if I set all variances to none. This made larger savannas and grasslands often have wrong type of terrain in the intended medium region, which then became a small region and went evil, so I would be interested if anyone knows how much variance there can be with variance turned off (and orthographic precipitation turned off); I also found marshes and swamps would be considered in the same regions, and lower and higher mountains would be considered the same region, as well as a volcano on a 100 elevation tile would join the same region.
I also found rivers would sometimes split an area intended to be a medium subregion, sometimes this would even happen for 2xsomething areas.
Also, I got some intermittent flooding, it seemed to get better if I added regular tiles with elevation below 100, or if I reduced the height of mountains, but i'm not clear on how corelated those were or how near the water needs to be.

Anyway, I wanted volcano+mountain+river+swamp+evil+savage+nearby towers+nearby goblins+if possible other nearby civilizations, and here is the parameters I used for that: (goblin raws were also modifies to only spawn on deserts, though I think I would need to up the map size to prevent dieoff of non-goblins by goblin causes; I'm not sure how much the high savagery or war between human/dwarf (or human/human and dwarf/dwarf) affects the dieoff rate; I suppose with a larger map I could spread out races so their influnce overlapped in a small area and then copy that and just check the small area, but living fortresses wasn't important for me, I would have even preferred a totally dead race but I guess that doesn't really happen right now.)

Spoiler (click to show/hide)

PatrikLundell · « **Reply #277 on:** February 20, 2021, 08:46:34 am »

You can't paint good/evil because it's assigned by DF internally during world gen. The reason is probably because good/evil are region features (recently modified by evil spreading from sources, but that's probably more of a temporary hack than a "real" implementation, as spheres will require a new framework during the map rewrite anyway), and regions are build from the other primitive parameters that are available (salinity isn't available either, but salty/brackish/fresh doesn't affect region type). If you define primitive parameters in neighboring tiles that result in the same region type they join up to form a single region.
Evilness can be hacked during/after world gen, but that's the current extent. It would be possible to have a script interrupt world gen immediately after assignment of evilness to override those values, but it's quite possible DF doesn't detect that and you'd have to hack the flora/fauna of the region as well to make it work.
As you say, a PSV world allows you to paint what becomes regions such that you can manually split them into good/neutral/evil based on size to a decent degree.

Variance shouldn't have any effect on a fully defined PSV world. However, if you specify only some of the parameters and let DF generate the others it will do so (and I generally use fully defined PSV world, so I don't know much about how the hybrid approach works).

The wiki should be able to tell you which biomes belong to the same region type. "Hills" and "Grassland" have the same biomes (savanna, etc.), while all the other biomes are present in a single region type only.

A volcano is a "feature" that exists within a tile, and so isn't tied to any particular region type, even if the cone may rise to mountain elevations.

Rivers shouldn't split regions, as they appear "on top" (they're "features"). I suspect the cases you've seen are grassland/hill regions (they differ only by Drainage).

Never seen any flooding (although someone recently said that happened, although it wasn't verified as far as I've seen). Lake formation happens during world gen and can't be controlled by PSVs, though. Elevation below 100 results in ocean tiles (which can result in very small oceans...), not lakes.

A major civ killing factor is megabeast/titan count. They tend to stomp out a lot of civs while they only have a single site if you have a lot of them.

Larger maps typically have more civs on them, but more civs means there's a greater chance some survive somewhere, so reducing the number of civs reduces the chance/risk of them being near each other.

You probably don't want to play a truly dead civ, as those a badly bugged: No travel possible, so no raiding. Your civ won't found sites (naturally), but you can't conquer any either, so you're relying on sites to spontaneously tie themselves to you for those nobles. If you lose your expedition leader it won't be replaced, which locks out a lot of functionality if that happens, but I think getting a mayor will get past that problem (there should be a new one elected the next summer).

javascripter · « **Reply #278 on:** February 20, 2021, 11:10:13 am »

Quote from: PatrikLundell on February 20, 2021, 08:46:34 am

You can't paint good/evil because it's assigned by DF internally during world gen.

I know this, but it still has bothered me a bit

Quote

Variance shouldn't have any effect on a fully defined PSV world. However, if you specify only some of the parameters and let DF generate the others it will do so (and I generally use fully defined PSV world, so I don't know much about how the hybrid approach works).

It seems i was incorrect here, the symbol changes for savanna/grassland, but it creates one big region with different symbols. It also seems regions depend on temperature somehow though and I left temperature to be randomly generated, which may also have been an effect, I guess in the future I should just set it to what I want everywhere.

Quote

The wiki should be able to tell you which biomes belong to the same region type. "Hills" and "Grassland" have the same biomes (savanna, etc.), while all the other biomes are present in a single region type only.

If I understand you correctly you are saying that hilly savanna and hilly grassland become the same region and savanna and grassland become the same region but it is not
my own testing tells me that broadleaf forest and conifer forest also become same region, as well as marsh and swamp becoming same region, so wiki isn't totally accurate but not especially difficult to figure it out (at least without rivers causing problems)

test world here:

Spoiler (click to show/hide)

Maybe I should do some more testing and put results to the wiki, since right now it has no information about regions and some more testing on this wouldn't be that much work.

Quote

A volcano is a "feature" that exists within a tile, and so isn't tied to any particular region type, even if the cone may rise to mountain elevations.

in the lower right corner of the previous test I have two 3x5 blocks of mountains separated by a 1 wide strip of deserts with 100 volcanism; I have it so only medium regions will go evil and those mountains go evil, while if I change the volcanism to 0 they don't.

Quote

Rivers shouldn't split regions, as they appear "on top" (they're "features"). I suspect the cases you've seen are grassland/hill regions (they differ only by Drainage).

After some more testing, I seem to be incorrect here as well, that must have been temperature effects then, or possibly flooding, not sure.

Quote

Never seen any flooding (although someone recently said that happened, although it wasn't verified as far as I've seen). Lake formation happens during world gen and can't be controlled by PSVs, though. Elevation below 100 results in ocean tiles (which can result in very small oceans...), not lakes.

I suspect the flooding and lakes are the same things, but they seem to be somewhat controllable by elevation modification for example, or for example by rivers; here is what my worldpainter screen looks like:

(https://ibb.co/S5hf4BW incase img tag doesn't work)

If I generate this world with 800 rivers, I sometimes get some small lakes:

(https://ibb.co/hsm8DHH incase img tag doesn't work)

But if I generate with 0 rivers I get a huge lake:

(https://ibb.co/Zg0WpPR incase img tag doesn't work)

I'm not sure but it also seems to be based on relative elevation, with a similar setup but larger map I was having problems with huge portions of the map becoming lakes, but by putting down a few sea tiles it seemed to prevent the lakes from showing up. I think reducing the sizes of the peaks helped as well, though I don't remember for sure and the reason I did that was because it made it must more likely for rivers to be near the volcano mountains.

Quote

You probably don't want to play a truly dead civ, as those a badly bugged: No travel possible, so no raiding. Your civ won't found sites (naturally), but you can't conquer any either, so you're relying on sites to spontaneously tie themselves to you for those nobles. If you lose your expedition leader it won't be replaced, which locks out a lot of functionality if that happens, but I think getting a mayor will get past that problem (there should be a new one elected the next summer).

not all those are problems for me, but also didn't realize so many bugs, and not a huge deal though, dying civilization is fine

PatrikLundell · « **Reply #279 on:** February 20, 2021, 12:20:55 pm »

Temperature affects whether it's tropical, temperate, or tundra, although it's only directly tied to temperature if the world doesn't have poles, otherwise tropical/temperate is controlled by latitude with temperature controlling the result in the subtropical zone.

The symbols generated for tiles is the one for a tree that's present there, while I'd prefer to see something representing the biome. The tree selected can differ for different tiles within the same biome.

You understood me correctly in the first part, but not the second. All kinds of forests belong to the Jungle region type, all kinds of ocean to the Ocean one, all kinds of wetlands to the Swamp one. All kinds of mostly open grass dominated terrain end up as either Hills or Grassland, depending on Drainage, so those to region types contain exactly the same sets of biomes. Unless I've missed one, the rest of the region types have a single biome in them.

Changing Volcanism probably scrambles the RNG, resulting in different results. The evilness settings for small/medium/large regions are not absolute, although they can be made mostly so by ensuring the requested number of tiles of that type in that kind of region exceeds the actual number of tiles, but it still doesn't give you full control.

Again, changing the number of rivers scrambles the RNG, but I believe lakes are generated to be connected to rivers in a somewhat logical way, so I would guess you might be able to guide river formation (and hence lake formation) to some extent by generating elevations such that there are clear valleys for rivers. However, I'd expect the result to be very much "somewhat".
Good to know that you mean lakes when talking about flooding. The other post about talked about oceans flooding land.

Any kind of map manipulation scrambles the associated RNG(s), so you'd need to make a large number of tests to draw any conclusions about effects from map changes as opposed from just getting a different result from using the RNG sequence differently.

javascripter · « **Reply #280 on:** February 20, 2021, 02:18:02 pm »

Quote from: PatrikLundell on February 20, 2021, 12:20:55 pm

The symbols generated for tiles is the one for a tree that's present there, while I'd prefer to see something representing the biome. The tree selected can differ for different tiles within the same biome.

I definitely agree with that but thats the way the game is; and you got the idea anyway

Quote

You understood me correctly in the first part, but not the second. All kinds of forests belong to the Jungle region type, all kinds of ocean to the Ocean one, all kinds of wetlands to the Swamp one. All kinds of mostly open grass dominated terrain end up as either Hills or Grassland, depending on Drainage, so those to region types contain exactly the same sets of biomes. Unless I've missed one, the rest of the region types have a single biome in them.

In that case information on biomes from wiki is basically just incomplete, though fairly easy to understand; I will add to advanced world gen though since it definitely should be there.

Quote

Changing Volcanism probably scrambles the RNG, resulting in different results. The evilness settings for small/medium/large regions are not absolute, although they can be made mostly so by ensuring the requested number of tiles of that type in that kind of region exceeds the actual number of tiles, but it still doesn't give you full control.

Still, if I ask for maximum evil tiles in medium regions and no evil tiles in small/large regions it is pretty clear anything evil must be a medium region.
if I make two areas of mountains both ~15-20 tiles and connect them with a definitely non-mountain tile and they always (as far as I've seen) become evil if that tile is a volcano, but not if it isn't a volcano (in which case dwarf civilizations are placed at difference places depending on the run for example) then it seems fairly unlikely to be random, unless I've gotten incredibly unlikely occurances with noncontiguous areas counting as regions sometimes it seems more likely that volcano tiles count as mountain tiles for the purposes of regions

another test I did, if you generate this world the volcano switches between the sides between each attempt to generate, but the side with the volcano is always the evil one:

Spoiler (click to show/hide)

Quote

Again, changing the number of rivers scrambles the RNG, but I believe lakes are generated to be connected to rivers in a somewhat logical way, so I would guess you might be able to guide river formation (and hence lake formation) to some extent by generating elevations such that there are clear valleys for rivers. However, I'd expect the result to be very much "somewhat".
Good to know that you mean lakes when talking about flooding. The other post about talked about oceans flooding land.

I didn't know what I meant, I just knew with a map like that it was regularly covering areas with water and then I aborted and tried again
eventually I decided maybe it was simulating that water couldn't flow anywhere and would pool there and tested some things to get rid of it and they sort of worked (still not exactly as I expected, but I don't know the physics/geology very well in real life and dwarf fortress physics are wonky sometimes)

PatrikLundell · « **Reply #281 on:** February 20, 2021, 06:30:20 pm »

I wouldn't be shocked if DF didn't always respect the elevation PSV value if it plonks down a volcano in a tile, in particular if that tile was surrounded by mountain tiles on at least two (out of 4 perpendicular) sides.

javascripter · « **Reply #282 on:** February 20, 2021, 11:36:28 pm »

Given that at least with one generated world that tile is totally covered with mountains, that seems to be what is happening.

Dushan · « **Reply #283 on:** March 19, 2021, 08:32:45 am »

Dont know how hard this would be, but im looking for a sand desert with a major river waterfall.

Kinda like a Death Valley/Nile/Niagara falls type thing.

If you can tell me how to generate a desert world with rivers im willing to look around for a Embark site myself.

Thank you in advance.

vjek · « **Reply #284 on:** March 19, 2021, 11:19:21 am »

This is what I can offer quickly, Dushan.
It's likely not exactly what you're looking for, but gives you an idea of a possible path forward..

Spoiler: desert-waterfall (click to show/hide)

An explanation..
A worldgen like this strongly encourages the path of a river to follow a predetermined course. A path that guarantees a certain length, that, coupled with a predictable rate of widening, allows you to get a major river to follow a certain course through the world.
You can do the same thing with smaller worlds, but the results are you get a less wide river.
In this example above, a waterfall is falling into a major river. Rather than a major river being the waterfall.
However, it does at least show you how to generate, predictably, a major river in a desert.

The tributaries in this world all feed the main river with 99 Rain. The major river itself has a rainfall value of 2, along it's path. You can see this by examining the PS_RF part of the worldgen pre-set values.

As far as artificially creating actual waterfalls (as in, creating & showing the waterfall icon on the embark interface, predictably) I haven't been successful in that endeavor with up to a painted 50Z drop in elevation along a river path. More elevation drop may be required, or it simply may not be possible with pre-set values. I'm not sure, yet.

News:

Author Topic: DF v0.47.01+ Worldgen Cookbook Thread (Read 107614 times)

PatrikLundell

Re: DF v0.47.01+ Worldgen Cookbook Thread

vjek

Re: DF v0.47.01+ Worldgen Cookbook Thread

horribleflesheater

Re: DF v0.47.01+ Worldgen Cookbook Thread

Crystal

Re: DF v0.47.01+ Worldgen Cookbook Thread

PatrikLundell

Re: DF v0.47.01+ Worldgen Cookbook Thread

BeardedShrimp

Re: DF v0.47.01+ Worldgen Cookbook Thread

javascripter

Re: DF v0.47.01+ Worldgen Cookbook Thread

PatrikLundell

Re: DF v0.47.01+ Worldgen Cookbook Thread

javascripter

Re: DF v0.47.01+ Worldgen Cookbook Thread

PatrikLundell

Re: DF v0.47.01+ Worldgen Cookbook Thread

javascripter

Re: DF v0.47.01+ Worldgen Cookbook Thread

PatrikLundell

Re: DF v0.47.01+ Worldgen Cookbook Thread

javascripter

Re: DF v0.47.01+ Worldgen Cookbook Thread

Dushan

Re: DF v0.47.01+ Worldgen Cookbook Thread

vjek

Re: DF v0.47.01+ Worldgen Cookbook Thread