Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[CobaltKobold]

Compete.

[Footkerchief]

but is it really that awkward? i have no problems with cheetahs next to kangaroos in-game, maybe they just happened to migrate that way? perhaps one of the sentient races moved them to a new landmass and they bred like crazy (think rabbits/toads in australia, camels for that matter too) sure it would be a good idea to add such a thing but untill it is i don't see it being a big problem amd if it is it shouldn't always end up that way, it should only be more likely for the same reason i stated above (parhaps have a migration chance that is used for both natrual migration and translocation by sentients)

But considering we are creating imaginary worlds, why should we place so much importance on forcing all fauna to coincide? Who is to say an imaginary continent might not have bizarre combinations of creatures within it? We have goblins in Dwarf Fortress; Kangaroos aren't even marsupials within the game. Why should as minor a thing as correspondence to actual biomes necessarily be so significant?

Bizarre combinations of creatures aren't a huge problem per se.  What I'm mostly concerned about is excessive homogeneity between biomes.  It's boring to walk into a temperate grassland and immediately know that it'll have exactly the same animals as every other temperate grassland.  The world becomes much more interesting when two biomes of the same type can have radically different organisms.  Sphere-linked regions will only partially alleviate this.

I do agree that adding some kind of tag that causes certain groups of creatures to spawn in a biome together and exclude others might be useful. However, rather than hackishly assigning continent ids, it might be better to determine a more emergent way of evolving animals as a true bug fix to the "Cheetahs and Kangaroos oh no" problem.

The method I mentioned is hackish, yes.  It was just the first vaguely workable method that came to mind.  A more flexible system would be nice, but I'm not sure what kind of "evolving" you're describing here.  An actual evolution simulation a) would be way too complicated and b) isn't necessarily going to produce the Earth fauna that the Ark Project aims to provide.

Perhaps biomes could be made more complex with "niches" and then animals could be selected to fill them based on size, food they consume, and ability to reproduce, among other things.

Niches are absolutely necessary for the wildlife simulation, but they won't solve this problem.  Biomes of the same type tend to have the same niches, meaning that you can't rely on niches to heterogenize the creature populations.  Incidentally this is why invasive species are so devastating -- a species from continent A not only evolves for the exact same niche as a species from continent B, but also does an unconditionally better job at it.  If you mix all of Earth's temperate grassland species in a single DF biome and allow them to compete, some of them will simply disappear.

e: I see G-Flex beat me to that point

[Nivim]

 On the subject of material properties, hasn't anyone asked a chemist (geologist, metallurgist) about some complete list? I thought these kinds of things had been tested and recorded before...Just in case, I'll ask my teacher/professor next time I get a chance (AP classes FTW...yet there are only 37 people out of ~320 in chemistry).

 DF is mainly theoretical; insanely accurate theory. I think the niche system could work, but only if it somehow accounted for the minor differences between animals. As the world is generating, numerous species would be wiped out from the imbalance, and others would be successful. I think if we brought the initial diversity high enough, this would just lead to a different semi-balanced world each time (I just visualized arctic marsupials for some reason, but that would require an adaptation system). It's kinda like how sometimes elves, goblins, humans, or dwarves have been wiped out by game start. Sometimes all of them if orcs are involved.

 I was "ninja'd" by Footkerchief; this is why I must conquer my attention span.

[Chthonic]

The second problem is that the Ark Project will result in slightly disturbing biome populations, like kangaroos and cheetahs inhabiting the same savanna.  DF has no real mechanism for controlling this, but I've been thinking about suggesting one.  It should be fairly straightforward -- each creature def specifies a "continent ID" or list of them (or a token like ANY_CONTINENT) and worldgen assigns one of those "continent IDs" to each landmass, which constrains what creatures can show up there.  So if you had a world full of islands, some of the islands would have North American fauna, others Australian, and so on.  This might be a little clumsy for some situations, I haven't really thought it through yet.

I don't really have a problem with this . . . it's a fantasy game, after all.  Would an "Australian" unicorn have a pouch?  I'd personally make it half-striped, like a quagga.

I think darkflagrance's idea for "niches" is good . . . maybe assign each animal a numerical value based on size and diet (corresponding to trophic level), and then allow each region a number of slots based on region size for each range of values.  Then each slot would "draw" appropriately valued creature cards from the region's deck.  A large region might end up with four species of large herbivore and two apex predators, while a small one might have only two large herbivores and a single large predator.  That way you get unique flavors.

edit: and by "region's deck" I mean "biome's deck"

[Neruz]

It would probably be more functional to be able to list 'companion' creatures in the creature data. When a creature is seeded into a biome, it's more likely to place it's 'companion' creatures in the biome and less likely to place non companion creatures into the biome.

This way you can chain different kinds of creatures together to get interesting and unique biomes.

[Footkerchief]

I think darkflagrance's idea for "niches" is good . . . maybe assign each animal a numerical value based on size and diet (corresponding to trophic level), and then allow each region a number of slots based on region size for each range of values.  Then each slot would "draw" appropriately valued creature cards from the region's deck.  A large region might end up with four species of large herbivore and two apex predators, while a small one might have only two large herbivores and a single large predator.  That way you get unique flavors.

Oh, if that's what darkflagrance meant, then yeah, that kind of approach may be worth pursuing.

It would probably be more functional to be able to list 'companion' creatures in the creature data. When a creature is seeded into a biome, it's more likely to place it's 'companion' creatures in the biome and less likely to place non companion creatures into the biome.

Defining graph relationships in the creature raws would be very messy and difficult to maintain.  I'd move toward a hierarchical system at least, i.e. grouping creatures in some fashion.

[G-Flex]

It would probably be more functional to be able to list 'companion' creatures in the creature data. When a creature is seeded into a biome, it's more likely to place it's 'companion' creatures in the biome and less likely to place non companion creatures into the biome.

Seems way too artificial.

If possible, a more natural/procedural approach would probably work better. If the game knows what animals are good at what, it can fill in the resulting niches semi-randomly and see how they work out over time.

[Askot Bokbondeler]

if we're having exotic creatures i demand exotic mythologies to go with them, animistic aboriginal spirits and mayan feathered serpent demons with entities for themselves

[Footkerchief]

If possible, a more natural/procedural approach would probably work better. If the game knows what animals are good at what, it can fill in the resulting niches semi-randomly and see how they work out over time.

There could also be a worldgen parameter that determines how much the "continent ID" (or some finer-grained method of geographical grouping) biases those semi-random selections.  So people could have very Earthlike biomes, or everything mixed together, or anything in between.

[tweinst]

It would probably be more functional to be able to list 'companion' creatures in the creature data. When a creature is seeded into a biome, it's more likely to place it's 'companion' creatures in the biome and less likely to place non companion creatures into the biome.

Defining graph relationships in the creature raws would be very messy and difficult to maintain.  I'd move toward a hierarchical system at least, i.e. grouping creatures in some fashion.

How about describing food characteristics for each creature, then matching predators up with prey? Perhaps some info about hunting characteristics would help there too, such as wolves hunting in packs, or giraffes having long necks.

For example, a tiger might be able to kill anything up to the size of a horse if it doesn't have good offensive weapons. Or a snake might be able to eat anything that it could swallow that doesn't have spines.

[Nadaka]

How about defining each creatures niche in the foodchain? each biome gets a random selection creatures for each niche, occasionally more than one.

[Chthonic]

I think all you need is trophic level and biome . . . no point in getting overcomplicated.  Similar biome-regions in widely distributed geographical regions are going to have similar niches at each trophic level, after all.

To blatantly oversimplify things, a kangaroo is just a particularly weird gazelle.

[jokermatt999]

Toady's stated that animal relationships are going to need to be redefined. Perhaps it'd be a good question to ask him how he plans to do that.

Also, I like the idea of putting animals together based on what they consume and how much of it. Yes, that's technically their "ecological niche", but it could be somewhat simplified.

Creature "gazelle" eats y units of plants, mostly grass. Creature "wolf" eats x units of meat, fresh, not scavenged. Can "wolf" take down "gazelle" (based on its hunting tactics)? "Wolf" would work as a predator for this biome.

This oversimplification is probably making some ecologists cry, but that's just my basic idea for what to do. Someone with real knowledge is certainly welcome to correct me.

[Lord Shonus]

I think a simple, if hackish, temporary solution is to only allow a certain number of each "category" of animal in each region. For example, a region might support ten or eleven kinds of small herbivores (rabbits, squirrels, etc.) four or five kinds of medium herbivores (Horses, zebra, etc.),  and two or three kinds of large ones (moose, elk, elephant, etc.), and one extra-large species (Triceratops, Brachiosaurus, etc.) Use a similar pattern for predators, with smaller populations. Not extremely realistic, but it would make each region distinct.

Th ebiggest issue with invasive species is not that they are better adapted for the environment, but that the ecosystem is badly adapted to them. Had the bunny evolved in australia, it would not have been so destructive.

[Askot Bokbondeler]

what about having generic template creatures and have them "adapt" in a generic simplified way? we could end up with large wolves hunting mountain elephants, and being named after the region they exist on, we'd be starting with real world animals, but fantasyfying them on the run

ofcourse this would have to be done with a bit more sesearch, so we could get a significant variety in characteristics, instead of just having biger or smaller, brands of the same animal