Bay 12 Games Forum

Many of us have experienced the hilarious "melting Rain" bug. As i understand it (and please correct me if i am wrong) the rain certain in regions was so hot (above 40°C) that i caused the fat and the flesh off a dwarf to melt - thus it caused massive internal bleeding and death.

The rain gets fixed with a change in the weather system by making the rain cooler (again correct me on this subject if i am wrong) thought similar things can happen in really hot biomes.

To humans (of the real-life) that does normally not happen because we have systems to regulate our inner temperature - we can even go to a Sauna which is far hotter. Our body-temperature hovers around 36.5°C normaly (actually between the boundary's 35.8 and 37.2 acording to wikipedia) - we achieve this regulation thanks to a number of mechanisms like:

against hypothermia:

- controlled circulation which allows us to spread heat over the available surface faster (or to loose less energy to the outside)
- shivering

againts overheating:

- sweating - or other kinds of evaporation cooling
- thermic windows (for things with fur)

and a number more which i dont to recap. Alas i hope yo see what i want to say (this is the actual suggestion): Each and every biological creature should have the ability to regulate its inner temperature actively so it can survive in its (choosen) biome to a certain amount of time. This shouldnt come without drawbacks - a hypothermic creature should burn fat faster (needs more food) while a overheated creature should loose water in bigger amounts (needs more drinks).

Thanks to the circulation system the energy a warm spatter of water adds to a dwarf should be distributed over the entire body or at least faster over a bigger surface.

Actual season related temperature regulation would also be neat for the the dwarf and adventure mode in ways that animals shift theyr furs and intelligent creatures wear appropriate clothing. Including Temperature related effects and clothing would be also neat ala "The thick clothed dwarves are laughting at the naked shivering dwarf."




Please excuse if i have written nonsense.

Ok. This is only a very minor suggestion but i have seen some Communities (mostly browser games) that had used the Forums logins for the bug-tracker. Thus if someone was registered on the forums (s)he hadnt to be registered again on mantis. Could we get the same for our tracker?

Its not that much work on the Php-side and it would make the forums more comfortable respectively the tracker more accessible.

Copulation preferences - simplified approach on mate selection and Animal variations

Ok this suggestion was inspired after reading the Ark-project thread. Said project has the problem that many genuses of a species are pretty much the same except for minor differences, so these genuses are lumped together in one creature. This isnt a bad thing,actually its pretty pragmatic, but i thought to myself that we can get this "just different in colouration" variants of a creature back into the game without braking the simplification in the raws.

My suggestion consists of two parts:

1. A change in Worldgen.

Upon worldgen (as i understand it) the game places a number of individuals. I suggest to place now groups of individuals which are pretty similar, say 20 red canary's instead 20 individual ones with different colours.

This would mean that the canary´s in one region stay red thanks to the new genetics system. This approach has 1 mayor problem: Blue canary´s.

Well so you dont get me wrong: I mean another group of the same base-raw but with another variation. Soon enough the region we are in will have a monopole of one canary variant or a amalgamation of the two kinds.

Thats there the second part of this suggestion should apply.

2. A change in behaviour: Copulation/sexual preference

Normally animals copulate at random.

To keep the two populations of canary´s apart the red canary´s should prefer red canary´s and blue ones their blue companions.

This can be done by a little bit of statistics Math stuff. Before a bird/mammal/whatever gets born its parents get determined. In the pairing (from the females view) of the parents you can now look for the Characteristics of the available males. If there is a father that is more similar to the females own creature variation (and preferable not a brother or something) this one is preferred over the other males.
You compare just the tags like the colour, the beak-size, body-size etc. and give the differences a certain score of points. Differences (for example!) that are higher then 10% get 10 points, over 20% 50 points, over 50% 100 points etc. Different colours / patterns get say 75 points. Being closely related could give 200 points.

The males with the lowest scores get chosen for mating.

With our canary example this means that, for a red canary female, every red canary male would have a score of 0 and every blue canary a score of 75. So our female could now at random choose from the available males of the red canary population. Only if no other red canary´s are around for mating or the all other red males are taken a blue canary should been chosen as mate.



Optional: Fitness and quirks:

Overall fitness (being stronger, faster etc.) should play a rule so differences in this direction should get a slightly lower weight then a negative difference (a weaker mate).

Quirks like a illness or being disabled should lower the mating chance a lot. This prevents that a very strong but limping male can propagate its quirks onto the next generation.

Optional: Lineage weighting:

The lineage of a creature can play a role. A red canary that has a pure red canary Ancestry could prefer red canary´s even more then a red canary which ancestry is slightly mixed.



The behaviours should variate a bit between individuals and in some cases should leave the borders of the own genus - Humanoids/humanoid (dwarf/elf, Goblin/rat-men) pairings, Feline/feline (Tiger/lion) etc.[/color]

Cause the question "Is it possible to build an advanced Siege AI" came up in the "Invading adventurers" Thread and i didnt want to disturb the ongoing discussion there, i decided to suggest an Approach on more advanced sieging Ai here.

I want to begin with an qoute from the above mentioned "Invading adventurers" Thread:

Quote from: Footkerchief on December 31, 2008, 02:24:32 pm

Quote from: LegacyCWAL on December 31, 2008, 01:15:25 pm

I'm sure that having siegers learn your defenses and take steps to counter them in future attacks has been suggested, but I thought a specific implementation could be counts of how many died and in what way.  So it would not just count deaths, but count deaths by way of:  melee combat; ranged combat; siege engine fire; kill-traps; cage traps; falling or other bridge-related actions; drowning; and burning by magma.  Then the next time, bring along some sort of counter to that defense.

For instance, if a lot of attackers are lost to traps, they might eventually start herding cannon fodder into them to set them off without "real" losses.  If a lot of attackers fell to Marksdwarves in a few sieges, they could bring along catapults that would fire projectiles that could splash through fortifications, or marksgoblins with Pavises.  If a lot of attackers drowned, they might bring an amphibious unit that could destroy underwater buildings.  And if there's a trend of attackers being chopped up by axedwarves, a few adventurers in =steel plate mail= would work them over but good.

Excellent ideas.  Also, the game could use civ-specific "traffic designations" so that invaders could try to avoid spots where their comrades died to traps earlier.

Excellent ideas or not, I am not sure that it's possible to create an advanced AI like that. Toady is a magician, so maybe he could code in an AI like that, but to be honest, I will be very happy if we gonna have a "working" tunneler system even.

First off i think it wouldnt be that hard to find an way for an siege AI.

The problem the current siegers have are from my point of view are 3.

- They dont learn from Mistakes
- They cant destroy Walls or cross moats and other stuff
- They are to straight forward.

To solve this problems i want to line out some ideas.

1. Learning

An Siege Ai should have an way to "learn" certain things. If you want to Learn you need informations. The Information system i envision has 3 parts:


- The in the quote mentioned "killcounter"
 
- Some sort of Map for the Ai. This map would have:

   - An plan of your the terrain.
   - An plan of encountered moats, walls
   - Mele-zones. - Zones were your Siegers are killed by the besieged in melecombat
   - Ranged-zones. - Zones were your Siegers are killed by the besieged with Bows/crossbows/Siegeweapons and from where the bolt came. 
   - Dangerzones. - Same as killzones but for falls, Smashers and Drowners.
   - Trapzones. - Same as killzones only for traps
   - Lootzones. - Quarters, Stockpiles, Baracks.
   (- Entrancepoints.)




- an Objectiv system which determines what has to be done to avoid loses and deal with "Stuff" at the same places as before by the Knowledge gathered before the recent siege.



The Informations for the first siege can be gathered by snatcher, Thiefs and "Scouts" (Invisible and Trapavoid).

Lets asume th Gobs try there first siege on your fortress and send out snatchers as well scouts. How would the information gathering work?

The Snatchers as the Scouts can see stuff right? So say an scout sees an wall with an marksdwarf on it or Fortification. The area of an Bowshot around the wall now gets marked as "Ranged-zone", if an siegeweapon is seen for an siegeweapons range. The wall itself gets marked naturaly too.

Meanwhile the Snatcher enters the fortress. He walks over some traps without activating them. The tiles with encountered traps of any sort are marked as Danger-zones. On his way trought the fort he sees too Baracks and Weapon-stockpiles which get marked as Mele-zones. An encountered r Bridge gets marked as Dangerzone. Quarters and valueable Stockpiles are marked as lootzones by our goblin.

Should one of our Scouts or Snatchers leave the map via the mapedge he brings the infos back to our Goblinculture.

Same goes if the siegers attack, get beaten, and retreat with some minor differences. Rangedzones get assigned to places were Gobs etc are killed by ranged weapons. This area gets now expanded one bowshot in the direction from there they did come.

Now kicks the real AI in by determining the priority of targets and how they are handeled (an "deal with order" if you want). If you are an sieger you would set prioritys too. Surely baracks and weaponstockpiles get the highest cause this places are the ones were the most resistance can be asumed, Thought diffenernt siegers have different prioritys. To reach that priority targets we can use the standart Pathing with some Modifications. Normaly the shortest direct path is seeked, here we seek the "safest" path. This can be done if we use our priviously marked "Zones" by giving this zones an higher value to path trought.


2. Dealing with "Stuff" on the Path in and Pathvalues.

We have now our path but we have to still deal with the things we will encounter.

Traps could be disabeled by by engineers or by driving "Meat" throught it they get an pathvalue of 20.

Atom Smasher could be disabeled by "supports" or by newlybuild walls. They should get an very high pathvalue like 10000.

Known Danger zones get no pathvalue instead they get marked an impassable so this way our siegers would  path around it.

Moats (pathvalue 25?) get "filled" or bridget as long they arent covered by an rangedzone. If they are covered by an Rangedzone the Sortingalgorithm of the prioritytargets rates the source of the rangedzone higher so it gets destroyed before the moat is crossed. "Destroyed" means killing marksdwarfs/violent deconstruction of dwarven artilery with own siegeweapons or Ranged military.

2.1 Walls Climbing and Digging

Not only digging should be possible for siegers building siegeweapons and other stuff should be an option to get around player and dwarf made defenses. Even climbing an wall could be an good tactic.

Walls get an very high Pathing value say 150. Walls that overlap with Rangedzones get sheduled for being shot with siegeengines if this Zones overlap with our path. If an wall without Rangedzone is in the way an there is only an long dangerous way around it or no way around it, it gets torn down by Buildingdestroyers (trolls) or if in range of one of the siegeweapons by those.
Our trolls should atack an by ranged weapons defended wall if an siegeweapon cant be brought in position for this wall. The Ai should here Rate the Trolls lower then siegeweapons and reenforce them with the aid of ranged troops.

Digging (if tunnelers get in) digging gets an value of 300. Damp and Hot walls should be avoided.

Climbing an wall should be possible only for some creatures thought climbing an wal schouldnt be so fast there should be an chance to fall and the sieger should have an very low hitchance if he attacks during climbing. Climbing in very heavy gear should be harder.

Since the Siege Ai knows what kind of resistance it has to deal with and how it should select the right kind of siegers too. An heavy trapped and walled of fort needs more siegeweapons and Trapdestroyers then an fort with an palisade and 50 Axemen.

(Additional tactics would be possible if "siegeshields" against Ranged weapons, Mobile bridges for crossing moats, and mobile Ramps could be moved too into the Batle too.)

3. Breaking straight forwardness

The straight forwardness of the sigers is one mayor problem. If there is one way in they use that way without thinking currently. Even with The Points 1 and 2 implemented the attacks the players will see would get predictable with the time.

This can be handeled by some things.

First of the Siege Ai could not only math out one way in it could math out  different a bit more unlikely ways.

Different plans at the same time. the Siegers could try to break in the fort on 2 or more different spots.

3.1 Sidebhavior (Tortureing the besieged).

The siegers can torture our besieged fort in different ways.

One way is shooting the surface fort to dust: The typical grinding siege if you want.

Poisoning the watersource: is another. If the fort has one mayor watersource like an river or brook it can be poisoned making the dwarfs inside of the fort sick.

Start Fire and theft: Once inside an fort it would be devasting if the siegers could destroy your Farmplots and even more so if they burn your wood and Foodstockpiles. The siegers could also grab valueables and better equipment if they get there Hands/Paws/Claws/Tentakles/Pseudopods on your fallen military or the weapon, armor and ammo Stockpiles.

Freeing captives: If there are captives in the fort te siegers could free them so they join in there fight or produce chaos if there are creatures found.

Slavehunting. This could be one of the mayor causes for an attack too. After the siegers broke into the fort they try to knockout or only wound Children and untrained dwarfs so the following slavedrivers can collect them.

Bridgeheads. Once inside the fort the sieger could errect Bridgeheads. By stationing guards and building provisoric defensesystems by manipulating your traps and with enought time setting up own traps.

Freeing HFS: Do i have to say more?




I hope this Suggestion was entertaining and not to full with mistakes. If there are question i try to answer them all. Feel free to discuss and to show me my failures if i made them (only logical ones please i know that my english is horrible).

I am currently modding some Dragon and Drake races for my personal fun. I want them to repoduce and a 3(Egg - Whelp -> Drageon / Drake) teir evolution for all of them. My Question is how do i mood my Eggs that they are:

1. still Creatures
2. immobile with no Atack
3. on ground

optional:
4. in litle clusters
5. defendet by the Mother

Just for info:

Some Whelps are Swarm creatures some of this with a ranged Atack and/or Flying or with none of that depending on the dragon / Drake race