1
DF Gameplay Questions / Invader climbing, jumping, etc
« on: February 05, 2017, 12:18:11 pm »
I've tried twice now to set up a fort to perform an in-depth study on the science of goblin gymnastic ability, only to see them both crash just days before a seasonal save. I don't have a lot of free time anymore, and what free time I do have I'd like to dedicate to my current fort, where I've been making informal experiments between invasion waves (I've lost track but about 2 to 2.5k invaders). So I'm afraid these observations are not as 'scientific' as I'd like. Since my current fort has all sorts of invaders, several of these observations are of Fortress Defense races. I've examined the raws and the only differences I can find re: climbing is some variations in their "CLIMBING_GAITS" speeds. There is a NATURAL_SKILL:CLIMBING tag but goblins don't have it and only the serpentmen have it in my world; for this reason I ignore experiments with them. I haven't seen them often enough as invaders to really comment on what difference that tag makes.
The experiments I've done are pretty simple and usually involve intentionally poorly build animal towers.
Tower 1: Starting with a single rough log pedestal at z, I built a 3x3 platform above at z+1 with fortifications in the cardinals and walls in the intercardinals. No roof. Cat as bait.
During a warwolf invasion, on multiple savescums I watched as the warwolves made a beeline straight for this tower.
Interesting observation #1: Stepping through carefully, I watched them leap straight up to the walls at z+1. Many of the diagrams floating in the forums and wiki make it appear that invaders use a sort of diagonal climbing, starting at the pedestal wall and then climbing upwards and outwards. This might be true for scaling tall walls but if the z-separation is just 1 between ground and wall, they will leap straight up instead.
Also, the warwolves only leaped and scaled the wall portions of the tower - on several reloads, I am 100% certain they ignored the fortifications entirely (see Tower 4). Once up on top of the wall, they simply hopped down and killed the bait.
Tower 2 is similar to Tower 1 except the base at z=0 was a 3x3 of rough wood, upon which was placed the same 3x3 tower platform as above (walls in corners, fortifications at the cardinals). It met the exact same fate as Tower 1, except the warwolves simply scaled the first z-level of wall. Once again, they did not climb the fortification sides, they always chose the corners where it was wall upon wall.
Tower 3 was to test the leaping mechanism and to verify that one of the structures on the wiki is not actually safe because of it. This is the "safe if walls protrude out at least two tiles"
In this case, I built a single rough wood pillar at z==0, a 5x5 rough wood wall platform at z+1, and a kitten at z+2. Invaders can see the kitten in the distance as they come down a cliff slope. The warwolves had no problem leaping up and scaling the edge of the platform, like so:
This wall might be safe if additional z-levels are present to prevent leaping, but I've yet to test that.
Tower 4 is similar to Tower 2, the only difference being that all the walls are replaced with fortifications, so the entire 3x3 platform is one tile surrounded by 8 fortifications. This is to test the theory that invaders can't actually climb fortifications - at all. A single cat is used as bait; no roof.
The warwolf invasion walked right by it, despite a perfectly visible cat.
At this stage, I am pretty much 100% convinced that invaders cannot actually climb through/over fortifications. I think in the rare few instances that I have seen invaders in my battlements before I enclosed them, they managed to use nearby walls to climb around the fortifications.
Experiment 5 - I like to use bait animals as ammo soaks, typically pasturing them with a gap between them and invaders. In my current fort, I didn't make this gap wide enough, and goblins simply leaped the 2z gap to butcher my cows. What was interesting was that afterwards, they stayed right where they were. Their entire army marched right past them into the fort and they twiddled their thumbs in the pasture instead of leaping back across to join them. However, when my melee dwarves butchered the army and came streaming out into sight - the goblins in the pasture immediately leaped across to attack.
Based on this and a number of similar observations, I believe that the climbing and jumping code is only activated when the invader has a hostile in sight. In other words, when invaders are doing their slow march into a fort, I don't believe they will climb even the simplest wall, or leap even the tiniest gap, unless they have a target in sight and are actively seeking to engage a target like an archer on top of the wall or wildlife/livestock on the other side. I believe there are two pathing codes; the "march into the fort" code (with its own mysterious "what are they targeting?") and the "Chaaaaaaaarge!" code when they sight something they want to kill. I think only the latter uses climbing and jumping.
So a simple experiment I tried was to place a cat in the middle of the typical invasion path, and surround it on all 8 sides with a rough wooden wall. It is therefore out of sight, but the walls are trivial to climb. It was completely ignored by invaders.
I don't have a vampire but I would love to follow up the experiment by placing a dwarf vampire there, to see if the presence of an actual fort citizen matters. I'm fairly certain that changes nothing, since in my experience invaders don't seem to path towards or know about citizens at all unless they're unlucky enough to come into direct sight.
A more useful test would be to simply start a fort surrounded by a basic, easily climbable wooden palisade, just 1-z tall. During an invasion, call the citizens in and bottle the fort up using a drawbridge. Make sure that not a single living thing is visible on the wall. I wager the invaders will not even path to the fort.
That's what I've got so far. I have a whole list of experiments I'd like to try (in particular, making the support column on the watchtowers 2z and 3z high to test jumping and with different materials to test climbing limits) and I'll update the thread when I ever get around to them.
Sadly, some of this is too close to anecdotal than proper 'science' but the current science on climbing and jumping is pretty murky and I feel this is not generally known. Please, if you've got contrary observations or simple experiment ideas, bring them up!
The experiments I've done are pretty simple and usually involve intentionally poorly build animal towers.
Tower 1: Starting with a single rough log pedestal at z, I built a 3x3 platform above at z+1 with fortifications in the cardinals and walls in the intercardinals. No roof. Cat as bait.
During a warwolf invasion, on multiple savescums I watched as the warwolves made a beeline straight for this tower.
Interesting observation #1: Stepping through carefully, I watched them leap straight up to the walls at z+1. Many of the diagrams floating in the forums and wiki make it appear that invaders use a sort of diagonal climbing, starting at the pedestal wall and then climbing upwards and outwards. This might be true for scaling tall walls but if the z-separation is just 1 between ground and wall, they will leap straight up instead.
Also, the warwolves only leaped and scaled the wall portions of the tower - on several reloads, I am 100% certain they ignored the fortifications entirely (see Tower 4). Once up on top of the wall, they simply hopped down and killed the bait.
Tower 2 is similar to Tower 1 except the base at z=0 was a 3x3 of rough wood, upon which was placed the same 3x3 tower platform as above (walls in corners, fortifications at the cardinals). It met the exact same fate as Tower 1, except the warwolves simply scaled the first z-level of wall. Once again, they did not climb the fortification sides, they always chose the corners where it was wall upon wall.
Tower 3 was to test the leaping mechanism and to verify that one of the structures on the wiki is not actually safe because of it. This is the "safe if walls protrude out at least two tiles"
Code: [Select]
z+2
z+1 ██_██
z+0 →→g█___
In this case, I built a single rough wood pillar at z==0, a 5x5 rough wood wall platform at z+1, and a kitten at z+2. Invaders can see the kitten in the distance as they come down a cliff slope. The warwolves had no problem leaping up and scaling the edge of the platform, like so:
Code: [Select]
z+2 ↗→ c
z+1 W█████
z+0 ↑ █___
This wall might be safe if additional z-levels are present to prevent leaping, but I've yet to test that.
Tower 4 is similar to Tower 2, the only difference being that all the walls are replaced with fortifications, so the entire 3x3 platform is one tile surrounded by 8 fortifications. This is to test the theory that invaders can't actually climb fortifications - at all. A single cat is used as bait; no roof.
The warwolf invasion walked right by it, despite a perfectly visible cat.
At this stage, I am pretty much 100% convinced that invaders cannot actually climb through/over fortifications. I think in the rare few instances that I have seen invaders in my battlements before I enclosed them, they managed to use nearby walls to climb around the fortifications.
Experiment 5 - I like to use bait animals as ammo soaks, typically pasturing them with a gap between them and invaders. In my current fort, I didn't make this gap wide enough, and goblins simply leaped the 2z gap to butcher my cows. What was interesting was that afterwards, they stayed right where they were. Their entire army marched right past them into the fort and they twiddled their thumbs in the pasture instead of leaping back across to join them. However, when my melee dwarves butchered the army and came streaming out into sight - the goblins in the pasture immediately leaped across to attack.
Based on this and a number of similar observations, I believe that the climbing and jumping code is only activated when the invader has a hostile in sight. In other words, when invaders are doing their slow march into a fort, I don't believe they will climb even the simplest wall, or leap even the tiniest gap, unless they have a target in sight and are actively seeking to engage a target like an archer on top of the wall or wildlife/livestock on the other side. I believe there are two pathing codes; the "march into the fort" code (with its own mysterious "what are they targeting?") and the "Chaaaaaaaarge!" code when they sight something they want to kill. I think only the latter uses climbing and jumping.
So a simple experiment I tried was to place a cat in the middle of the typical invasion path, and surround it on all 8 sides with a rough wooden wall. It is therefore out of sight, but the walls are trivial to climb. It was completely ignored by invaders.
I don't have a vampire but I would love to follow up the experiment by placing a dwarf vampire there, to see if the presence of an actual fort citizen matters. I'm fairly certain that changes nothing, since in my experience invaders don't seem to path towards or know about citizens at all unless they're unlucky enough to come into direct sight.
A more useful test would be to simply start a fort surrounded by a basic, easily climbable wooden palisade, just 1-z tall. During an invasion, call the citizens in and bottle the fort up using a drawbridge. Make sure that not a single living thing is visible on the wall. I wager the invaders will not even path to the fort.
That's what I've got so far. I have a whole list of experiments I'd like to try (in particular, making the support column on the watchtowers 2z and 3z high to test jumping and with different materials to test climbing limits) and I'll update the thread when I ever get around to them.
Sadly, some of this is too close to anecdotal than proper 'science' but the current science on climbing and jumping is pretty murky and I feel this is not generally known. Please, if you've got contrary observations or simple experiment ideas, bring them up!
