Print

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

[Pirate Bob]

Not research, but a simple obserevation - in 40.01 I witnessed a bowgoblin emty his entire load of arrrows into a single capybara, before he went and finished the poor creature in close combat. He was carrying copper bow and arrows.

Did the goblin empty his quiver *into* the capybara, or just *at* it?  Small creatures are very hard to hit with ranged weapons, so my guess is he didn't hit it at all.  If you think he actually hit it, then I will have to put this on my list of future !!SCIENCE!!.  Also, that cat disturbs me. 

The cat knows what you did.
I didn't read the logs carefully, but SoundSense implied that the animal was hit. The capybara was already unconcious and severely wounded, but alive.

Sounds like !!SCIENCE!! is needed.  This should be pretty easy for me to test.  If I don't post something in a few days, feel free to remind (i.e. PM) me.  Probably what happened is only a few of the goblin's arrows hit the poor rodent, but I will run a test to make sure there isn't something funny going on with very small creatures.  I have previously found that creature size had no impact on the effectiveness of armor against bolts, but I don't think I went to extreme sizes.

It does seem easier to reflect them with shields, however.
Playing adventure mode, I encountered a Goblin Bowman who shot all but 2 of his copper arrows at me before trying to flee.
All of those arrows would have hit me if I didn't block with my shield, with some help from batting them out of the air (ninja dwarves) and simple dodging.
Note that I was a Proficient Shield User and Dodger.
I was unscathed.
Only one or two of the arrows actually missed without my interference.

I'm going to make an assumption that you are somewhat smarter than the AI, and kept yourself (and hence your shield) pointed towards the enemy.  At least in arena tests, dwarves being shot from behind don't even turn around to see what's hitting them, so that could render them somewhat vulnerable to ranged attacks...

[Mel_Vixen]

I think the problems here are that DF assumes that you have a continuous force behind the bolt and that the Armor-metal is only paperthin so to say. Imagine a block of balistic gelatin, if you fire a bullet at it, it will bore in but somewhere down the way it will stop.

If the simulation assumes that the projectile has a continuous force behind it, it would go all the way through.

The second point of failure might be that the Armor is virtually paperthin in the simulation so that the projectile dosent dissipate enough energy (if at all) on the Armor since the thickness of the armor isnt taken into account.
 
Also bolts should be more deadly against Chainmail - if the bolt hit, normally it should rupture one or two rings of the mail thus removing any resistance from the armor. As it stands now i would think the [STRUCTURAL_ELASTICITY_CHAIN_ALL] token (that raises the "STRAIN_AT_YIELD" to 50000) is what makes the chainmail better at blocking arrows/bolts.

[cikulisu]

isn't it kind of ridiculous that arrows get through steel plate by default anyway? i mean, at very close ranges maybe. but at long? the arrow will bleed off a ton of it's kinetic force at even medium ranges. no way a high quality steel armored dwarf should die to a +bone bolt+

just seems silly. if he gets hit in the throat or something maybe, but the chest? makes no sense.

[Pirate Bob]

I think the problems here are that DF assumes that you have a continuous force behind the bolt and that the Armor-metal is only paperthin so to say. Imagine a block of balistic gelatin, if you fire a bullet at it, it will bore in but somewhere down the way it will stop.

If the simulation assumes that the projectile has a continuous force behind it, it would go all the way through.

The second point of failure might be that the Armor is virtually paperthin in the simulation so that the projectile dosent dissipate enough energy (if at all) on the Armor since the thickness of the armor isnt taken into account.

I'm not sure I understand what you are saying, but I don't think it's correct.  The reason bolts/arrows defeat plate armor is very simple.  They are modeled as being very massive (about 1 kg), traveling at very high speed, and having a very small impact area.  There is nothing about the force being applied continuously or the armor being very thin.  As far as I can tell (from extensive testing, see the wiki article and previous thread I linked in the OP), the mechanics for armor are fine, more or less (good armor stops melee attacks most of the time).  It's just that bolts and arrows have such enourmous momentum that they cannot be stopped.

Also bolts should be more deadly against Chainmail - if the bolt hit, normally it should rupture one or two rings of the mail thus removing any resistance from the armor. As it stands now i would think the [STRUCTURAL_ELASTICITY_CHAIN_ALL] token (that raises the "STRAIN_AT_YIELD" to 50000) is what makes the chainmail better at blocking arrows/bolts.

Agreed.  The [STRUCTURAL_ELASTICITY_CHAIN_ALL] causes armor to convert incoming edged attacks to blunt attacks.  This makes them slightly less deadly, but still quite devastating in the case of bolts/arrows.  I believe that chainmail can fail to convert edged to blunt if the attack has extremely high momentum, but even bolts/arrows do not have enough momentum for this, and I can't remember if this is true.

isn't it kind of ridiculous that arrows get through steel plate by default anyway? i mean, at very close ranges maybe. but at long? the arrow will bleed off a ton of it's kinetic force at even medium ranges. no way a high quality steel armored dwarf should die to a +bone bolt+

just seems silly. if he gets hit in the throat or something maybe, but the chest? makes no sense.

By default, plate armor does not cover the neck or face (including the throat) or upper arms, and therefore would provide no protection to these areas even if it did normally stop bolts.  Chainmail covers the neck and upper arms, but I don't think it covers the face.

Since this thread seems pretty well derailed anyway, what I would ideally like to see is modeling projectiles as hitting armor at some angle, with the momentum transferred (and hence damage and armor penetration) proportional to the cosine of the incident angle.  This would mean that even crappy armor would have some chance to deflect projectiles if they hit at a glancing angle, while a perfectly true hit might penetrate even good armor.  In the simplest case, you could just still assume each part is hit with likelihood proportional to its size, and treat all parts as spheres.  Then the probability to hit at given angle is just a simple function which does not depend on body part size [Edit - just set the angle equal to asin(sqrt(Rand)), where Rand is a random number equally distributed between 0 and 1].

If you want to get even more realistic, you could allow shooters to target a given body part, and then calculate how close they come to hitting the center of this (based on skill, range etc).  A miss would then have a chance of hitting other connected body parts, depending on their size and how far away they are from the targeted part.  This would require somewhat more complicated calculations, but would be fun as you could realistically target body parts.  Also, this would make legendary archers more deadly than novices, as they would be able to hit targets dead center.

I suspect Toady has something more detailed (if not this, then something else) in mind, which is why he's not bothered to fix the current system - he probably wants to do a major rewrite at some future time to make ranged attacks more realistic.

[Pirate Bob]

Not research, but a simple obserevation - in 40.01 I witnessed a bowgoblin emty his entire load of arrrows into a single capybara, before he went and finished the poor creature in close combat. He was carrying copper bow and arrows.

I did some testing vs. capybaras, and it doesn't appear anything strange is going on.  It does appear that I don't know what a capybara is (I thought they were rat sized), as they are actually rather large (45000 vs. 60000 for a dwarf).  It takes on average 12 hits to kill a capybara (using copper crossbows and bolts).  While projectiles are extremely good at incapacitating targets (via chipped/broken bones), they only occasionally hit vital organs causing instant death.

The rate of an unskilled dwarf hitting a capybara appears pretty high, although a few of them needed upwards of 40 shots to kill their targets, and that was with the target in a 1x1 cell 2 spaces away from the shooter (through a fortification).  If the capys could dodge, the hit rate would be lower.  Anyhow, since capybaras are the size of large dogs, it makes sense they take a bit of damage to kill, so it is not particularly odd for it to take more than 30 shots to kill one, although typical would be more like 15-20.  I searched through the logs, and there are no strange deflections etc.  It was an interesting experiment though...

[cephalo]

I didn't play that much in 34.x, but I didn't get the impression that ranged combat was particularly deadly. That may be because in 40d, crossbows were like machine guns. A single crossbowman could mow down 6 goblins before they closed to melee range on flat ground. After that, in 31.x ranged combat was heavily nerfed and marksdwarves wouldn't train in any case. What we wanted at the time, was for crossbows to be deadly, but also slow to reload. Is that not the case now?

[Pirate Bob]

I didn't play that much in 34.x, but I didn't get the impression that ranged combat was particularly deadly. That may be because in 40d, crossbows were like machine guns. A single crossbowman could mow down 6 goblins before they closed to melee range on flat ground. After that, in 31.x ranged combat was heavily nerfed and marksdwarves wouldn't train in any case. What we wanted at the time, was for crossbows to be deadly, but also slow to reload. Is that not the case now?

The behavior of crossbows against unarmored targets is quite good now.  I'm not sure if "deadly" is quite the right word, but hits from bolts cause lots of pain (chipped bones, torn internal organs, etc).  This leaves the target fighting at a severe disadvantage due to pain modifiers, and they are easily mowed down in melee.  It takes many hits to actually kill something with a crossbow alone, as you have to either get a lucky shot to the head, heart, lungs, etc, or enough hits that they bleed out.  They also fire at a fairly slow rate. 

The only issue with them currently is they completely disregard all plate armor, including adamantium, due to the incredibly high momentum of crossbow bolts.  I have found that lowering the momentum of bolts/arrows make no significant difference in the damage to unarmored targets, while allowing armor to have intermediate effectiveness, which can be tuned by adjusting the bolt momentum (roughly equal to [SHOOT_FORCE]) to somewhere between 50-100. 

All of this appears to be completely unchanged since 34.11.

[cephalo]

Back in the day, people argued that crossbows were specifically designed to pierce armor. I'm not sure if that's true, but I think they sited certain history channel shows reenacting the effect of crossbows on sheet metal or something.

[therahedwig]

Back in the day, people argued that crossbows were specifically designed to pierce armor. I'm not sure if that's true, but I think they sited certain history channel shows reenacting the effect of crossbows on sheet metal or something.

Perhaps, but when you actually go into the bug-reports, the bug itself shows not as being actually a problem of 'metal doesn't stop bolt' but rather, 'crossbow is equivelant to railgun'.

There's a rounding error in the algorithm to calculate the speed of a crossbow bolt, which causes all crossbow bolts to shoot at the speed of sound. Something which, regardless of how you feel about historical accuracy, is not intended.

[thvaz]

Crossbow bolts could pierce plate armor in short ranges, but the resulting damage wouldn't be enough to kill or even incapacitate the victim. Longbows couldn't pierce plate amor either.

Some say that most of the damage caused by arrow/bolts volleys were psychological. Most people wounded by arrows, even in pre-plate armor era, died by infection instead of direct damage by the arrows/bolts.

[Agent_Irons]

Crossbow bolts could pierce plate armor in short ranges, but the resulting damage wouldn't be enough to kill or even incapacitate the victim. Longbows couldn't pierce plate amor either.

Some say that most of the damage caused by arrow/bolts volleys were psychological. Most people wounded by arrows, even in pre-plate armor era, died by infection instead of direct damage by the arrows/bolts.

Currently, with vanilla raws, there is no decrease in lethality by applying any amount of armor. Even wooden bolts penetrate adamantine armor 100% of the time. In the vanilla testing in the great !!SCIENCE!! threads of versions past, with many thousands of marksdwarves firing many many thousands of bolts, a handful of deflections were recorded against armored targets. All of them were deflections off an eyeball, not deflections off armor. The only decrease in lethality observed was using metal bolts against armor of the same metal. In this case, rarely, the bolt would merge with the armor or something and turn it into a hammerblow instead of an bullet, letting it jam hands into wrists and so on. Notably upped the suppressive capability, but lowered the lethality(per shot, not overall. overall lethality was always 100%. Sometimes it took longer, though) by 10% or so.

Nerf your SHOOT_FORCE to 50 and your SHOOT_MAXVEL to 200. Steel turns aside wooden and bone bolts ~100% of the time, adamantine is bulletproof(but doesn't cover your throat and eyes etc) and copper/bronze armor is fairly good but not infallible protection against wood/bone but nearly useless against copper/iron/silver bolts. NB: your wooden bolts will be slightly less effective against unarmored targets, but still deadly. It's up to you whether this counts as a nerf or not. On the one hand, armored opponents will be a little harder to murder and hunting may take two trips. On the other hand, now your adamantine hammerlords need fear nothing.

[Pirate Bob]

Back in the day, people argued that crossbows were specifically designed to pierce armor. I'm not sure if that's true, but I think they sited certain history channel shows reenacting the effect of crossbows on sheet metal or something.

Perhaps, but when you actually go into the bug-reports, the bug itself shows not as being actually a problem of 'metal doesn't stop bolt' but rather, 'crossbow is equivelant to railgun'.

There's a rounding error in the algorithm to calculate the speed of a crossbow bolt, which causes all crossbow bolts to shoot at the speed of sound. Something which, regardless of how you feel about historical accuracy, is not intended.

It is true that there's a rounding error in the algorithm, but that does not actually cause the very high momentum.  That just causes heavier metals (copper, silver) to be more effective at piercing armor than lighter metals (iron, steel) if you mod bolts to have lower momentum.  With vanilla parameters the rounding error is irrelevant.

Bolts all pierce armor because they have a very small contact area (2) and a very large applied force (1000). 

I agree that arguments can be made that bolts should pierce armor some to most of the time.  I personally think that steel armor should stop bolts a large (say at least 50%) of the time, but could be convinced otherwise.

However, I think it's ridiculous that adamantine armor does absolutely nothing against crossbow bolts.  It is my understanding that, in everything but name (due to copyright concerns), adamantine is supposed to behave like mithril, which can completely stop a spear wielded by a cave troll (or was it an ogre?). 

Regardless, it's completely irrelevant what any of us think.  I pretty sure Toady knows how he wants armors to behave, and when he gets around to it he will adjust crossbows to respect whatever this is.  It's also my understanding that the current ranged attack system is just a placeholder for something more detailed, and that Toady very quickly "balanced" ranged attacks in 34.11 by increasing their power until a squad of marksdwarves would be of similar power to melee attackers.   It is now absolutely true that, on average, crossbows are decently well balanced.  In fort mode, you are extremely unlikely to ever notice the failure of armor.  It is really only in adventure mode that you're likely to notice the steel/candy helm you worked so hard to get did nothing against a kobold bowman.

I would not be at all surprised if Toady does not rebalance ranged attacks in this version, and instead waits until some future time when he does a full update of the ranged attack system.  If by some chance he does, I hope that all the testing we have done will help him to choose parameters that he feels are appropriate.  Otherwise, they should be useful to anyone who wants to mod the game to make bolts behave however they want.  I have an example in my sig on DFFD if anyone is interested, and I am happy to help with designing parameters to get any behavior you want.

[cephalo]

I agree that arguments can be made that bolts should pierce armor some to most of the time.  I personally think that steel armor should stop bolts a large (say at least 50%) of the time, but could be convinced otherwise.

I think that is a reasonable target. I don't think candy armor should be bullet proof, but maybe 80 percent effective against bolts. I think the material of the bolt should matter, but only a little.

[Shizmoo]

Its a gamebreaking bug which im surprised Toady hasn't made a priority of. Same with throwing.

[Pirate Bob]

I agree that arguments can be made that bolts should pierce armor some to most of the time.  I personally think that steel armor should stop bolts a large (say at least 50%) of the time, but could be convinced otherwise.

I think that is a reasonable target. I don't think candy armor should be bullet proof, but maybe 80 percent effective against bolts. I think the material of the bolt should matter, but only a little.

Within the current armor and ranged attack system, I don't think it's possible for candy to be 80% effective and other armors to have any effectiveness at all, although I would have to check this.  Candy is so much better than everything else that if you make bolts powerful enough to penetrate it with any frequency they penetrate everything else 100% of the time.

Keep in mind that plate armor does not cover all areas - the upper arms, neck and face are exposed.  Therefore even with 100% effective armor you do not get 100% protection.

I would like to see that ranged attack system adjusted so that there is more variance in the probability a bolt will penetrate armor.  Right now, it goes from 100% penetration to 100% protection over a fairly narrow range of bolt momentum.  If, for example, Toady added in simple modeling of the angle at which bolts impact armor, then this would give even the worst (i.e. copper) armor a slight chance to deflect bolts if they hit at a glancing angle, and momenta could be set so that a dead on hit will penetrate even adamantine.  I gave a more complete description of this a few posts back. 

I am working on testing how melee attacks are blocked by armor now, and will post a new thread on that soon.  I plan to also compare with ranged attacks, and assuming the differences are as large as I suspect, I might try to model how some slightly more detailed physics (such as deflection angle) could be used to give a more realistic result.  This would be largely academic, as I suspect Toady already has his own ideas of how to improve ranged attacks that he will implement when he gets around to it.