Print

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

[expwnent]

Excellent work. Posting to watch.

[Deimos56]

Out of curiosity, would materials modded to be ludicrously heavy or absurdly light have different results?

[krenshala]

Assuming the current theory, as proposed by the OP, holds then denser materials will do more overall damage per hit.  e.g., slade bolts should do massive damage per hit, assuming the other variables are at least relatively close to those of iron.

[Urist Da Vinci]

...In any case, first I'd like to identify all parameters which influence ranged damage. I suppose the next one is MAX_EDGE of the ammo used. Does anyone know of any other parameters likely to be used in damage calculations for ranged weapons? Apart from quality, which I believe cannot be changed in the object testing arena (Can it?).

I assume something other than MAX_EDGE is used, because all metals apart from adamantine (100000) have the same value for MAX_EDGE, 10000. (Wood and Bone have 1000). If MAX_EDGE and SOLID_DENSITY are the only two parameters affecting ranged damage, then Heavier is always better as far as metals goes (again, except for adamantine)....

http://www.bay12forums.com/smf/index.php?topic=30026.msg1007133#msg1007133

Currently shear yield/fracture is used to determine edge effectiveness and the effectiveness of a material against a cut.  Impact yield/fracture is used to determine resistance against impacts.  Solid density will increase weight and therefore the impact of attacks.

Testing indicates SHEAR_YIELD acts like the Mohs scale of mineral hardness and determines what can cut what. I created a silver colossus and a copper colossus in the arena, and then tried having a dwarf grand master attack them repeatedly with copper and silver swords. The silver sword can dent the copper colossus, but the copper sword is deflected off the silver colossus. See also this bug report for a fix to obsidian swords, which can't cut as good as wood swords: http://www.bay12games.com/dwarves/mantisbt/view.php?id=5346

I'd also suggest modding the test subjects to remove random body size variations:

Code: [Select]

[BODY_APPEARANCE_MODIFIER:HEIGHT:75:95:98:100:102:105:125]
[APP_MOD_IMPORTANCE:500]
[BODY_APPEARANCE_MODIFIER:BROADNESS:75:95:98:100:102:105:125]
[APP_MOD_IMPORTANCE:500]It appears that all arena creatures are created with average ability scores, but the height is still chosen from the random set. This was first noticed while sciencing butchering results.

[Mura]

Vindication. It is sweet. 

[Urist Da Vinci]

...Thus it can be concluded that ammo weight is indeed used in some manner in ranged crossbow attacks, adding to armor penetration. This effect is practically unnoticeable on unarmored opponents, possibly due to the powerful parameters of the vanilla crossbow. If these parameters were weakened more data could be gathered. Other future works include testing the effects of sharpness, both sharpness and weight, testing default materials (instead of modified as opposed to modified ones), testing the effect of Marksdwarf, Archer, Shield skills, etc. etc.

For reference, the vanilla crossbow is so powerful, if you increase the contact area of the bolts from 2 to 40000 (battleaxe-sized), the bolts will chop off limbs and even decapitate/bisect creatures.

[misko27]

...Thus it can be concluded that ammo weight is indeed used in some manner in ranged crossbow attacks, adding to armor penetration. This effect is practically unnoticeable on unarmored opponents, possibly due to the powerful parameters of the vanilla crossbow. If these parameters were weakened more data could be gathered. Other future works include testing the effects of sharpness, both sharpness and weight, testing default materials (instead of modified as opposed to modified ones), testing the effect of Marksdwarf, Archer, Shield skills, etc. etc.

For reference, the vanilla crossbow is so powerful, if you increase the contact area of the bolts from 2 to 40000 (battleaxe-sized), the bolts will chop off limbs and even decapitate/bisect creatures.

On the other hand, if your launching battle-axes, why shouldn't they cut limbs?

Overall, absolutely Brilliant Science. the ultimate question of sharpness versus weight, how they relate, how does one compensate for the other, and ultimately, which is the superior metal, have yet to be answered. But when it does, I can only hope it looks as fancy as your study.
*applause*

[Podesta]

...Thus it can be concluded that ammo weight is indeed used in some manner in ranged crossbow attacks, adding to armor penetration. This effect is practically unnoticeable on unarmored opponents, possibly due to the powerful parameters of the vanilla crossbow. If these parameters were weakened more data could be gathered. Other future works include testing the effects of sharpness, both sharpness and weight, testing default materials (instead of modified as opposed to modified ones), testing the effect of Marksdwarf, Archer, Shield skills, etc. etc.

For reference, the vanilla crossbow is so powerful, if you increase the contact area of the bolts from 2 to 40000 (battleaxe-sized), the bolts will chop off limbs and even decapitate/bisect creatures.

This seems to be the common sense, and at the expense of sounding stupid, I dont think 20 iron bolts to kill someone who is 3 meters away so much overpowered...

Edit: forgot to congratulate OP. Looking forward for more papers like this.

[misko27]

...Thus it can be concluded that ammo weight is indeed used in some manner in ranged crossbow attacks, adding to armor penetration. This effect is practically unnoticeable on unarmored opponents, possibly due to the powerful parameters of the vanilla crossbow. If these parameters were weakened more data could be gathered. Other future works include testing the effects of sharpness, both sharpness and weight, testing default materials (instead of modified as opposed to modified ones), testing the effect of Marksdwarf, Archer, Shield skills, etc. etc.

For reference, the vanilla crossbow is so powerful, if you increase the contact area of the bolts from 2 to 40000 (battleaxe-sized), the bolts will chop off limbs and even decapitate/bisect creatures.

This seems to be the common sense, and at the expense of sounding stupid, I dont think 20 iron bolts to kill someone who is 3 meters away so much overpowered...

Edit: forgot to congratulate OP. Looking forward for more papers like this.

Refering to the way they often pucnh through armor. I apreciate it in general, but in adventure mode its a down-right killer.

[Loud Whispers]

Refering to the way they often pucnh through armor. I apreciate it in general, but in adventure mode its a down-right killer.

Meh. The bolts aren't what kill you, they just horribly incapacitate you. If you've got super dwarven willpower/endurance you can make it out there with a bolt sticking through your leg or something, otherwise you're going to die to a spearbold.

[Zivilin]

Hey Zivilin,

someone on reddit, latexified your post - your research in shiny research doc format. - http://dl.dropbox.com/u/2891221/DF%20study.pdf

Very... impressive O_o

Bones are very noticibly better than wood when it comes to bolt, I checked the raw and a lot of it's stats are much better than default, when it comes to everything other than mass ( I think MAX_EDGE is at default, so that too). My dwarves that uses bone for hunting takes down animals with fewer bolts and with far more peneration and tearing in combat log than with wooden bolts.

I think I'll add bone vs wood (same density) to my To Do list, out of curiosity. And this further confirms my suspicion that there are parameters other than MAX_EDGE/SOLID_DENSITY used in damage calculation.

Would love to see some iron vs. silver vs. copper numbers.

At some point I will definitely attempt to do a large ammo vs armor study on all major materials. For the moment I'm trying to refine my methods on some smaller projects. I'm getting a lot of good data and ideas from the feedback to this thread

...In any case, first I'd like to identify all parameters which influence ranged damage. I suppose the next one is MAX_EDGE of the ammo used. Does anyone know of any other parameters likely to be used in damage calculations for ranged weapons? Apart from quality, which I believe cannot be changed in the object testing arena (Can it?).

I assume something other than MAX_EDGE is used, because all metals apart from adamantine (100000) have the same value for MAX_EDGE, 10000. (Wood and Bone have 1000). If MAX_EDGE and SOLID_DENSITY are the only two parameters affecting ranged damage, then Heavier is always better as far as metals goes (again, except for adamantine)....

http://www.bay12forums.com/smf/index.php?topic=30026.msg1007133#msg1007133

Currently shear yield/fracture is used to determine edge effectiveness and the effectiveness of a material against a cut.  Impact yield/fracture is used to determine resistance against impacts.  Solid density will increase weight and therefore the impact of attacks.

Testing indicates SHEAR_YIELD acts like the Mohs scale of mineral hardness and determines what can cut what. I created a silver colossus and a copper colossus in the arena, and then tried having a dwarf grand master attack them repeatedly with copper and silver swords. The silver sword can dent the copper colossus, but the copper sword is deflected off the silver colossus. See also this bug report for a fix to obsidian swords, which can't cut as good as wood swords: http://www.bay12games.com/dwarves/mantisbt/view.php?id=5346

I'd also suggest modding the test subjects to remove random body size variations:

Code: [Select]

[BODY_APPEARANCE_MODIFIER:HEIGHT:75:95:98:100:102:105:125]
[APP_MOD_IMPORTANCE:500]
[BODY_APPEARANCE_MODIFIER:BROADNESS:75:95:98:100:102:105:125]
[APP_MOD_IMPORTANCE:500]It appears that all arena creatures are created with average ability scores, but the height is still chosen from the random set. This was first noticed while sciencing butchering results.

Like this! Thank You, for the interesting referred posts and suggestions, i'll definitely be using them. I suspected that the shear parameters could be somehow involved, now I have confirmation. I'll still test it, though.

...Thus it can be concluded that ammo weight is indeed used in some manner in ranged crossbow attacks, adding to armor penetration. This effect is practically unnoticeable on unarmored opponents, possibly due to the powerful parameters of the vanilla crossbow. If these parameters were weakened more data could be gathered. Other future works include testing the effects of sharpness, both sharpness and weight, testing default materials (instead of modified as opposed to modified ones), testing the effect of Marksdwarf, Archer, Shield skills, etc. etc.

For reference, the vanilla crossbow is so powerful, if you increase the contact area of the bolts from 2 to 40000 (battleaxe-sized), the bolts will chop off limbs and even decapitate/bisect creatures.

This seems to be the common sense, and at the expense of sounding stupid, I dont think 20 iron bolts to kill someone who is 3 meters away so much overpowered...

True, I used 'overpowered' here based more on empirical observations from the game itself (i.e. losing significant amounts of armor-clad dwarfs to archers etc), rather than the results of the experiment. There is even a Broken Arrow which attempts to address this by lowering launch force on ranged weapons and decreasing ammo penetration.

Also, the criterion I chose (average number of bolts used to kill) may not be the best for judging weapon power. In fact, the parameter I wanted to use in the first place was something based on wound types. I still think this will be better for most comparative studies, especially with ammo vs armors scenarios: How many type A bolts pierced through type A armor? How many were deflected? How many type B bolts pierced through type A armor? How many were deflected? How many type C... etc etc. I suspect there is very little point to unarmored tests, because dwarf skin/muscle is not very good at stopping piercing attacks on its own (predictable, really). The number of bolts to kill is a criterion perhaps better suited to testing the Marksdwarf skill? Again, I do not know the battle mechanics, so I can only hypothesize, but I think it wouldn't be a stretch if Master Marksdwarfs hit critical shots (Head, organs) more often than Novice ones. This is certainly testable, in any case.

[Di]

First of all, great job Zivilin.

Secondly, if one wants to have more accurate criterion for bolt effectiveness, I could suggest analyzing gamelog.txt. Of course doing that by self would be tedious, but it'd the most accurate data if someone wrote a program that'd check on how many times key phrases (like bruising/tearing skin/fat/muscle, bruising/chipping/shattering/jamming bone/skull) are encountered by which bolt material they're preceded and by which armor they're followed. Unfortunately, I myself can't do it at the current moment.

Finally, I went ahead and conducted vanilla bolts test by op-post method. I've tested wooden, copper and adamantine bolts.
As everyone already knows, wooden bolts are crappy, 20 out of 50 iron armored peasants without shields survived,  the average bolt amount for the is 50,3. Metal bolts groups had no survivors copper scoring slightly better 12,1 against 12,6 for bluemetal.

[krenshala]

Also, the criterion I chose (average number of bolts used to kill) may not be the best for judging weapon power. In fact, the parameter I wanted to use in the first place was something based on wound types. I still think this will be better for most comparative studies, especially with ammo vs armors scenarios: How many type A bolts pierced through type A armor? How many were deflected? How many type B bolts pierced through type A armor? How many were deflected? How many type C... etc etc. I suspect there is very little point to unarmored tests, because dwarf skin/muscle is not very good at stopping piercing attacks on its own (predictable, really). The number of bolts to kill is a criterion perhaps better suited to testing the Marksdwarf skill?

You could use unarmored (and unclothed) dwarves as the ultimate control group for your bolt type versus armor type testing.  This could also give a bit more info on bolt type verses flesh and blood as well, which might be worth it for its own sake.

While typing this I thought of a more complex testing method that might provide in some better, or at least more interesting, results. Take your current setup, but have the target be webbed in place so he can't move.  I know for melee testing this would ensure headshots (and the need for helms on the target to test armor) but I'm not sure if that is the case for ranged attacks against the webbed target.

This is based on reports of (sometimes otherwise unarmoured) dwarves with helms being perpetually attacked by giant cave spiders while webbed because the spider cannot pierce the helm, but always attacks the head because the victim is incapacitated by the webbing.

[Zivilin]

While typing this I thought of a more complex testing method that might provide in some better, or at least more interesting, results. Take your current setup, but have the target be webbed in place so he can't move.  I know for melee testing this would ensure headshots (and the need for helms on the target to test armor) but I'm not sure if that is the case for ranged attacks against the webbed target.

This is based on reports of (sometimes otherwise unarmoured) dwarves with helms being perpetually attacked by giant cave spiders while webbed because the spider cannot pierce the helm, but always attacks the head because the victim is incapacitated by the webbing.

Logged in the To Do list

Secondly, if one wants to have more accurate criterion for bolt effectiveness, I could suggest analyzing gamelog.txt. Of course doing that by self would be tedious, but it'd the most accurate data if someone wrote a program that'd check on how many times key phrases (like bruising/tearing skin/fat/muscle, bruising/chipping/shattering/jamming bone/skull) are encountered by which bolt material they're preceded and by which armor they're followed. Unfortunately, I myself can't do it at the current moment.

I didn't know gamelogs were stored, thank You for the information! I'll brush up on my programming skills and see if I can properly utilise this treasure trove of data.

In the meantime, I used the tedious "Look through reports and note in Excel" to do a small test on the MAX_EDGE parameter. Since the amount of useful data gained is small, I will publish this as a short in this thread instead of a full article.

The Effect of the MAX_EDGE Parameter on Crossbow Performance

METHODOLOGY

Spoiler (click to show/hide)

The experimental setup is the same as in the original post. The test groups were chosen as follows:

Team A Weapon: Cedar crossbows Ammo: 100 iron bolts Armor: None Skills: Proficient Archer Proficient Marksdwarf

| Team B | Weapon: Cedar crossbows | Ammo: 100 sharp iron bolts | Armor: None | Skills: Proficient Archer | Proficient Marksdwarf | | | |

| Team C | Weapon: Cedar crossbows | Ammo: 100 very sharp iron bolts | Armor: None | Skills: Proficient Archer | Proficient Marksdwarf | | | |

| Team V | Weapon: None | Ammo: None | Armor: iron breastplate |        iron mail shirt |        iron greaves |        iron helm |        2x iron gauntlet |        2x iron high boot | Skills: None

Iron: regular iron with MAX_EDGE = 10'000.
Sharp Iron: iron with MAX_EDGE = 20'000.
Very sharp iron: iron with MAX_EDGE = 100'000.

Three setups were tested: Team A vs Team V, Team B vs Team V and Team C vs Team V. In each setup each team consisted of 120 dwarfs.

Two criterions were chosen to analyse the effects of crossbow damage:

Criterion 1: The average number of bolts used to kill the armored dwarf (on a sample group of 120 dwarfs)

Criterion 2: The percentage breakdown of types of injuries inflicted on the dwarf. This was tested by reading the injury reports of 20 V dwarfs from each setup, which resulted in obtaining from 250 to 350 samples per setup. Only the most damaging injury from an injury message was considered per bolt, and only damage caused directly by that bolt, For example, if the injury message read:

The flying {sharp iron bolt} strikes Dwarf 179 in the lower body, bruising the muscle and tearing the lower spine's nervous tissue through the small iron greaves!

Then this was classified as a spinal tear.

If the report read:

The flying {sharp iron bolt} strikes Dwarf 133 in the upper body, bruising the muscle, jamming the left false ribs through the left lung and tearing the left lung!

Then this was classified as a jamming, since the arguably more severe lung damage was caused by the jammed bone, not the bolt itself.

The damage classifiers and their order of superiority was chosen as follows:

Blunt damage
Bruise < Chip < Jam < Fracture < Shatter

Edged damage
Tear < Severe < Spinal Tear < Brain tear

RESULTS

Spoiler (click to show/hide)

The results were organised into a table for easy analysis:

CONCLUSIONS

Spoiler (click to show/hide)

The results are somewhat surprising, since they imply that the MAX_EDGE parameter had almost no impact on crossbow performance, particularly on injury types, whose percentages seem to remain constant across the board. Most damage remained blunt (bruise/chip/fracture/shatter) type even for very sharp iron bolts.

The very sharp iron bolts appear to have a slightly smaller Average-bolts-to-kill value, but this may be within statistical error. The results are surprising due to the general acceptance of MAX_EDGE being an important parameter used for calculating weapon damage. This was also found by Theoretical code analysis. Possible reason for this discrepancy:

• There are some additional interactions which are unknown e.g. MAX_EDGE could possibly be used more for slashing damage (when weapon has larger Contact Area with Edge type damage?)
• the test is too limited in scope and the effect of the MAX_EDGE parameter becomes more pronounced under different conditions e.g. when 'softer' bolts (copper?) are used against superior armor (e.g. steel)

Specific conclusions: Iron bolts with increased MAX_EDGE parameter perform just as well as regular iron bolts against opponents clad in regular iron armor.

No general conclusions can be drawn.

As a bonus, here is how the only registered Bleed Out death happened:

Spoiler (click to show/hide)

Talk about unlucky

[expwnent]

Perhaps iron already has a high enough max edge that raising it doesn't help. Have you tried lowering it significantly?