Jump to content

Melee Damage Investigation


Recommended Posts

Today I set out to confirm that stats affect melee the way I thought they did, since I was so off on spells. This time my observations were more accurate, but there's still some interesting data. I tested with the Obsidian Spear and Icy Longsword and recorded drained HP values as well (only striking when not near full health). The base character was again 10 Strength, 10 Melee and Spear skills, 10 Berserker, and NOT Mighty Warrior. The average damage (again, against enemies with no armor):

 

Stat Change   Obsid  IcySw  Drain  + Obs  + Icy  + Dr
-----------   -----  -----  -----  -----  -----  -----
Base          121.2   82.4   10.9
+10 Strength  166.7  113.9   15.1   45.2   31.5    4.2
+10 Sw/Sp     141.7  103.1   11.0   20.5   20.7    0.1
+10 Berserk   150.0  110.5   16.0   28.8   28.1    5.1
Mighty War.   141.4   99.4   11.1   20.2   17.0    0.2
Bless Spell   136.6   97.1    7.8   15.4   14.7   -3.1

 

First off, the drained HP varied pretty heavily and is probably entirely random and not dependent on any of these factors.

 

The Obsidian Spear claims to do 1-7 damage per skill level, and the Icy Longsword says 1-5. However, this only seemed to affect Strength. Strength however favored the Spear more than we would expect from random die rolls (average of 4 for spear and 3 for sword, versus the 3-to-2 ratio observed). That last part could be random variation, as this data was (as expected) less well-behaved than the spell data, and I didn't feel like doing the 400 blows. wink.gif

 

I couldn't test Roman Training with a Celt. When I tested it less rigorously with my Roman, it appeared similar to melee weapon skill. That makes sense -- Berserker has to have something better about it, given Roman Training's crazy resistance bonuses.

 

At level 14 at least, Mighty Warrior appears to be equivalent to TEN levels of regular weapons skills! Wow. That's an amazing bargain.

 

The main take-home here appears to be that Strength is great; but just how much better it is than the other skills depends on whether you plan on using the Obsidian Spear (or whatever that crazy pike is), or the Icy Longsword.

Link to comment
Share on other sites

  • 2 months later...

I was intrested in damage calculation too, so here's what I found out...

 

Total damage breaks down like this:

 

Each point of Strength adds 1 level

Every odd point of Spears/Melee Weaps adds 1 level

Every even point of Roman Training adds 1 level

Each point of Berserker adds 1 level

Mighty Warrior bonus: charlvl/4 + 2 levels

Blessing pool weapon bonus: +3 levels

Bless spell bonus: +4 levels

Berserk bonus (woad): +5 levels

And the weapon itself adds base-1 levels of damage (for example, obsidian spear, base 6-42=6*(1-7), so 5 levels of damage).

Also there can be bonuses on items (such as Warrior's ring, +2 levels), and a few other things I didn't check (such as curse = -4 levels).

 

So lets say, there is lvl 18 Mighty Warrior, with STR 7, Spears 11, Roman tr. 9, wielding blessed Obsidian Spear.

That's total of 6+7+6+4+3+5=31 levels of damage.

Which means a dice of 1-7 will be rolled 31 times, each result adding up to the total damage.

Link to comment
Share on other sites

Interesting numbers. Some of these make a lot of sense (i.e., the Mighty Warrior formula) but they fail to account for a few things, such as the fact that weapon damage die affects Strength more than Melee Weapons. I'm interested to hear how you arrived at these numbers.

Link to comment
Share on other sites

Quote:
Originally written by Yama:
but they fail to account for a few things, such as the fact that weapon damage die affects Strength more than Melee Weapons.
Uh. Not sure what you mean...

+10 pts of strength is the same as +19 or +20 pts of melee weapons, so yea, there must be a big difference if you're comparing between +10 of each.
Moreover, any increase will be more noticeable on the spear because of the better dice: 1-7 vs 1-5 (and the bigger the increase, the bigger will be the difference between the two).

Lets put your stats to use:

Base levels of damage with 10 Strength, 10 Melee and Spear skills, 10 Berserker:
OS - 10+5+10+5=30
Icy - 10+5+10+3=28
So the average damage is:
OS - 30*(1-7)/2 = 120
Icy - 28*(1-5)/2 = 84

Average with Bless spell:
OS - 136
Icy - 96

Average with +10 Spears/MeleeW:
OS - 140
Icy - 99

Average with +10 Str:
OS - 160
Icy - 114

All of this is reflected in your results quite nicely. And the only thing that stands out - relatively big difference in the results of +10 Berserker vs +10 Strength using obsidian spear - could be attributed to RNG, and generally, the margin of error (surely it can be at least 5% or so) in your empirical test.
Link to comment
Share on other sites

If you look at the columns in my chart showing change in average damage (the "+ Obs" and "+ Icy" colums) you will notice that for Melee Weapons skill, Berserker skill, Mighty Warrior trait and Bless spell, the gains are nearly identical for both weapons. For Strength skill however, the gains are about 50% greater for the Spear. The discrepancy between nearly identical and 50% greater is too wide for me to chalk up to random chance when using 20 data points for each average.

 

You claim that "any increase will be more noticeable on the spear", but in fact this was not the case; Strength was the only increase that was significantly more noticeable on the spear.

 

Additionally, if each point of Berserker added 1 level, we would expect the +10 Berserker results to be similar to the +10 Strength results, but they aren't.

Link to comment
Share on other sites

Quote:
Originally written by Yama:
You claim that "any increase will be more noticeable on the spear", but in fact this was not the case;
That's not just some claim, that's logic.
dice 1-7 vs 1-5
avg 4 vs 3, difference = 1
avg 5*4 vs 5*3, difference = 5
avg 10*4 vs 10*3, difference = 10
and so on.

Couldn't care less if you use your main data (which is prone to RNG fluctuation and loss of precision when rounding to begin with) to deduce some other data (which further lowers precision) and then jump to conclusions based on these new numbers.
What matters is, if you check the numbers from my previous post with the numbers from your test (first two columns), they differ very reasonably. How reasonable is reasonably, depends on your sample size - how many repetitions you did of each, because my formulas do not depend on that. The only error my info can have is human error, but the same can also be true for your data too, and I assure you that I have a habit of double-checking mine until it hurts.

Base (+0): 121.2, 82.4 vs 120, 84
+10 str (+10): 166.7, 113.9 vs 160, 114
+10 mw/sp (+5): 141.7, 103.1 vs 140, 99
+10 bserk (+10): 150.0, 110.5 vs 160, 114
14(?) Mighty (+5): 141.4, 99.4 vs 140, 99
Bless (+4): 136.6, 97.1 vs 136, 96

A couple of points difference is quite reasonable when we're talking hundred points, given the nature of your data. Even with 400 repeats, 5% margin is to be expected. (Almost) everything of the above doesn't exceed that, which quite frankly should make you happy, not sceptical.
As I said, the only thing that sticks out is your Strength vs Berserker for Spear, which I'd prefer to write off as a product of that random fluctuation unfortunately going opposite directions - which was acknowledged to be quite possible in the 1st post of this thread by yourself.
And the next biggest difference between your data and mine is melee skill for Icy (103.1 vs 99) which by itself isn't that huge, but unfortunately doesn't translate into your conclusion too well. Everything else is almost to a tee, sometimes even remarkably so.

I could check some Strength vs Berserk cases again, to see if maybe there's some weird modifier hidden somewhere that I've missed, but frankly such idea doesn't sound too probable to me, given how blunt the rest of the damage system is. Damage for "spears" and "melee weapons" is generated identically as far as I can tell, and levels from Strength are added before any checks of what kinda weapon the attacker is using.

Another thing that just occured to me, is that maybe there are some minor differences between different versions.
All my info is for Windows version of N:R.
Link to comment
Share on other sites

How is your info for the Windows version if it's based on my data, which was taken from the Mac version? This is the question that I was originally asking: what is your data?

 

As to the rest, my answer remains: Random fluctuation does not explain 0% difference versus 50% difference. I'm not interested in arguing it further unless there is more data to discuss.

Link to comment
Share on other sites

Quote:
Originally written by Yama:
How is your info for the Windows version if it's based on my data, which was taken from the Mac version? This is the question that I was originally asking: what is your data?
If he knows what order different checks are made by the game when calculating damage, then I think I know why he's reluctant to discuss the source of his data. winkwink
Link to comment
Share on other sites

Quote:
Originally written by Yama:
How is your info for the Windows version if it's based on my data[/i]
It's not based on your data at all. How did you come up with that nonsense? I presented you your data with my data side by side, since you were having troubles noticing it in my previous posts and comparing it with your tests. Aparently you still do.
Plz increase your reading comprehension skill before replying. Kthx.
Link to comment
Share on other sites

Quote:
Originally written by Maimonides:
Predictions are not data.
Your definition of data is different from mine, that's all.

As for the evidence, right now I have a strong impression that OP is stressing this issue for reasons that are not related to the topic at all. That pretty much anything I say would be countered with the same "oh, but my test tells otherwise", coupled with an attemp to discredit whatever evidence I present. That's already happened when I posted what you call "predictions", and it's going to happen again.
Since I didn't feel like playing these kind of games, especially when I'm physically outnumbered to begin with, and because the evidence I can present wouldn't make any difference even if there was genuine interest in it, I tried some other ways to present you the required proof. Which isn't good enough for you... fine, but what can I do (or should I even care).

I say it wouldn't make any difference, because most likely you are clueless as to what I would say is true or not. Which you shouldn't take offense at, as it's impossible for anyone to be expert in every field, usually each of us is layman in most. There are things or theories in our lives that we simply have to believe in (as the effort wasted trying to get the question "how?" answered is simply not worth it). Or declare them bogus and believe in something else, that's always an option too. smile In any case, it's still a matter of belief.

I mean I could say that, for example, this particular snippet
Code:
0041B475  |>  8B8D 4C125300           MOV ECX,DWORD PTR SS:[EBP+53124C]0041B47B  |.  0FB79424 A2000000       MOVZX EDX,WORD PTR SS:[ESP+A2]0041B483  |.  8B85 48125300           MOV EAX,DWORD PTR SS:[EBP+531248]0041B489  |.  C1E9 08                 SHR ECX,80041B48C  |.  F6C1 40                 TEST CL,400041B48F  |.  895424 18               MOV DWORD PTR SS:[ESP+18],EDX0041B493  |.  894424 2C               MOV DWORD PTR SS:[ESP+2C],EAX0041B497  |.  74 0E                   JE SHORT Netherga.0041B4A70041B499  |.  83C7 03                 ADD EDI,30041B49C  |.  834424 10 0A            ADD DWORD PTR SS:[ESP+10],0A0041B4A1  |.  EB 04                   JMP SHORT Netherga.0041B4A70041B4A3  |>  8B7C24 14               MOV EDI,DWORD PTR SS:[ESP+14]0041B4A7  |>  8BB424 48010000         MOV ESI,DWORD PTR SS:[ESP+148]
from the disassembly of Nethergate.exe tells me that blessing pools give +3 dmg levels and +10% tohit bonus. I could even go great lengths explaining you the steps I took to get to that particular snippet, and how I made myself pretty damn sure that it means what I believe it means.
I could do that for every bit of my info. It would take much more than this page to do so, and in the end, I'm not sure it would be successful, as there's a lot of side information to take in when you start from 0, and I'm not particularly good at explaining things. So what's the difference if I've said it or not?

I know I've grown to trust my skills more than any test consisting of limited repetition and calculation of average result from a pseudorandom number generator neither you nor I know much about (even though I'm not overruling possibility that sometimes my skills may be not good enough. And I would like to say once again - OP data is close enough to the output from my formulas, only with some difference which can be easily discarded as expected margin of error in the method used to collect that data, close enough to reassure my belief that I haven't missed anything of importance in my research, so that data from independent source was useful to me).
What do you trust, is up to you.

Anyway, now is a pretty good opportunity to cry out "evil haxor" and run me out of your town wink
Link to comment
Share on other sites

Heal_plz, all people were asking you to do is to explain your data-collection methods, rather than just stating conclusions. That's standard practice in any science; it's not a sign of distrust. It would have been much simpler for you to say "I disassembled the binary" the first time someone asked, rather than to dance around the question.

 

Anyway, this whole discussion got me curious, so I went ahead and ran some tests of my own. I used the method that heal_plz suggested a few posts back: edit "corescendata2.txt" and set one weapon's it_damage_per_level property to 1, and its it_bonus to zero. (Actually, I did this to three weapons: one melee, one spear, and one sling.) Then this weapon will do its damage in "one-sided dice", so there are no random numbers and no need to average a large data set. (You do need to remember to add any blocked damage to the total.)

 

I created a new party, gave one of each modified weapons to each party member, and experimented with training their skills to different levels.

 

I verified most of heal_plz's earlier statements (didn't try the blessing pool, since this is a new party) and found a few other things as well. In particular:

 

Strength gives no to-hit bonus, and gives +1 damage die per level with hand-to-hand weapons. It has no effect on ranged weapons.

 

Dexterity gives +5% to hit per level with all weapons, and +1 damage die per level with ranged weapons.

 

Weapon skill gives +5% to hit per level, and +1 damage die per two levels (with the damage increasing on odd levels).

 

Berserk gives no hit bonus, and +1 damage die per level with hand-to-hand weapons. It doesn't affect ranged weapons. (So for combat purposes, it's exactly like strength.)

 

Roman training gives +5% to hit per two levels (incremented on odd levels) and +1 damage die per two levels (incremented on even). It applies equally to hand-to-hand and ranged weapons.

 

The "Mighty Warrior" trait gives +10% to hit and +2 dice of damage. Every four character levels (at 4th, 8th, 12th, etc), these bonuses increase by an additional +5% to hit and +1 damage die.

 

The "Battle Rage" spell gives +20% to hit and +4 damage dice, apparently independent of the skills or traits of the druid who casts it.

 

Each level of the weapon's it_bonus property adds +5% to hit and +1 die of damage. Thus, weapons listed on-screen as doing more than one "base" die of damage will also have better hit rolls.

 

 

On the defensive side:

 

Each level of dexterity gives your enemies -5% to hit you.

 

A shield spell, apparently regardless of the skills or traits of the druid that casts it, gives enemies -15% to hit. It also seems to reduce damage by some fixed amount (which may depend om the druid's skill; I haven't checked), rather than by a percentage like armor does.

 

The defense skill is interesting: it apparently reduces the enemy's hit chance by a random amount each time they swing, from 0% (no reduction at all) up to -5% per level of defense.

 

I'm curious to verify heal_plz's statements (in another thread) about the Armor Use skill...but I imagine they're correct as well. Seems like he or she is skilled at disassembly and code reading, though less so at communication.

Link to comment
Share on other sites

Quote:
Originally written by divergence:
It would have been much simpler for you to say "I disassembled the binary" the first time someone asked
I suppose you have a point, but does it really change anything? Anyone can say those words, this alone does not increase credibility and provides no "evidence". It's still a matter of personal trust, and still anyone can reply "but my test result is different".

Quote:
Originally written by divergence:
On the defensive side:

Each level of dexterity gives your enemies -5% to hit you.

A shield spell, apparently regardless of the skills or traits of the druid that casts it, gives enemies -15% to hit. It also seems to reduce damage by some fixed amount (which may depend om the druid's skill; I haven't checked), rather than by a percentage like armor does.

The defense skill is interesting: it apparently reduces the enemy's hit chance by a random amount each time they swing, from 0% (no reduction at all) up to -5% per level of defense.
I believe your observations are correct.
And shield spell damage reduction? It gives -3 dmg levels to the attacker. Doesn't depend on caster's stats, just like bless spell.
Link to comment
Share on other sites

Quote:
Originally written by heal plz:
I suppose you have a point, but does it really change anything? Anyone can say those words, this alone does not increase credibility and provides no "evidence". It's still a matter of personal trust, and still anyone can reply "but my test result is different".
There is a difference between "trust me, it does X" and "trust me, I'm convinced it does X because I've done Y to confirm it", you know. For what it's worth, now that I know you're not just pulling claims out of random orifices I'm much more inclined to believe you, and I have no particular reason to disbelieve you.

And not everyone can say those words -- someone who at least knows that disassembling the binary is a valid way of obtaining information about game mechanics is, on the whole, more likely to know what the hell he's talking about than someone who doesn't. There are at least two separate issues here, trusting in your honesty and trusting in your competence to interpret data, and further evidence goes at least some way toward bolstering the latter kind of trust.
Link to comment
Share on other sites

Quote:
Originally written by heal plz:
I suppose you have a point, but does it really change anything? Anyone can say those words, this alone does not increase credibility...
Well, speaking for myself alone here...yes, knowing how you reached your conclusions does increase your credibility. It puts your results in context; we know where they come from, even if we haven't pored over every line of code with you.

Granted, it's still possible you may have made a mistake in interpreting the code, but it's easier to assess your results now that we know how you got them. And the fact that you and I reached the same conclusions via different methods lends me more confidence than if we had repeated exactly the same experiments.

Quote:
And shield spell damage reduction? It gives -3 dmg levels to the attacker. Doesn't depend on caster's stats, just like bless spell.[/QB]
Ah, that makes sense. No wonder the poor rats and goblins in the early game can't do any damage through a shield spell; it reduces them to zero dice or less.
Link to comment
Share on other sites

Quote:
Originally written by heal plz:
I mean I could say that, for example, this particular snippet
Code:
0041B475  |>  8B8D 4C125300           MOV ECX,DWORD PTR SS:[EBP+53124C]0041B47B  |.  0FB79424 A2000000       MOVZX EDX,WORD PTR SS:[ESP+A2]0041B483  |.  8B85 48125300           MOV EAX,DWORD PTR SS:[EBP+531248]0041B489  |.  C1E9 08                 SHR ECX,80041B48C  |.  F6C1 40                 TEST CL,400041B48F  |.  895424 18               MOV DWORD PTR SS:[ESP+18],EDX0041B493  |.  894424 2C               MOV DWORD PTR SS:[ESP+2C],EAX0041B497  |.  74 0E                   JE SHORT Netherga.0041B4A70041B499  |.  83C7 03                 ADD EDI,30041B49C  |.  834424 10 0A            ADD DWORD PTR SS:[ESP+10],0A0041B4A1  |.  EB 04                   JMP SHORT Netherga.0041B4A70041B4A3  |>  8B7C24 14               MOV EDI,DWORD PTR SS:[ESP+14]0041B4A7  |>  8BB424 48010000         MOV ESI,DWORD PTR SS:[ESP+148]
So what's the difference if I've said it or not?
For those who skilled enough, your snippet tells nothing but that at some condition one value get increased by 10, and another by 3 ) But for those who don't it does give some mystical evidience )
Anyway, I agree in both cases it's matter of personal belief - trust, or not to. Well, I do ^_^
I'm impressed with your patience and determination to RE game mechanics! Sometimes I too catch myself on desire to fire up IDA, but always can't convince myself that results woth efforts frown
PS
Hmm...your code formatting looks unfamilair. What disassembler do you use?
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...