(Edit on 9/18/15: NovaRain gave me a link to http://www.mediafire.com/download/da...2calc_docs.rar , which contains a "partially reverse-engineered Fallout 2 combat algorithm", and should render this post obsolete. I still cannot reconcile some results from my testing though, and if the YAAM mod really is the cause, then I'm pretty sure it was some unintended side-effect.)

I understand that aimed shots to different areas have different bonuses to crit chance, but that "this number isn't documented anywhere" (from the critical hit table at the fallout wikia; I can't seem to link it here) If somebody can give definitive numbers, then this post becomes obsolete.

- Despite the in-game text and the wiki's critical effect table saying "ignore armor" for some critical hits, a small degree of armor still remains to slightly soften the blow, but at least under YAAM, any DT mod can easily blow that remainder completely away.

- For non-critical hits(at least), damage seems to be rounded *up*. Under YAAM at least.

- Weapon penetrate reduces DT by 80% + rounding down the remaining DT. i.e. It rounds up it's effect, i.e. it's "at least 80% reduction".

- The wiki gives magneto-laser pistol as having a special -5 DT *and* -50% DR property, but either using fw2edit destroyed some special property(unlikely imo), or it is really just the weapon penetrate effect.

- Similarly, I cannot see the wiki's stated shotgun mechanics(either the range-dependent accuracy bonus or the range-dependent damage reduction). i.e. in my testing, I could not detect any unique shotgun characteristics. (but you could create one by using f2wedit to give a burst of 2, or something else beyond my creative/modding capacity )

The rest of this post is regarding my empirical data and other tests; you could glance at my sample sizes, but you'll probably want to ignore everything else. It's more for (1) the critical-minded people who want to check my methods + thought process + interpretations (please let me know when you find mistakes), and (2) for people who would like to independently do testing, and/or stack their sample sizes on top of mine.

--------------------------------------------------------------------------------------------------------------------

- I have taken a fixed damage LE Red Ryder BB gun(lowering its damage using f2wedit), and varied small guns skill between 107% and 297%, shooting at Sulik(combat armor) at noon with a 1 Luck, 10 Perception character, with no perks, at range 20. Here is my current data and sample sizes:

At small guns = 107%,

Eyes : 67.67% (sample size: 150) (accuracy was at 68%)

Head : 45.88% (sample size: 320)

Groin : 34.10% (sample size: 473)

Arm : 33.52% (sample size: 197)

Leg : 26.34% (sample size: 395)

Torso : 5.36% (sample size: 440)

Uncalled : 4.66% (sample size: 265)

At small guns = 297%,

Eyes : 75.00% (sample size: 150)

Head : 55.88% (sample size: 320)

Groin : 43.10% (sample size: 898)

Arm : 40.11% (sample size: 343)

Leg : 33.72% (sample size: 265)

- Sulik wasn't wearing combat armor for the head. This is an unintentional carryover from when I started. I may eventually do split tests for the head, but will lump the results together unless and until I find a significant difference.

- "Armor bypassing effects" are not 100%; my "wild guess" is that it reduces both DT and DR by 80%, e.g. 5/40% becomes 1/8%. Have not done the required varied armors and varied BB gun damage to pin this down.

- Regarding the wiki critical effects table damage multipliers for torso, you can easily test it yourself by getting sniper with luck 10 and shooting a hundred shots or something. I will not be doing so, and thus it may take a long time before I get a large enough sample size and update(since crit rate is so low for torso).

- Similarly, while monitoring the critical effects for groin hits, the wiki's critical effect table wasn't able to explain my results, hence the extra section further down below.

- As a related note, under RP, ddraw.ini allows for modification of accuracy modifiers for different parts. Personal appreciation to timeslip and the other creators for sfall(and killap, and haen. <3 )

7 dmg - 5 = 2 dmg, 2 x 0.6 = 1.2. Actual result was 2.

19 dmg - 5 = 14 dmg, 14 x 0.6 = 8.4. Actual result was 9.

- It would appear that damage was rounded up. Now what if it's below 1? I switched to the laser pistol:

Combat armor has 8/60% against laser.

9 dmg - 8 = 1 dmg, 1 x 0.4 = 0.4. Actual result was 1.

- Still seems to hold true(i.e. still rounding up). Now I switch to advanced power armor, still with the laser pistol:

Advanced power armor has 19/90% against laser.

19 dmg - 19 = 0 dmg. Actual result was 0.

20 dmg - 19 = 1 dmg, 1 x 0.1 = 0.1. Actual result was 1. (still true)

29 dmg - 19 = 10 dmg, 10 x 0.1 = 1. Actual result was 1.

30 dmg - 19 = 11 dmg, 11 x 0.1 = 1.1. Actual result was 2. (still true)

Yet so far, from at least a few places that I've read, they tell me that it rounds off to the nearest number. My current guess is that this is due to YAAM, but it's just a wild guess. (somebody without YAAM can easily test this if they need/want to find out for themselves). In addition, I am not yet assuming that the same applies for critical hits.

Here I temporarily gave the LE BB weapon penetrate through cubik's f2wedit. Setting damage to 3 and letting sulik don metal armor(4/30%):

- If weapon penetrate "reduces DT by 80%", then 4 x 20% = 0.8.

- If this 0.8 is not rounded off beforehand, I would expect to see 2.2 x 0.7 = 1.54.

- If this 0.8 is rounded down to 0, I would then expect to see 3 x 0.7 = 2.1.

Actual damage was 3. (which also means that 0.8 couldn't have been rounded up to 1)

Used f2wedit to give the regular pistol weapon penetrate, then used both weapons to shoot at combat armor and advanced armor over a good range of values(from 3 to 20+). Exactly identical, and exactly in accordance with what I've learnt about weapon penetrate and rounding up non-crit damages.

I said earlier that I think there might be a problem with the wiki's critical effect table. Here is my data and current hypothesis.

- Firstly, these are the 6(or 7) text crit results from a 10 dmg BB gun against Sulik's combat armor:

- 5 dmg : "Ouch! That had to hurt"

- 5 dmg : "...

- 5 dmg : "The pain is too much for him, and he collapses like a rag"

- 12 dmg: "...and he's not wearing a cup, either"

- 12 dmg: "without protection, he falls over, groaning in agony"

- 17 dmg: usually "groaning in agony", or if Sulik is unlucky, "collapses like a rag"

I think there are at least two anomalies:

(1) I believe the bolded and underlined is a text error; I also think that there is a corresponding error on the wiki's table, i.e. 21-45's "ignore armor".

(2) The wiki's critical hit table gives 71-90 as "x2 damage, knockout effect"(and nothing else). But combat armor reduces a base damage of 10 to exactly 3, and 3x2 is 6, which is absent from the results.

My guess is that the game is working well, but in a way that is not completely in accordance with the wiki's critical effect table. Which is why I will record my current hypothesis here:

0-20 => x 1.5 dmg

21-45* => x 1.5 dmg....................(end - 4?) roll , pass = knockdown , fail = knockout

46-70 => x 1.5 dmg......................(end - 2?) roll , pass = knockout , fail = bypass armor

71-90 => bypass armor, x 1.5 dmg..(end - 6?)roll , pass = remain standing, fail = knockdown

91-100 => bypass armor, x 2 dmg....(end - ?) roll , pass = knockdown , fail = knockout

* with FAKE bypass armor text for 21-45

I now assume that Sulik starts with 9 endurance, and also that "groaning in agony" = knockdown and "collapses like a rag" = knockout. Then the predicted probablilties are:

"Ouch! That had to hurt" (5 dmg)-----------------> 0-20 ========================> 20%

"

"...collapses like a rag" (5 dmg)------> failed 21-45 OR passed 46-70 => 0.5x25% + 0.7x25% = 30%

"...not wearing a cup" (12 dmg)-----> failed 46-70 OR passed 71-90 => 0.3x25% + 0.3x20% = 13.5%

"w/o protection...groaning..." (12 dmg)-----------> failed 71-90 ===========> 0.7 x 20% = 14%

either knockdown or knockout (17 dmg)----------> 91-100(combining both roll cases) ===== > 10%

Current sample size: 562(disregarding skill differences)

Empirical | Predicted

20.1% | 20%

11.0% | 12.5%

31.9% | 30%

12.6% | 13.5%

13.4% | 14%

11.0% | 10%

My hypothesis may be totally or partially wrong(and the endurance roll modifiers are definitely only estimates). Eventually, as sample size grows, I will re-evaluate my hypothesis and also look at the wiki's critical effect table again to see if it can explain the results.

I also use the 10dmg LE BB gun crits to combat armor sulik's groin to get a clue. If I assume that 12 represents the x1.5 modifier and 17 represents the x2 modifier, and if I assume that these modifiers are applied after the (greatly) reduced armor effects, then:

- I think I can rule out the rounding of damage before the application of the crit effect modifier. (i.e. 8 x 1.5 wouldn't do the trick, since 8 x 2 isn't 17)

- I think it follows that you apply the x1.5 or x2 to the un-rounded damage, and then either round to the nearest number, or you round down. (rounding up would be impossible: anything 8 or less couldn't produce 17, yet 8.001 would produce 13 and 17 instead)

- If you round to the nearest, then the value after reduced-armor lies between 8.25-8.33. If you round down, then the value would lie between 8.5-8.66.

- If armor-bypassing uniformly bypasses 80% of DT and DR, then you would get 1/8%, i.e. (10 - 1) x 92%, = 8.28. I haven't really thought it through yet, and would definitely need to vary the armor type and/or BB gun damage to pin it down, so this 80% uniform armor reduction is almost purely a speculation. Perhaps a modder can enlighten me on to what degree armor is bypassed; but whatever it may be, in my testing, whatever that remains can continue to be cut through by YAAM's DT mod for complete negation of armor.

For the record, here's roughly what I did(whether relevant or not):

- RP 2.3.3, YAAM(totally fresh, I haven't played again for many, many years). Not sure of my CD version; should be some US version, with "Windows 7/Vista" written on it.

- F2se and F2wedit were the only two tools I used.

- Fresh game, Mingan, naturally levelled up to 2 in the process of getting straight to Sulik, put the skill points in small guns

- F2se manipulation to get desired items, stats and skill levels. (including all stats to 10, except 1 luck)

- Used F2wedit to set weapon damage; "weapon long range perk" and other values unchanged.

- Ran out to the wilderness desert with Sulik, set the clock to go to noon.

* Fixed seperate skills in seperate save game slots to eliminate constant tinkering with F2se.

* Range was always consciously fixed at 20

* Care was taken to always reload after a knockdown(to account for possible influence)

* Due to me fearing that computer-generated "random" outcomes aren't truly independent from each other(e.g. a specific unlikely crit effect not happening forever and then happening 3 times in a row), I have made an effort to not stop just after a long unlikely string. I get a "vague" feeling that the random number generator might forcibly correct its probabilities within 100+ hits, but this is *extremely* subjective. One way or another, fair and large sample sizes should give greater overall accuracy.

* Other than increasing sample sizes, if I have time, other investigations include (1) skill = 202%(mid-point) to further check for linearity, (2) varying sulik's armor wearing plus (3) varying range to see if it is "excess accuracy" or just pure skill level, (4) checking for the presence of innate melee bonus crit chance

* Anybody can always replicate my conditions and stack sample sizes on top of what I have given here(assuming you trust my recording accuracy).

I understand that aimed shots to different areas have different bonuses to crit chance, but that "this number isn't documented anywhere" (from the critical hit table at the fallout wikia; I can't seem to link it here) If somebody can give definitive numbers, then this post becomes obsolete.

*Update log:*

- (31/10/14) Uncalled thrust attacks from a knife at 297% gave ~5+% crit, i.e. crit contribution from melee skill seems to be lower than small guns. (sample size = 265)

- (1/11/14) I think I confirmed that at least under YAAM, "weapon penetrate" rounds down the DT before any other calculations. (combined weapon penetrate with DT mod 4, brotherhood armor's 8/40% was identical with combat armor's 5/40%)

- (31/10/14) Uncalled thrust attacks from a knife at 297% gave ~5+% crit, i.e. crit contribution from melee skill seems to be lower than small guns. (sample size = 265)

- (1/11/14) I think I confirmed that at least under YAAM, "weapon penetrate" rounds down the DT before any other calculations. (combined weapon penetrate with DT mod 4, brotherhood armor's 8/40% was identical with combat armor's 5/40%)

**:**__Current empirical estimates__**20 small guns skill = 1% crit chance (edit on 9/18/15: I still can't reconcile this with the mechanic revealed in fo2calg.pdf; either my sample sizes are too small, or YAAM inadvertently somehow produced the effect that I am seeing, or something...what I mean is that in fo2calg.pdf, there should not be any bonus to crit chance beyond 95% accuracy, yet in my tests there was a consistent difference between 107% small arms skill and 297% small arms skill, @ mostly 95% accuracy throughout)**

Eyes : +60%

Head : +40%

Groin and arms : +30%

Legs : +20%

Torso : +0%

Critical effect table at the fallout wikia for groin may be incorrect(some torso multipliers might be off too). Will update.

Tentative conclusions from other tests(with my RP 2.3.3 with YAAM):Eyes : +60%

Head : +40%

Groin and arms : +30%

Legs : +20%

Torso : +0%

Critical effect table at the fallout wikia for groin may be incorrect(some torso multipliers might be off too). Will update.

- Despite the in-game text and the wiki's critical effect table saying "ignore armor" for some critical hits, a small degree of armor still remains to slightly soften the blow, but at least under YAAM, any DT mod can easily blow that remainder completely away.

- For non-critical hits(at least), damage seems to be rounded *up*. Under YAAM at least.

- Weapon penetrate reduces DT by 80% + rounding down the remaining DT. i.e. It rounds up it's effect, i.e. it's "at least 80% reduction".

- The wiki gives magneto-laser pistol as having a special -5 DT *and* -50% DR property, but either using fw2edit destroyed some special property(unlikely imo), or it is really just the weapon penetrate effect.

- Similarly, I cannot see the wiki's stated shotgun mechanics(either the range-dependent accuracy bonus or the range-dependent damage reduction). i.e. in my testing, I could not detect any unique shotgun characteristics. (but you could create one by using f2wedit to give a burst of 2, or something else beyond my creative/modding capacity )

The rest of this post is regarding my empirical data and other tests; you could glance at my sample sizes, but you'll probably want to ignore everything else. It's more for (1) the critical-minded people who want to check my methods + thought process + interpretations (please let me know when you find mistakes), and (2) for people who would like to independently do testing, and/or stack their sample sizes on top of mine.

--------------------------------------------------------------------------------------------------------------------

__:__

Method and dataMethod and data

- I have taken a fixed damage LE Red Ryder BB gun(lowering its damage using f2wedit), and varied small guns skill between 107% and 297%, shooting at Sulik(combat armor) at noon with a 1 Luck, 10 Perception character, with no perks, at range 20. Here is my current data and sample sizes:

At small guns = 107%,

Eyes : 67.67% (sample size: 150) (accuracy was at 68%)

Head : 45.88% (sample size: 320)

Groin : 34.10% (sample size: 473)

Arm : 33.52% (sample size: 197)

Leg : 26.34% (sample size: 395)

Torso : 5.36% (sample size: 440)

Uncalled : 4.66% (sample size: 265)

At small guns = 297%,

Eyes : 75.00% (sample size: 150)

Head : 55.88% (sample size: 320)

Groin : 43.10% (sample size: 898)

Arm : 40.11% (sample size: 343)

Leg : 33.72% (sample size: 265)

__Notes:__- Sulik wasn't wearing combat armor for the head. This is an unintentional carryover from when I started. I may eventually do split tests for the head, but will lump the results together unless and until I find a significant difference.

- "Armor bypassing effects" are not 100%; my "wild guess" is that it reduces both DT and DR by 80%, e.g. 5/40% becomes 1/8%. Have not done the required varied armors and varied BB gun damage to pin this down.

- Regarding the wiki critical effects table damage multipliers for torso, you can easily test it yourself by getting sniper with luck 10 and shooting a hundred shots or something. I will not be doing so, and thus it may take a long time before I get a large enough sample size and update(since crit rate is so low for torso).

- Similarly, while monitoring the critical effects for groin hits, the wiki's critical effect table wasn't able to explain my results, hence the extra section further down below.

- As a related note, under RP, ddraw.ini allows for modification of accuracy modifiers for different parts. Personal appreciation to timeslip and the other creators for sfall(and killap, and haen. <3 )

__About rounding up damage when armor-bypassing effects are not acheived(YAAM-specific?):__

- Combat armor has 5/40% against regular-type damage. BB gun:7 dmg - 5 = 2 dmg, 2 x 0.6 = 1.2. Actual result was 2.

19 dmg - 5 = 14 dmg, 14 x 0.6 = 8.4. Actual result was 9.

- It would appear that damage was rounded up. Now what if it's below 1? I switched to the laser pistol:

Combat armor has 8/60% against laser.

9 dmg - 8 = 1 dmg, 1 x 0.4 = 0.4. Actual result was 1.

- Still seems to hold true(i.e. still rounding up). Now I switch to advanced power armor, still with the laser pistol:

Advanced power armor has 19/90% against laser.

19 dmg - 19 = 0 dmg. Actual result was 0.

20 dmg - 19 = 1 dmg, 1 x 0.1 = 0.1. Actual result was 1. (still true)

29 dmg - 19 = 10 dmg, 10 x 0.1 = 1. Actual result was 1.

30 dmg - 19 = 11 dmg, 11 x 0.1 = 1.1. Actual result was 2. (still true)

Yet so far, from at least a few places that I've read, they tell me that it rounds off to the nearest number. My current guess is that this is due to YAAM, but it's just a wild guess. (somebody without YAAM can easily test this if they need/want to find out for themselves). In addition, I am not yet assuming that the same applies for critical hits.

__:__

About weapon penetrateAbout weapon penetrate

Here I temporarily gave the LE BB weapon penetrate through cubik's f2wedit. Setting damage to 3 and letting sulik don metal armor(4/30%):

- If weapon penetrate "reduces DT by 80%", then 4 x 20% = 0.8.

- If this 0.8 is not rounded off beforehand, I would expect to see 2.2 x 0.7 = 1.54.

- If this 0.8 is rounded down to 0, I would then expect to see 3 x 0.7 = 2.1.

Actual damage was 3. (which also means that 0.8 couldn't have been rounded up to 1)

__:__

About the magneto-laser pistolAbout the magneto-laser pistol

Used f2wedit to give the regular pistol weapon penetrate, then used both weapons to shoot at combat armor and advanced armor over a good range of values(from 3 to 20+). Exactly identical, and exactly in accordance with what I've learnt about weapon penetrate and rounding up non-crit damages.

About groin crit effects:

About groin crit effects:

I said earlier that I think there might be a problem with the wiki's critical effect table. Here is my data and current hypothesis.

- Firstly, these are the 6(or 7) text crit results from a 10 dmg BB gun against Sulik's combat armor:

- 5 dmg : "Ouch! That had to hurt"

- 5 dmg : "...

__, he falls over, groaning in agony"__**without protection**- 5 dmg : "The pain is too much for him, and he collapses like a rag"

- 12 dmg: "...and he's not wearing a cup, either"

- 12 dmg: "without protection, he falls over, groaning in agony"

- 17 dmg: usually "groaning in agony", or if Sulik is unlucky, "collapses like a rag"

I think there are at least two anomalies:

(1) I believe the bolded and underlined is a text error; I also think that there is a corresponding error on the wiki's table, i.e. 21-45's "ignore armor".

(2) The wiki's critical hit table gives 71-90 as "x2 damage, knockout effect"(and nothing else). But combat armor reduces a base damage of 10 to exactly 3, and 3x2 is 6, which is absent from the results.

My guess is that the game is working well, but in a way that is not completely in accordance with the wiki's critical effect table. Which is why I will record my current hypothesis here:

0-20 => x 1.5 dmg

21-45* => x 1.5 dmg....................(end - 4?) roll , pass = knockdown , fail = knockout

46-70 => x 1.5 dmg......................(end - 2?) roll , pass = knockout , fail = bypass armor

71-90 => bypass armor, x 1.5 dmg..(end - 6?)roll , pass = remain standing, fail = knockdown

91-100 => bypass armor, x 2 dmg....(end - ?) roll , pass = knockdown , fail = knockout

* with FAKE bypass armor text for 21-45

I now assume that Sulik starts with 9 endurance, and also that "groaning in agony" = knockdown and "collapses like a rag" = knockout. Then the predicted probablilties are:

"Ouch! That had to hurt" (5 dmg)-----------------> 0-20 ========================> 20%

"

__...groaning..." (5 dmg)---> passed 21-45 ========= > 0.5 x 25% = 12.5%__**w/o protection**"...collapses like a rag" (5 dmg)------> failed 21-45 OR passed 46-70 => 0.5x25% + 0.7x25% = 30%

"...not wearing a cup" (12 dmg)-----> failed 46-70 OR passed 71-90 => 0.3x25% + 0.3x20% = 13.5%

"w/o protection...groaning..." (12 dmg)-----------> failed 71-90 ===========> 0.7 x 20% = 14%

either knockdown or knockout (17 dmg)----------> 91-100(combining both roll cases) ===== > 10%

Current sample size: 562(disregarding skill differences)

Empirical | Predicted

20.1% | 20%

11.0% | 12.5%

31.9% | 30%

12.6% | 13.5%

13.4% | 14%

11.0% | 10%

My hypothesis may be totally or partially wrong(and the endurance roll modifiers are definitely only estimates). Eventually, as sample size grows, I will re-evaluate my hypothesis and also look at the wiki's critical effect table again to see if it can explain the results.

__:__

About armor-bypassing effectsAbout armor-bypassing effects

I also use the 10dmg LE BB gun crits to combat armor sulik's groin to get a clue. If I assume that 12 represents the x1.5 modifier and 17 represents the x2 modifier, and if I assume that these modifiers are applied after the (greatly) reduced armor effects, then:

- I think I can rule out the rounding of damage before the application of the crit effect modifier. (i.e. 8 x 1.5 wouldn't do the trick, since 8 x 2 isn't 17)

- I think it follows that you apply the x1.5 or x2 to the un-rounded damage, and then either round to the nearest number, or you round down. (rounding up would be impossible: anything 8 or less couldn't produce 17, yet 8.001 would produce 13 and 17 instead)

- If you round to the nearest, then the value after reduced-armor lies between 8.25-8.33. If you round down, then the value would lie between 8.5-8.66.

- If armor-bypassing uniformly bypasses 80% of DT and DR, then you would get 1/8%, i.e. (10 - 1) x 92%, = 8.28. I haven't really thought it through yet, and would definitely need to vary the armor type and/or BB gun damage to pin it down, so this 80% uniform armor reduction is almost purely a speculation. Perhaps a modder can enlighten me on to what degree armor is bypassed; but whatever it may be, in my testing, whatever that remains can continue to be cut through by YAAM's DT mod for complete negation of armor.

__"Experiment conditions"__For the record, here's roughly what I did(whether relevant or not):

- RP 2.3.3, YAAM(totally fresh, I haven't played again for many, many years). Not sure of my CD version; should be some US version, with "Windows 7/Vista" written on it.

- F2se and F2wedit were the only two tools I used.

- Fresh game, Mingan, naturally levelled up to 2 in the process of getting straight to Sulik, put the skill points in small guns

- F2se manipulation to get desired items, stats and skill levels. (including all stats to 10, except 1 luck)

- Used F2wedit to set weapon damage; "weapon long range perk" and other values unchanged.

- Ran out to the wilderness desert with Sulik, set the clock to go to noon.

* Fixed seperate skills in seperate save game slots to eliminate constant tinkering with F2se.

* Range was always consciously fixed at 20

* Care was taken to always reload after a knockdown(to account for possible influence)

* Due to me fearing that computer-generated "random" outcomes aren't truly independent from each other(e.g. a specific unlikely crit effect not happening forever and then happening 3 times in a row), I have made an effort to not stop just after a long unlikely string. I get a "vague" feeling that the random number generator might forcibly correct its probabilities within 100+ hits, but this is *extremely* subjective. One way or another, fair and large sample sizes should give greater overall accuracy.

* Other than increasing sample sizes, if I have time, other investigations include (1) skill = 202%(mid-point) to further check for linearity, (2) varying sulik's armor wearing plus (3) varying range to see if it is "excess accuracy" or just pure skill level, (4) checking for the presence of innate melee bonus crit chance

* Anybody can always replicate my conditions and stack sample sizes on top of what I have given here(assuming you trust my recording accuracy).

Last edited: