Damage Modifiers for Aliens

[MikeTheRed]

Another one nailed down, Z! Good work.

 

That's the third time we've seen 11 crop up as a "point of completion":

• 11 experience points to make the highest skill tier (2-6)
  
• 11 experience points to ensure a Bravery increase
  
• 11 damage to ensure >0 (1-3) Fatal Wounds
  

Although arguably the two experience point ones are the same thing.

 

If you can get around to the armor facing question in post 198, that would be great. (Did you collect armor data along the way?) I would think just 1,000 trials or so would be more than enough to show whether/how much the side armor is hit in a front-to-front encounter.

 

Hi Danial! LTNT!

[zaimoni]

Conceptually, it would be simpler to invoke the RNG twice:

* first check, range 0-10: ask if number is strictly less than the damage.

* second check, range 1-3: actual number of fatal wounds.

 

Of course, the efficient way would be to take one number from range 0-32 and parse it for the above.

 

That is: the measured probabilities plausibly are approximating 0, 1/11, 2/11, ... 11/11 .

[Zombie]

I agree... That is very cool :-)

 

Nice work as always, Zombie. Kudos.

 

So with what you're saying about numbers greater than 10, does that the next line would be:

 

Damage	0	 Ant 1,2,3   Ant 1-3
>10	   0%	  33.33%	  100%

 

Just checking to see if I've gotten the grasp of it.

 

P.S. Have you posted this table and equation on the wiki yet?

Yes indeed, 0 fatal wounds never happen so therefore the number of fatals is always between 1 and 3 when damage > 10. And because there is an equal chance of getting those wounds, the probability is 33.33%. Haven't added anything to the wiki yet, as I'm still doing some tests on this subject and I want to try and tie everything together at some point.

 

If you can get around to the armor facing question in post 198, that would be great. (Did you collect armor data along the way?) I would think just 1,000 trials or so would be more than enough to show whether/how much the side armor is hit in a front-to-front encounter.

I didn't actually test this per-se yet as my test subjects all had 0 armor, but I did have a look through my fatal wound data. (See, BB's logger keeps track of fatal wounds on each bodypart). I'll give some preliminary results here, but please take this info with a grain of salt as it deals with fatals (not armor) and it was a quick 'n dirty calculation.

 

First of all, while looking at where fatal wounds were happening on the "testers" body, I noticed there is a correlation between facing and location of the wound. The two would always match. That's to say that if the target is facing away from the shooter, fatals would primarily happen on the torso (basically, this equates to rear armor). However, torso fatals seemed to max out at about 45-47% of the time, the other 50% of the time fatal wounds happened elsewhere on the body (the arms and legs mostly, but they never happened on the head or front of the body).

 

I suppose the easiest way of showing the % of torso fatals would be to just draw a picture and post it.

 

 

The red squares are where the alien test subjects are standing, the yellow squares are empty and the green square with the X through it is where the shooter stands. As you can see, torso FW's seem to lower as the angle of attack increases. Still, I'm not happy with these results as my shooter is kneeling (which might preclude head FW's), isn't standing directly behind the line of aliens (there are 2 rows of separation in my testing scenario) and the results are not symmetrical. The non-symmetry part is what worries me the most as shooting to the left or right should have the same result (soldiers do not favor a hand for shooting, at least, they shouldn't)!

 

I'll try and do a comprehensive mapping of fatal wounds on my test data to see if that clears up any of these problems. If not, some tweaking of the saved game and some more trials may be necessary. Don't worry, we'll get to the bottom of this one yet folks!

 

- Zombie

[Danial]

Just a quick question about the Stun Damage:

 

You said the probability of higher numbers is low, so I was just wondering what the graph of probabilities looks like - Is it half a bell curve? Or roughly linear? It's hard to tell from your 0-7 example above.

 

I was also wondering how they'd choose the number. If if was purely random, each prob. should be the same, right? As in the Fatal Hits calculations. I've been wondering the same thing about the weapon Damage - How do you create a random number within a bell curve? My statistical maths is a little rusty so you may have to humor me

[Zombie]

Just a quick question about the Stun Damage:

 

You said the probability of higher numbers is low, so I was just wondering what the graph of probabilities looks like - Is it half a bell curve? Or roughly linear? It's hard to tell from your 0-7 example above.

 

I was also wondering how they'd choose the number. If if was purely random, each prob. should be the same, right? As in the Fatal Hits calculations.

Hmmm, yeah the 0-7 stun example was only to give an idea of the mechanics. It's probably better to show what happens with the AP-100 shell.

 

First, some background. AP-100 shells do 100 damage on average while the max against a unit is 200 damage. The game rolls a number between 0 and 200 to determine how much damage to apply to a unit (the distribution is flatline since the probability of each damage number happening is the same). Ignoring armor, this number is subtracted from the health rating. The damage number which the game just rolled is also used to calculate stun. The max is INT(Damage/4) and then the game rolls another random number between 0 and max stun (again, the distribution is flatline). As the damage increases, so does the stun. But there is overlap.

 

As a quick overview, let's look at an AP-10 shell. The damage range will be between 0 and 20. If the damage is between 0 and 3, the only possible outcome is 0 stun. Damages between 4 and 7 produce either 0 or 1 stun, and so on and so forth till we get to 20:

 

         Counts for 11000
       Theoret     Actual
Stun
 0       991         988
 1       771         762
 2       661         685
 3       588         578
 4       533         520
 5       489         504
 6       452         441
 7       420         459
 8       393         372
 9       369         376
10       347         344
11       327         322
12       308         294
13       291         304
14       276         307
15       261         254
16       247         241
17       234         210
18       222         213
19       210         204
20       199         190
21       189         176
22       179         192
23       169         163
24       160         151
25       151         152
26       143         166
27       135         148
28       127         100
29       119         136
30       112         103
31       105         104
32        98          86
33        91         103
34        85          71
35        79          91
36        73          72
37        67          68
38        61          52
39        55          54
40        50          34
41        44          39
42        39          37
43        34          27
44        29          32
45        24          18
46        19          18
47        15          16
48        10          11
49         5          10
50         1           2

As you can (or can't) see, the theoretical and actual counts match up almost exactly. Perhaps a graph would show it better:

 

 

That should make it pretty obvious what is going on. If not, there's always an excel spreadsheet: AP_100_Stun_Probabilities.zip

 

- Zombie

[Danial]

Thanks. The graph makes it much easier to comprehend (for me anyway).

 

I didn't realise the weapon damage probabilities were equal. Cool.

[Alitorious]

If you want to get technical, Zombie, that graph is a histogram (or is it a frequency distribution?), not a probability graph. I just love seeing this kind of work, nonetheless.

 

(Oh, and your sig is hilarious! )

[Zombie]

Histogram, frequency distribution, probability graph... you know what I mean.

 

One of these days I'm going to have to look at the fatal wounds data again. But probably a new testing scenario is warranted. Then we'll know what body parts are affected and at what rates.

 

- Zombie

[Alitorious]

Try varying at what height the shooter/target is at, too. Like on bales of hay or skyranger ramps or something - see if that increases the chance of leg wounds or head wounds. (I just had a rocket launcher guy on the ramp of the skyranger hit the soldier in front of him with a small rocket. By accident, of course! Funny thing was: Rocket guy died, but the soldier in front survived! (Both were wearing personal armour). She was left with two fatal wounds on her head, but it was her under armour that seemed to take the hit.

 

Or maybe that has more to do with ammo types - I'm pretty sure the rocket would have hit her head, and that's where the wounds were applied, but it was her under armour that got diminished because it was explosive type?

 

Either way, we look forward to the results eagerly .

 

 

Oh, and I think histogram and frequency distribution are the same thing. They show how many times a result falls into each class. A probability distribution would have the left side labelled as % chance (or a decimal).

 

A line like you have is for continuous data, but stun damage in the game is discrete. You can't have 3.282 points of stun damage . So, a histogram should have full-width vertical bars instead of a line. The probability is actually dependent on the area of the result you want - in both histograms and Probability Density Functions (pdf).

 

 

Either way, doesn't matter one bit. This is all for fun. I should get engineering classes out of my head!

[Danial]

Ok, I have another quick question:

 

Besides the actual weapon's damage to Health, are Fatal Wounds and Stun the only other directly affected stats (for the target)?

[Zombie]

@ Danial: Well, the target is obviously going to see a reduction of armor if the damage potential is high enough. Also, morale will be affected by a small amount (depending on Bravery I believe). And of course, if damage was done, FA and TA will be reduced too. Other than that, everything remains the same.

 

@ Alitorious: For the fatal wound testing, my soldier was kneeling with 2 tiles of separation between him and the target(s) and not once did the target(s) get a wound in the leg. You would think that the closer a shooter is to the ground, the more likely you are going to inflict a leg wound on the target. Not the case. All the wounds happened to the torso or sides. (The head was always uninjured too). But like you say, varying the height of the target may play a role.

 

Of course, I did my tests on alien targets which normally do not get fatal wounds, so perhaps aliens cannot get wounds to the legs, arms or head. What I might do is just get a bunch of unarmored soldiers together and use them as targets instead. That way, I'm not exploiting any Mind Control features on aliens as "friendly units". Setting this type of scenario up would also take little time: especially on a base defense mission where I can get 40 of my guys to spawn (well, 39 test subjects and 1 shooter). Heck, now that I think of it I could run perhaps 2 or 4 simultaneous mini-tests with varying soldier distances/heights).

 

- Zombie

[Zombie]

Can't remember if I mentioned this before, so here it goes. Long ago, NKF happened to find the damage modifier info in the executable (thanks again!). After a short discussion, I was able to figure out the ordering. One thing bothered me though: there were 14 damage modifier categories, but the OSG only lists 13. At the time I figured it was a dummy record, but I'm doubting that now.

 

Collecting all the alien stats and getting into the modding/editing aspect presented me with with a deeper understanding of what is going on. The aliens and units all have an assigned damage modifier category ranging from 1 to 13. But none of them have a value of 0 (the "dummy" category). What could that first category be? Well, I'm starting to think it is for terrain and items as Apocalypse has a category for it. Haven't really got around to testing if this theory is correct, but I'll get to it sooner or later. Figured I should keep everyone updated...

 

- Zombie

[Bomb Bloke]

Well, the wiki at present only details those other 13 categories (I assume they were taken from the OSG?). I'd assume terrain/items to be 100% suspectable to everything except perhaps melee - Does the executable back this up with the "0" category?

[MikeTheRed]

I see where the GEOSCAPE Alien Stats [27] lists the DM groupings. That entry says "The actual modifiers are listed elsewhere in the executable." Did we establish/put in the wiki, where those are, in GEOSCAPE?

 

Hi all! *waves*

[Bomb Bloke]

Hiya Mike.

 

They wouldn't be in the GeoScape engine... Well, not exactly. They'd go in the Tactical engine instead (because it actually uses the things).

 

The GeoScape engine does, on the other hand, dictate which damage modifying group a unit belongs to (And it writes it to UniRef.Dat, offset 55 for each record). The exceptions are, of course, new zombies and chryssalids.

[MikeTheRed]

Heya Bomb! Great to see you. Sorry I never got around to analyzing the data from your great shot-intersection application... the importance of the results vs. the time involved to analyze the data vs. my interest in other things... sigh. I've still got tons of raw data if anybody wants it... anyway,

 

Where are the DM lookups in TACTICAL then? Somebody seems to know this, but I don't recall seeing it documented anywhere. I and Danial once spent some time trying to hack it out. More precisely, Danial once spent time trying to get me to hack it out. Now that you mention it, I may have screwed up by not byte searching something besides GEO... or did I search across all subdirectories? Ah well.

[Danial]

More precisely, Danial once spent time trying to get me to hack it out.

I did no such thing

 

Yes, I do still visit this site from time to time

 

It's been very quiet here lately though...

[Knan]

What do you expect, it's summer

 

(And besides, I just got sidetracked into attacking british convoys, usually getting whacked pretty bad in the process by the friendly neighbourhood destroyers.)

[Zombie]

Where are the DM lookups in TACTICAL then? Somebody seems to know this, but I don't recall seeing it documented anywhere. I and Danial once spent some time trying to hack it out. More precisely, Danial once spent time trying to get me to hack it out. Now that you mention it, I may have screwed up by not byte searching something besides GEO... or did I search across all subdirectories? Ah well.

The Damage Modifier info is found in the CE executable at byte 450137 (or 450237 if you open it with MS-Edit with 100 column widths). I start with the executable at byte 1 so if my numbers don't match what you see, subtract 1. Or, just look for a long string of "d d d d d d d d d" with the occasional odd character thrown in.

 

I added some of the damage modifier stuff over at the wiki a while back. All the numbers were taken from the GAME, not the OSG. Doesn't matter really as the numbers both agree. And for the reason I mentioned in my post above, I thought the first category (0) was a dummy entry, not a valid DM category so this wasn't added at the wiki. Please note that in a real scenario some of these numbers may not be right. See, the DM's are correct, its just that the game overrides them in tactical sometimes (I'm thinking of the 0% susceptiblity to fire in armor in particular).

 

The first category has 100 for everything (including melee), so terrain and items could be affected. In a pure HTH attack without a melee weapon, no, you would not be able to attack items or terrain. But if you edit a weapon to shoot bullets with a melee damage type, I'd assume it would work (at least for terrain, items can only be destryed with HE). Still, haven't had much time to test this out.

 

- Zombie

[Zombie]

I finally got some time and fooled around with the first damage modifier column (category 0). First I sent my troops out on a desert mission to see what kind of damage the normal pistol could dish out against sand. All it did was scorch it a little bit. Ok, with the baseline established I now edited the first byte (AP Damage Modifier for suspected terrain) to 1000. When I reentered the mission with the changes, the pistol was destroying the sand and turning it into glass like the Blaster Bomb does. Just for giggles, I cranked it up to 65535. Pistol did the same thing to the sand, but now was able to easily cut throught the walls of the Small Scout. Whee! And the best thing about this is that the extra boost in power only affected terrain, not units. So any stray hits didn't kill my men. Quite an effective way to open stuff up if you catch my drift.

 

I also tested what would happen if I cranked the incendiary DM for personal armor up to 65535. Still nothing. My men didn't recieve any damage. This proves that specific attributes of the suit of armor must override any associated damage modifiers. Neat

 

- Zombie

[Danial]

Another mysterious mystery solved!

[MikeTheRed]

Good work, Z. Danial, I guess I should've thought to search through my TACTICAL, since I have the DOS version. I thought I searched a bunch... anyway. Live and learn. And otherwise bow to Zombie's fuller understanding. Anyway - Great to see you, Dan! How've things been for ya, friend?

 

Z, we've seen the aliens that fall into the various DM groups, as per e.g. GEOSCAPE alien stats [27]. But have you posted the DMs for each group? To your alien stats page or whatever? Or has anyone posted them? Sorry for not checking... easier to ask!

[Zombie]

Yes, those are posted on the Damage page under Damage#Damage_Modifiers. I didn't add the terrain column yet as the table is quite cramped right now. I think I can sneak it in by converting it to strict HTML though.

 

- Zombie

[Danial]

Anyway - Great to see you, Dan! How've things been for ya, friend?

Good... except for the cold I have at the moment

 

Still dreaming of the day an X-COM shooter will be made

[Bomb Bloke]

Hasn't that Half Life mod been released yet?

 

Now that I think about it, I thought I downloaded some X-Com Counterstrike maps... Wonder where they got to...