Abulafia

Okim, you're full of useful information as ever. I've not been able to get the start_AI_unit_equip console command to work, as yet, but I'm probably using the wrong name, or wrong screen. I'll probably have a play with it over the weekend. Thanks again, Abulafia

Hmm, surprising answer to the testing with skilled shooters but it actually makes sense. Skilled shooters get an extra bonus to accuracy when making aimed shots, on top of the bonuses they get to maximum and effective range of their weapons. This means you can see quite a significant drop in hit chance for highly skilled shooters using snap shots, compared to their aimed shots. Skilled shooters are still better than unskilled shooters at snap shots, as they still get the bonuses to effective range and maximum range to the weapon they're using. Numbers bit: It turns out shooters with high skill get an extra bonus to range FLT and accuracy FLT, which doesn't show up on the inventory screens, but only when using aimed shots. Bonus to both values = 0.1 (Skill Modifier -1) * MODIFIED accuracy FLT Where Skill Modifier is the coefficient for their skill score (see my post on how skill affects to hit chance) and the Modified accuracy FLT is the accuracy FLT after skill increases are factored in initially (formula for this in the same post). This is the first time I've encountered an increase to range FLT being based off accuracy FLT, usually it's the other way around.

Bleh nothing is ever simple - there's a slight complication to what I thought was the formula, where skill very, very slightly increases effective range faster than maximum range most noticably for innacurate weapons being used by highly skilled soldiers. The change is minimal (the maximum possible effect is with a superheroic +4 shooter using a 1% accuracy weapon, and in that scenario his adjusted accuracy is 42%, not 36% as the old formula predicted). I'll redo the Heroic skill graph, shortly. The Average skill graph is still correct. I've also got a Superheroic+4 Graph to add in, which shows the maximimum possible effect of skill (you'll need a fully implanted and chipped cyborg with the appropriate bonus training to get to SH+4, but the effects are quite impressive). Horrible numbers bit: Ye gods. I don't think I can actually model the true accuracy calculation for skill, as the squad screen rounds displayed accuracy to 0 decimal places. I've got an approximate correction that seems out by less than 1%, so calculated scores and displayed values will be right after rounding. It's certainly close enough for effective weapon balancing. For a skill factor Sk (from the initial table): Adjusted Range FLT = Original Range FLT *Sk (as before, this is nice and simple) Adjusted Accuracy FLT = a + ((1-Sk)r * (1.025+0.18x(x-0.2))) where a = original accuracy FLT, r = original range FLT, x = 1-(a/r) (i.e. 1 - original squad screen accuracy expressed as a decimal)

Slightly surprisingly, for Aftershock, Dodge works in a pretty logical fashion, it turns out. I admit being slightly surprised it works at all, and thanks go to Okim whose observations of shooting at his super-dodgy cultists gave me the incentive to test properly. Dodge works as the direct opposite to shooter skill, so the target's dodge skill modifies the range and accurate range of a weapon exactly like shooter skill, except it lowers values rather than raises them. What this means is, like shooter skill, dodge affects lower accuracy weapons more than high accuracy ones. In fact, with a high enough dodge skill, weapons below a certain accuracy can't actually hit you. Also shooter skill and target dodge skill neatly cancel each other out - a superheroic shooter firing at a superheroic dodger has exactly the same hit chance as an average shooter firing at a target with average dodge. Minimum weapon accuracy required to be able to hit vs. target's dodge skill: Good 5% Very Good 10% Excellent 15% Heroic 22% Superheroic 27% SH+1 34% SH+2 41% SH+3 48% SH+4 55% The ugly numbers bit: Dodge uses the same co-efficients as shooter skill i.e. Awful 0.907029478 Poor 0.952380952 Average 1 Good 1.05 Very Good 1.1025 Excellent 1.157625 Heroic 1.21550625 Superheroic 1.276281563 SH+1 1.340095641 SH+2 1.407100423 SH+3 1.477455444 SH+4 1.551328216 Target dodge skill reduces range FLT and accuracy FLT by (S-1)*Ra Where S=the skill co-efficient and Ra=original range FLT So an average rifleman shooting an M4 carbine (Range FLT = 40, Accuracy FLT =22) at a target with Superheroic dodge skill will calculate the attack with (1-1.27628)*40 = 11.05 subtracted from both Range FLT and Accuracy FLT.

General things to know about recoil: Recoil shortens the range and accurate range of the weapon proportionally, which is why the accuracy figure for a weapon doesn't change between burst fire and single shot, but the range figure does. Recoil does not affect the first shot in a burst, so the first round fired will have identical accuracy and range to a single shot. Recoil is applied to each shot in a burst in turn. The third shot of a 20 round burst has the same recoil penalty as the third shot of a 3 round burst, for weapons of equal recoil. As a rough rule of thumb, every shot after the first lowers your accuracy by 2.5% per point of recoil. This is actually exactly right for the third shot of a burst, but isn't quite true for earlier and later shots. A recoil value of 0 means every shot fires with the original accuracy. Accuracy drops towards zero for every shot after the first as recoil approaches 20. The ugly stuff: Recoil reduces both range FLT and accuracy FLT to a fraction of their former values. Note that recoil doesn't work like shooter skill, where the accuracy FLT change is based off the range FLT score. So the third shot of a burst with a recoil 5 weapon will be calculated with a range FLT of 0.75 of the original range FLT and an accuracy FLT of 0.75 of the original accuracy FLT. The modified range FLT and accuracy FLT for the nth shot of a burst from a weapon with recoil R is = ((1-0.05R)^(0.5))^(n-1) Knowing me, there will be graphs to follow A recoil value of 20+ does very weird things, as the game engine is trying to cope with the square root of a negative number. First shot has normal hit chance, second shot has 100% hit chance, third shot 0 hit chance. Personally I'd avoid modding in anything that will give a modified recoil score of 20 or greater. Abulafia

Okim you are a star, as always! Masses of useful information here, thank you very much. Just to clarify a couple of your points - When you say "Skills are locked to attributes and will change taking into account the real attributes values when this soldier is first loaded" do you mean that the skill scores in the .txt file don't matter at all? or that you change the recruit's attributes to be what you want, then use the values from the edited savegame? Will the skill effects from trainings get added in automatically, if you do set the trainings manually (like extra hitpoints from gunman or leader)? Is there anything special you needed to do to mod in armour add-ons for the cultist armours? I notice cultist light armour has only one add on slot, so I presume copy/paste from a light armoured laputa recruit won't work if they have two add-ons? Did you get drones to work by building the drone of choice in the squad screen then copy/pasting the WHOLE text including subrace into borg-guard2.txt? Does the race of the recruit you use to generate the inventory code have to be the same as the subrace entry equip_name STR? so use cyborgs for borgs, psionics for priestesses and humans for muckstars? Many, many thanks again. I've loved how much you've been able to change the enemies in ACM, and it's really helpful of you to share your knowledge. Abulafia.

As a side effect of trying to test if enemy anatomy (level 3 stalker skill) works at all, it seems it does work with guns. Test results (200 aimed shots with a WRB needler modded to do "Psi" damage, using a soldier with super heroic enemy skill, heroic close combat skill, point blank range so 100% hit chance, if you care for detail) were - Enemy anatomy skill, non commando weapon - 41 crits No enemy anatomy skill, commando weapon - 48 crits Enemy anatomy skill, commando weapon - 152 crits Which is pretty convincing that enemy anatomy is correctly tripling crit chance, even with a firearm. Next question is, can melee weapons actually crit? I can't make a damn katana crit at all, so the question of whether enemy anatomy works in melee, which is what I was originally testing, is still open. At first I thought the designers had been clever and made crit chance an inverse function of attack speed (I'd modded the katana down to 0.1s swings to make the testing faster) but testing with time FLT and longshot FLT both set to 5.0 still gave me not a single crit. Has anyone out there actually seen a critical hit with a melee weapon? if so I can start looking for things that are stopping the crits happening, but it's currently looking like melee weapons don't crit.

Thanks Sunflash, more to come I hope. Thanks DerelictableX! 4. Any idea how random it is? Knowing that it adds in pairs is useful. 5. Exactly what I needed! I'm guessing prob INT is the probability of picking that type as a squad member (out of the sum of all the prob INTs for possible squad members) while min INT is the minimum number found in the squad?

Hi all, more questions I'm afraid. This time I'm trying to change enemies stats, skills and equipment. As I currently understand how this works: share/configs/units/unit_phase.txt lists which unit uses which skills and equipment file share/configs/ai_units contains folders for each enemy type which in turn contain .txt files for their loadouts for each phase The individual files list the loadout for a specific enemy - so share/configs/ai_units/borg/cultist-borg_assault.txt lists the stats for an enemy cyborg, of the "assault" class for when the game is in the "cultist" phase (assuming unit_phase.txt hasn't been changed). Each unit has scores for each training, each stat and each skill, where training levels are 0-3 and stat and skill levels are 0-11, with 0 being Awful and 11 being SuperHeroic +4 First of all, am I right on this lot, so far? Secondly, there's some stuff I don't understand about the actual loadout files: 1. Dexterity and agility seem the wrong way round. From comparing save game files with actual ingame stats, the second CUnitAttribute RECORD entry appears to determine agility and the third dexterity. 2. Stats are listed in the format: CUnitAttribute RECORD level INT 2 distLevel INT 2 bonus INT 0 levelString STR "RwBvAG8AZAA=" Where level INT is the actual score for the stat, bonus is any stat bonuses (from cyborg limbs, say) but bonus is already added into level INT, and I have no idea what distLevel INT and levelString STR do. Is it safe to just set distLevel INT = level INT and leave levelString STR unchanged to make everything work ok? also if you have an enemy cyborg with implants, do you have to set bonus INT correctly for the implants you've given him? 3. Skills are listed in the format: skills ARRAY 14 CUnitSkill RECORD value FLT 3.5 level INT 4 bonus INT 0 levelString STR "VgBlAHIAeQAgAEcAbwBvAGQA" Where value FLT is the calculated value of the skill from the stats, level INT is the actual rating of the skill (so value FLT rounded nearest) and bonus INT is any skill bonuses, again these are already added into the level INT value. Do you have to correctly calculate value FLT for each skill to avoid an error? if not, because hp and skills have their own values, can you ignore enemy stats, and just give them whatever hp and skill scores you want, and leave their stat scores unchanged? 4. Each file contains a squad array of, usually 6 or 7, variations to the unit type. Does the variant type get picked at random, or are they cycled through in order? So if 4 borg_assault enemies spawn, will they automatically be types 1,2,3 and 4 or will they each be a random type from 1-6? 5. share/configs/units/squad.txt determines the maximum number enemies (and allies I'm guessing) for each mission type. What determines the makeup of the individual squads? A cultist assault squad contains between 7 and 10 members (unmodded game) but what determines how many of those are borg_assault types rather than, say, seer_sniper? 6. Enemy armaments contain the information em Item RECORD itemTemplateStringId STR "M60E3" loadSaveMimeName STR "TQA2ADAAIABNAGEAYwBoAGkAbgBlAGcAdQBuACAAMAAxAA==" storeItemId INT 80 tagId INT 11822 is all of this apart from itemTemplateStringId STR just used to store savegame info? I notice that the actual item IDs from the equipment.txt file aren't found in the enemy loadout files, so does just changing the name, height, width and loaded ammo of a weapon change the enemy's armament? Many thanks to anyone who can help with any of these. Abulafia.

I've finished up testing out my theory on the 100% accuracy effect - where adding multiple accuracy enhancing add ons to a weapon gave it a 100% hit chance out to maximum range. I've found that the number of accuracy add ons doesn't matter, what's causing the bug is the to hit formula failing when a weapon's accurate range (accuracy FLT) becomes equal to or greater than the weapon's maximum range (range FLT). In the unmodded game, and in several mods, it's fairly easy to increase accuracy FLT beyond range FLT for most weapons, because most weapon add ons don't modify range, just accuracy. Just putting a gyroscope (1.15 accuracy multiplier) and a movement finder (1.2 accuracy multiplier) on a SPAS 12 (range FLT 11 accuracy FLT 8.5) will raise accuracy FLT to (1.15*1.2*8.5) 11.73 and make the to hit formula go nuts. A quick and easy fix if you don't want to screw around a lot with weapon values is to change all accuracy modifying add ons to modify range by the same amount they modify accuracy. So in the above example, if a gyro and movement finder also increased range by 1.15 and 1.2, respectively, the SPAS' accuracy FLT will still be increased to 11.73, but its range FLT will also increase to (1.15*1.2*11) 15.18, keeping it higher than the modified accuracy FLT value and preventing the 100% accuracy effect from kicking in.

Finish up accuracy testing on aimed shot vs. snapshot. Results are pretty clear that aimed shots use the "normal" weapon values, while taking a snapshot lowers range FLT by 10% and lowers accuracy FLT by 10% of range FLT. I've only tested so far with an "Average" skilled shooter, so it's theoretically possible that snapshots lower shooter skill by 2 levels rather than a flat 10% adjustment to range and accuracy, as going from average to awful is very very close to a 10% drop. I'll do some quick tests with a Superheroic +4 shooter over the next couple of days, which should clarify this.

Tests on weapon add-ons now done. Weapon add on adjustments are calculated before skills. i.e. bonuses for skill use the range FLT and accuracy FLT scores after they have been multiplied by weapon add on and ammunition modifiers.

Thanks Okim! I've not looked at armour varients. This is a way of having enemy units have different graphics for a bit of variety?

Okim, if you want to remove the backpacks from an armour do you just set backpack STR to 0, or delete the backpack STR line completely for that armour, or something else?

Big bad and dangerous, from the times I've played ACM That's cool, and you'll have tested by look and feel anyway. If in the future you want a quick and dirty fix to avoid the auto 100% effect, just ensure all weapon mods that increase accuracy also increase range by the same amount. Incidentally did anyone ever work out a way to make the accuracy effect of add ons show in the equipment screen? It would be damn handy for my testing.